The Money Mustache Community
Learning, Sharing, and Teaching => Investor Alley => Topic started by: MustacheAndaHalf on May 18, 2016, 11:21:53 PM
-
I'm trying an experiment with a tiny fraction of my portfolio, where I see what it's like to apply momentum algorithms to a sort of tax-managed mini-fund.
Here's the criteria I imposed on this experiment:
* momentum involves lots of buying and selling, so I need to limit the costs involved with trades
* insignificant compared to my portfolio (under 1%)
* benchmark against S&P 500
* limit the tax impact
So what I decided on is sector rotation using Vanguard funds. That way I'm covering the market (roughly), trades are $0, and expense ratio of the holdings are also low. I allocate 20% gains over the year, and will skip buy/sell until this "tax gain budget" has enough room. I'll start posting results and observations later in this thread.
Important caveat: this experiment will prove nothing. Meaningful data would require years, and I don't plan to run it that long. The best fund is still the Vanguard Total Stock Market fund, not a momentum fund.
-
I started this experiment 2016-02-17.
(2016-02-17) I intended to use price momentum, and used proximity to 52-week high on accident. The momentum experiment started with a mistake, buying Vanguard Utilities ETF (VPU). Oddly enough, the next day momentum pushed VPU to the top of both criteria. From this point on, I decided to weight both criteria equally: use the 12 month gain minus the past month for half the fund, and the sector closest to a 52 week high for the other.
(2016-03-22) In past five weeks, S&P 500 +5.3% and VPU +6.4%. Momentum now pegs consumer staples (VDC) and telecomm (VOX) as the funds to buy, so I sell VPU and buy VDC and VOX equally. Which is another mistake: I just went over the tax gain budget for the first 5 weeks. I allocated a max gain of 4%, but just cashed in a gain of 6.4%. So I'll be locked out of trades until the tax gain budget catches up.
(2016-04-27) Couldn't switch funds for lack of tax gain budget (1.5% / 5 weeks). Keeping VOX and VDC.
Between 2016-03-22 and today (2016-05-18):
S&P 500 ETF ("VOO") +0.1%
Cons Staples + Telecom ("VDC" & "VOX") +0.0%
-
Interesting strategy. I'm currently getting my masters in finance at the University of Utah; my professor was talking about this strategy and how there is an opportunity to beat the market with this strategy.
Blackrock has recently created some "smart beta ETFs" with one fund as a momentum etf. If you look up MTUM and compare it against SPY, it actually has done better for the last 3 years (41% vs. 26.7%). It will be interesting to watch yours unravel, and whether a company like vanguard picks up on these strategies easy enough to put into an ETF (I doubt it). MTUM has a .15% expense ratio; I've been tempted to put it into my portfolio.
-
I see nothing wrong with using some "play money" to try different investing strategies. Keep us posted. MTUM sounds interesting. The outperformance over 3 years is solid.
Best to keep "play money" limited to 5% of the portfolio or less. I've got about 1% in a small microbrewery, and will be putting in another 0.5% to another.
-
teltic - Note the experiment here is more limited, and probably worse than a fund which buys and sells stocks based on momentum. Unlike the fund, the mixture of stocks in a sector will have mixed momentum. I could very well believe momentum beats the S&P 500 while sector rotation does not. But in order to save on buying/selling costs, I picked sectors owing to the low cost ETFs available at Vanguard. So you might want to avoid sector rotation as a full portfolio strategy.
PhysicianOnFIRE - Play money here is under 1% of my portfolio. Unlike a small cap or value tilt, some sectors are only 1% of the economy and could behave dramatically different from the S&P 500. But in practice I don't expect too much difference from 100+ stocks picked from one specific sector.
Tomorrow is my next date to make a purchase, and currently Utilities are in the lead on both momentum measures I'm using. So I should have an update tomorrow.
-
The following two sectors were held for 10 weeks, from 3/22 to 6/1:
Cons Def (VDC), buy $133.74, sell $135.63, gain of +1.4%
Telecomm (VOX), buy $90.68, sell $91.43, gain of +0.8%
Return from 3/22 to 6/1 was +1.1% versus S&P 500 return of +2.6%
The strictness of the fund's tax management impacted performance. On 4/22 the fund was prevented from switching to Utilities (+1.6% for May) because of a strict tax gains budget. Instead the fund remained in telecomm and consumer staples (+0.0% and +0.8% for May) as they lost momentum. To address this the gains budget will be changed to be more flexible: +4% per quarter, and +4% during the year (total +20% max realized gains).
Overall record from 2/17 to 6/1 is +7.6% for experimental fund, +8.1% S&P 500.
-
This momentum experiment has 200% turnover in 5 months. Here's the prior purchase decisions:
2/17: utilities
3/22: consumer staples, telecom
6/01: utilities
Since 6/1, Utilities are up +7.1% while the S&P 500 lost -0.5%. Note decisions for the upcoming 5 weeks can change quickly - yesterday REIT topped one of the signals, today it's back to utilities for both momentum signals. So the experiment keeps utilities for the next 5 weeks.
Overall record from 2/17 to 7/6 is +15.2% momentum experiment, +7.6% S&P 500.
-
This momentum experiment has 200% turnover in 5 months. Here's the prior purchase decisions:
2/17: utilities
3/22: consumer staples, telecom
6/01: utilities
Since 6/1, Utilities are up +7.1% while the S&P 500 lost -0.5%. Note decisions for the upcoming 5 weeks can change quickly - yesterday REIT topped one of the signals, today it's back to utilities for both momentum signals. So the experiment keeps utilities for the next 5 weeks.
Overall record from 2/17 to 7/6 is +15.2% momentum experiment, +7.6% S&P 500.
Are returns adjusted for tax efficiency? Just curious, as transaction costs, spread and tax inefficiency usually eats up most arbitrage type strategies that might work on paper.
-
This momentum experiment has 200% turnover in 5 months.
...
Overall record from 2/17 to 7/6 is +15.2% momentum experiment, +7.6% S&P 500.
Are returns adjusted for tax efficiency? Just curious, as transaction costs, spread and tax inefficiency usually eats up most arbitrage type strategies that might work on paper.
No, and that's important. The longer this experiment runs, the more taxes matter. When momentum pays short-term gains, that money goes to the IRS the next year and at a higher tax rate. The S&P 500 doesn't sell, and doesn't owe tax yet - so it earns money in the meantime, and years later pays a lower tax rate on long-term gains. Right now momentum got lucky - the economy took a hit while it had 100% utilities. So even a tax hit in the first 5 months won't erase the +15% vs +8% gap... but I expect that gap is temporary, and taxes matter.
Your other questions are why I selected Vanguard ETFs. Vanguard ETFs are traded frequently and cost $0 for Vanguard account holders. I track actual purchases: the bid-ask spread is reflected in the purchase and sale price, and even includes a tiny SEC transaction fee. The benchmark S&P 500 ETF is typically $0.01 bid-ask spread on $195/share, so although I don't purchase the benchmark there's very little bid-ask spread there.
I hope to add after-tax performance in the update next month, which will be much more accurate and spreads the word about viewing investing with after-tax efficiency in mind.
-
Utilities performed poorly this past month relative to the S&P 500. This momentum experiment was 100% in utilities, and took the full brunt of this gap. Measuring performance since 2/17/2016:
Ignoring taxes, momentum experiment +9.06% versus +12.00% benchmark.
After taxes, momentum experiment +6.79% versus 10.20% benchmark.
I'm assigning the benchmark S&P 500 a tax rate of 15% and momentum a tax rate of 25%. This captures the common S&P 500 buy-and-hold approach that involves long-term capital gains rates. In general S&P 500 involves a choice of waiting to sell at long-term rates, while momentum does not. Note 25% is the median tax rate in the U.S., with 15% being the corresponding long-term capital gains rate.
One switch in how the fund picks sectors: buy the tied sectors. When sectors are closely grouped at the top of one momentum measure, all of the sectors within 3% of the top will be purchased equally. This has occurred several times in the past - one near miss involved 4 sectors tied for new highs. Since it wasn't a scheduled purchase date, it had no impact. But these ties represent information the fund is ignoring, and buying the tied sectors equally should better represent which sectors have momentum.
Momentum switched from 100% utilities to four sectors of 25% each:
25% utilities (performance momentum)
25% telecomm (0.998 tie)
25% information technology (closest to 52-week highest price)
25% materials (0.9833 tie)
-
I skipped the 9/14 posting, which breaks the audit trail of these posts. Sector momentum for industrial and telecommunication sectors fell to low and both were sold. That caused the fund to buy more utilities, and more info tech. Performance was bad then, and will remain bad now (relative to S&P 500 benchmark, incl taxes).
This month is still bad, but more interesting. One of the annoying things with the momentum measures is selling and buying of the same sectors. For example, utilities lacked momentum and both the 1 month holding and 4 month holding were sold. Replacing utilities is telecommunications (which was sold last month, and has now been bought again).
The other momentum measure struck a 3-way tie between information technology, energy, and financials. So 2/3rds of the existing information technology holding was sold, making room for an equal split between the three. I think this is the first time either energy or financials had any momentum since the experiment started in Feb 2016.
The sector momentum experiment is +6.2% so far, or +4.7% after taxes (ST 25% rate).
The S&P 500 benchmark is +10.3% so far, or +8.8% after taxes (LT 15% rate).
-
Following!
Thanks for the honest and (semi)faithful updates. :)
-
Interesting stuff! I really appreciate the updates! Maybe at year end I'll help nerd out and see if holding both would lower your standard deviation.
-
Appreciate the feedback - wasn't sure if the thread was filling a void or just speaking into a void. :)
Consider reading the paper "FTSE Momentum Factor" to learn what I learned. I picked two of the momentum measures they studied, as they described them. But momentum of Vanguard sector funds was purely a guess on my part based on $0 cost to trade and dividing the market up as finely as possible. That's fine for an experiment, but if you put money behind the idea it's all based on a guess.
http://www.ftse.com/products/downloads/FTSE_Momentum_Factor_Paper.pdf
I'd also like to highlight how flaky sector momentum can be. Last week's purchase turned over about 83% of the fund's assets compared to the prior purchase. If this week was another checkpoint, it would have turned over 83% again, reverting back to the September holdings. In the first 8 months of this sector momentum experiment, turnover has been 400%. Often when I check a day early, the sectors differ from the next day. Actually "financials", one hour later was no longer part of a 3-way tie.
-
Appreciate the feedback - wasn't sure if the thread was filling a void or just speaking into a void. :)
Consider reading the paper "FTSE Momentum Factor" to learn what I learned. I picked two of the momentum measures they studied, as they described them. But momentum of Vanguard sector funds was purely a guess on my part based on $0 cost to trade and dividing the market up as finely as possible. That's fine for an experiment, but if you put money behind the idea it's all based on a guess.
http://www.ftse.com/products/downloads/FTSE_Momentum_Factor_Paper.pdf
I'd also like to highlight how flaky sector momentum can be. Last week's purchase turned over about 83% of the fund's assets compared to the prior purchase. If this week was another checkpoint, it would have turned over 83% again, reverting back to the September holdings. In the first 8 months of this sector momentum experiment, turnover has been 400%. Often when I check a day early, the sectors differ from the next day. Actually "financials", one hour later was no longer part of a 3-way tie.
Odd.
Whenever miles explained his momentum strategy, he always claimed that turnover/any action at all was very rare.
I wonder if that's because:
A) Of the time frame he was talking about--he hadn't had much activity to that point, but recently he has, like you, due to market movements,
B) He has a significantly different momentum strategy than you, so he actually does have quite a bit less turnover than you, or
C) Other
-
Odd.
Whenever miles explained his momentum strategy, he always claimed that turnover/any action at all was very rare.
I wonder if that's because:
A) Of the time frame he was talking about--he hadn't had much activity to that point, but recently he has, like you, due to market movements,
B) He has a significantly different momentum strategy than you, so he actually does have quite a bit less turnover than you, or
C) Other
Per his blog post on the subject, Dual Momentum compared the relative performance of global asset classes, and not sectors.
For example, comparing the 6 month look back of VTI, VXUS and short term T-Bills. Such broad Indexes would not be as volatile as sectors.
-
Miles' last annual momentum update on his blog [also last post there] was 15 Nov 2015 so perhaps he will post another update shortly or he may have gone off the strategy since last year?
-
Thanks for the clarification, Tux!
-
I'll clarify what I use, but please discuss momentum in general in a new thread or this one will be sidetracked.
Per the white paper I just mentioned, this experiment uses the past 12 months ignoring the last month. So 11 months of data, skipping the most recent month. The two momentum measures are percentage gain, and proximity to 52-week high. If one sector is closer to it's 52-week high than any other, it has more momentum. Similarly, whichever sector gained the most in the 11 of the past 12 months (ignoring prior month) has momentum.
I take the worst and best momentum sectors, and mark those as 0% and 100%. All sectors above 97% are considered to tie, and are used in the next purchase. Vanguard's sector funds are bought in whole shares, so any cash of less than 1 share is leftover. Otherwise the fund is 100% invested in sector funds at all times - no bonds.
My goal for this thread is to document the purchase decisions, mistakes and all, so that a record is created showing the purchases made. Someone "auditing" this thread could go back and trace each purchase and verify the performance. I also want to answer questions about this experiment as they arise, but I'd like to keep my thread from veering off that focus.
-
Vanguard has 11 sector funds (REIT, Energy, etc) which together roughly represent the stock market. I take 52 weeks of price history, subtract 4 weeks of price history, and use that number to rank the 11 sector funds. Separately, I measure 52-week high vs current price, and again rank those 11 sector funds. The top fund of each criteria gets 50% of the investment.
Note that I'm using blocks of the stock market (the sector funds) which may not have very good momentum signals. The stocks in each sector are likely a mixture, and I'm using sector funds as an experiment. If Vanguard had another way of dividing the market into ~11 pieces and $0 transaction costs, I might use that. This in no way is a vote for using sector funds and momentum - it's an experiment.
Ties are tricky to explain. If the worst fund is -10% and the best fund is +20%, there is a 30% gap between them. I rank all the other sector funds from 0.00 to 1.00 using the best and worst performers as the two extremes. Anything at 0.97 to 1.00 is tied with the top performing 1.00 scoring sector. All funds involved in the tie are purchased equally. This way when sectors are very close in top performance, I capture both. Note the idea of a "tie" was introduced during the experiment, not at the start.
-
The U.S. election switched which sectors had momentum, creating a 5-way tie. Half of this fund invests in "price momentum", which is the current price divided by the highest price in the past 12 months. With 5 sectors near new highs, those 5 form a tie. The other half is "performance momentum", which invests in the best 12 month performance minus the past 1 month performance (sample 11 months of performance, then wait 1 month to buy it).
It might help to illustrate the buys and sells by percentage of fund assets:
50% telecom sold (VOX), 50% utilities bought (VPU)
6.7% finance sold (VFH) to switch from 3-way tie to 5-way tie
6.7% energy sold (VDE) to switch from 3-way tie to 5-way tie
16.7% information tech (VGT) sold, no longer part of 3-way tie
10% industrials (VIS) bought as part of 5-way tie
10% materials (VAW) bought as part of 5-way tie
10% consumer discretionary (VCR) bought as part of 5-way tie
I didn't run the performance numbers yet, but it looks like the fund beat the S&P 500 for the past 5 weeks by a slim margin. S&P 500 +3% (2.6% after tax) and momentum +4% (+3% after tax). Note the greater tax impact reduces momentum's lead by about half.
-
On the last trading day of the year the momentum experiment sold most of it's prior funds and bought:
50% Vanguard telecom (VOX, was 0%)
50% Vanguard Energy (VDE, up from 10%)
Vanguard Financials (VFH) did well but probably not enough to rescue the other holdings. I'm tried to implement an easier way to automatically calculate profit but hit snags, so I don't have the final results yet. I'm posting now so I don't stall this update while I work on the profit for the year (Feb - Dec).
-
Just XIRR in excel based on deposit dates/amounts and ending value.
-
I'll keep that in mind in case I don't go through with the automation I'm attempting. I'd like a setup that both calculates profit and provides automated suggestions of the next investment. And a further complication is money I used for a second experiment was merged in (like a failed fund!).
My other experiment was with small/mid/large value/growth splits of the market. But most of the time was just alternating between large/growth and large/value, so I ended it. It did not seem to split the market into very many pieces, which I felt is important. So that money merged into the sector momentum experiment, and I backdated its performance to match the rest of the funds.
It would also help if I had more time to make changes to the sector momentum spreadsheets. I've devised a spreadsheet with more detailed tracking, but only got it half filled in. And then I need to make calculations and automation part of it.
Even now, I still have to manually sort the sector funds by performance. I haven't had time to automate that - I think it would involve the raw data being in one place, and the sorted data being developed from the raw data.
-
Spreadsheet that pulls in quotes and spits out keep same or move based on the criteria you set.
-
On the last trading day of the year the momentum experiment sold most of it's prior funds and bought:
50% Vanguard telecom (VOX, was 0%)
50% Vanguard Energy (VDE, up from 10%)
Vanguard Financials (VFH) did well but probably not enough to rescue the other holdings. I'm tried to implement an easier way to automatically calculate profit but hit snags, so I don't have the final results yet. I'm posting now so I don't stall this update while I work on the profit for the year (Feb - Dec).
Does Fidelity have similar equal funds for VOX, VDE, etc?
-
For the spreadsheet crowd, SORT() came in handy.
As to sector momentum elsewhere, keep in mind this is not proven - it's an experiment I thought up. I thought how can I divide the market into the most slices, with no cost transactions. So having given that warning, you could look at iShares sector funds which I believe are free to buy & sell at Fidelity. But iShares has duplication in it's list of sectors, so you'll need to narrow their full list down to just one ETF per sector.
-
In today's wrestling contest... we have the benchmark S&P 500 in this corner, gaining +1% over the past 5 weeks. And in the other corner... oh no! It's Trump trashing the momentum experiment! Owch, that looks like it hurt. And momentum goes down -2% for the 5 week period.
S&P 500: +1.1% applying 15% tax rate: 0.9%
momentum: -2.0% applying 25% cap loss: -1.5% (IRS shares capital losses)
I've stopped making a running total of momentum performance, as I need to separate the experiment into the additional funding period and the period after that. Haven't kept up the spreadsheets enough. The momentum experiment turns 1 year old this month. Momentum suffers during volatility (signal says buy, and after buying things change), so maybe this can become a perverse experiment in negative performance. Here's the buy/sell decisions based on momentum this time:
-50% sold VDE (energy)
-50% sold VOX (telecom)
+50% buy VCR (consumer staples)
+25% buy VAW (materials) 2-way tie
+25% buy VFH (financials) 2-way tie
-
Thanks for the update. I checked over at Miles' blog to see if there was an update since it's been over a year from his last momentum post. He's not posting any more on that topic it seems.
-
Thanks for the update. I checked over at Miles' blog to see if there was an update since it's been over a year from his last momentum post. He's not posting any more on that topic it seems.
Yes, his blog as a whole has been really dead since he was banned from this forum for some poor communication choices. This momentum strategy is pretty interesting and M+1/2, I appreciate the regular updates.
It was at that time he started his miles hacking thing, and that has kept him busy, from what I understand.
-
I'm trying an experiment with a tiny fraction of my portfolio, where I see what it's like to apply momentum algorithms to a sort of tax-managed mini-fund.
Here's the criteria I imposed on this experiment:
* momentum involves lots of buying and selling, so I need to limit the costs involved with trades
* insignificant compared to my portfolio (under 1%)
* benchmark against S&P 500
* limit the tax impact
So what I decided on is sector rotation using Vanguard funds. That way I'm covering the market (roughly), trades are $0, and expense ratio of the holdings are also low. I allocate 20% gains over the year, and will skip buy/sell until this "tax gain budget" has enough room. I'll start posting results and observations later in this thread.
Important caveat: this experiment will prove nothing. Meaningful data would require years, and I don't plan to run it that long. The best fund is still the Vanguard Total Stock Market fund, not a momentum fund.
Good luck with the trading. I wrote a post about my advice for people who want to try out a trading strategy:
MOD EDIT: Spam link removed.
In short, keep it a small part of your portfolio, do it in a retirement account, know your edge over other investors, and track your performance. If you aren't doing well, quit trading and don't try to chase losses. Good luck!
-
WallStreetPhysician - I am not asking for advice in this thread. You seem more interested in quoting your website than understanding the purpose of the thread by going through the posts. Please don't tangent the thread into something it's not.
Just to refresh the difference between this experiment and other momentum approaches, this experiment is worse. It takes the 11 Vanguard sector funds and switches between them based on two momentum factors: sector with most gain, sector closest to it's 52 week high point. The switch happens every 5 weeks. Nobody has suggested sectors can be used to capture momentum, so this is truly an experiment in that regard.
I've pieced together the following performance from this thread, and from some scattered data:
2/17 - 10/19 : +10.3% S&P 500 and +6.2% sector momentum experiment (+4% S&P)
10/19 - 11/23 : +4% S&P 500 and +3% sector momentum (+1% S&P)
11/23 - 12/28 : +1.8% S&P 500 and +3.8% sector momentum (-2% S&P)
12/28 - 2/1 : +1% S&P 500 and -2% sector momentum (+3% S&P)
Overall I roughly estimate +18% for S&P 500 (pre-tax) and +11% for sector momentum (pre-tax) over the experiment so far (~50 weeks). Taxes are 15% for S&P 500 (assumes 1+ yr holding) and 25% for sector momentum (always sells within 1 yr), which gives S&P 500 +15% after tax, and sector momentum +8% after tax. I double checked the S&P 500 performance on Google Finance and also got +18%.
After almost one year, roughly +7% better performance from S&P 500 investing than sector momentum. This does not say anything about other ways to invest in momentum (individual stocks).
-
This experimental fund buys whatever sectors have momentum every 5 weeks. I plan to start using the term "5W" to describe this experimental fund, since it takes less words to say "5W fund". It does not imply this idea is smart or successful because it has a label - it fell behind the S&P 500 in the past 5 weeks, so "S&P 500" is a much better label to work with.
Performance momentum picked VFH (Financial) since it's performance was highest of the sectors. For stock price momentum, VGT (Info Tech) was closest to it's 52 week high. Both of these had no ties, so the fund switched to solely these two:
VGT, now 50%
VFH, was 25%, now 50%
VAW, was 25%, sold
VCR, was 50%, sold
Side note: poor execution probably added ~0.1% to the cost of two trades. Two trade executions were strangely outside the bid-ask range before and after the trade. The cause is less interesting than the fact that without the actual trade, this cost wouldn't show up.
Performance of S&P 500 in the past 5 weeks was about 4.3%, or 3.7% after LT tax.
Performance of 5W fund in past 5 weeks was about 2.1%, or 1.6% after ST tax.
In both cases, 5W fund fell -2% behind it's benchmark for the past 5 weeks.
-
MustacheAndaHalf, I appreciate that you continue to update this thread. It allows people to see how your experiment is working in real time, rather than a retrospective that has the potential to skew to whatever you wanted the results to be.
How long do you plan on keeping this experiment going, and are you doing so at this point to further the experiment, or because you really believe it's a better potential investment strategy than VTSAX or whatever broad index fund you want to use as a benchmark?
-
I mostly keep it going out of.. uh, momentum. I don't believe this idea will be profitable - it has too many compromises, and the S&P 500 is an excellent and unrelenting opponent to any active investing attempt. For example, the "financials" sector itself contains both stocks with and without momentum, but this experiment can only buy an entire sector at a time. So "pure momentum" funds can pick and choose their stocks, this 5W fund cannot.
But I actually continue it to learn. I don't often dwell on the tax aspect, but that's been interesting. I started wanting every aspect to show up - including these trades on my personal tax bill. So I set a maximum gain budget to limit the tax impact. My budget was too inflexible and was hit immediately. I learned to provide quarterly "max gains" and a sweeping amount for the year. It's also clear that the fund switches so often that everything is short-term taxable gains or losses. I have recently decided this isn't so interesting, and am moving the fund into a tax deferred account so it has no impact on my personal taxes.
Ranking all sectors by hand using two criteria is annoying. First I automated the calculation, but had to do my own sorting. Then I setup automatic sorting and tie detection, and it's much easier to know what to purchase next.
The 5W fund is embarrassingly short-sighted. For example, on day one of the experiment I bought the wrong sector. I was doing calculations by hand. The next day? It was the correct purchase. Sometimes the top ranked sector changes within hours, sometimes days. It's also interesting to me how often I make mistakes, and what precautions I need to make against that.
Overall I continue because of learning from it, and I should probably post more about that topic.
-
Per my post, imam curious for data on your tax, spread, and transaction costs as annual % numbers. Please post if you get a chance. For example, i have also observed the risks of ETF spreads being manipulated especially at market open resulting in trades outside the spread (hidden broker dealer profit sniping)
Per your earlier post, I started including tax brackets. Notice how I now mention "after tax" profit, using LT and ST tax rates. I assume 5W fund always sells within a year (I assume 25% tax) while S&P 500 is buy/hold (I assume 15% tax).
I never trade exactly at market open, but I have traded within 30 min of the opening bell. I definitely saw purchases outside the listed bid-ask spread this last time, but that's unusual. In general I don't track this specifically, but I do track every purchase made. So when the 5W fund pays more because of a larger spread, that will be reflected in profits (less bought, less profit).
-
Finished redoing spreadsheets on profits, with the following updated numbers. Profits are from 2/17/2016 through 3/10/2017, roughly 13 months.
5W fund, pre-tax +14.5%. With ST tax of 25%, +10.9%
S&P 500, pre-tax +22.4%. With LT tax of 15%, +19.1%
The "5W" sector momentum experiment trailed it's benchmark by about 8% both before and after taxes.
I have spreadsheets to track purchase, sell, and performance. The performance spreadsheet wasn't keeping up with ties well, since those were not part of the original experiment. Sometimes a sector pulls ahead of a prior tie, causing existing holdings to be increased. Now the prior position plus any buys and sells are represented in spreadsheet I use to track performance. It's once again automated and represents performance over the entire experiment (which started Feb 17, 2016).
-
The short answer is "no" - the S&P 500 benchmark beat sector momentum by 8%.
5W spent 40% of it's time in utilities, 20% in telecomm, etc. Applying those percentages to the annual return of each sector shows a return about halfway between sector momentum and S&P 500. Meaning the timing of sector momentum investments subtracted from returns.
The S&P 500 performed (+25%, year ending 2/28) roughly in the middle of the market's sectors: between the 5th ranked energy sector (+29.1%) and 6th ranked consumer discretionary (+18.4%). But sector momentum tied the 2nd worst sector of telecomm (13.9%), only beating consumer staples (+12.1%). It's amazingly bad to beat only 1 sector out of 11.
My definitions of momentum were taken from an MSCI white paper on momentum, but applied to Vanguard sector ETFs. I wouldn't meddle with the windows of time (52 weeks ignoring the past 4 weeks) used by academics to measure if the market exhibits momentum. That would entail contesting academics on their home turf to re-invent momentum.
One possible change could be the timing. But the whole goal of a momentum fund is to switch from investments that have lost momentum to those that have gained momentum. So making fewer purchase windows per year (say only 4 instead of 10) would result in less momentum investing. Since the 35 day window was selected to avoid "wash sales", making even more frequent purchases would involve making decisions about sectors that have capital losses but fall under the IRS wash sale provisions.
Another change could be the sampling of momentum. Instead of once every 5 weeks, momentum could be checked a number of times a day for a number of days. If yesterday utilities had momentum but today finance does, both could be purchased in equal measure since they had momentum equally during the measuring period. I have no data to decide if this helps, and it would certainly be more troublesome to automate.
Ultimately I don't have data pointing to changes that could improve 5W sector momentum. Based on the data I have, the S&P 500 index remains a superior investment.
-
I'm altering several polices of this experimental fund.
1) Purchases will be made twice as often, every 2.5 weeks. The fund previously bought every 5 weeks on Wednesday, with the next trade scheduled for Wednesday, April 12th. The next trade date will be Thursday, April 13th. The trade after that will be Monday, May 1st. The fund will continue purchasing every 2.5 weeks, alternating between Thursday and Monday.
2) Since trades can happen within 30 days of each other, wash sales are now possible. Wash sales delay realizing capital losses. In practice the fund exists in a tax deferred space, so this situation will need to be simulated based on the experimental "fund manager's" understanding of U.S. tax law and wash sales.
3) Certain sectors are 3% of the market, while others are more than 15%. I speculate these small sectors are more interesting, and should be included in ties more often than other sectors. To implement this speculation, if the tiny sectors get within 95% of the top sector's performance, they will be part of a tie. Other sectors will need to get within 98% of the top sector's performance to be part of a tie. The tiny/small sectors are those with about 3% of the market: REIT, utilities, basic materials, and telecommunications.
4) The worst sector will no longer influence ties. Performance will simply be the top sector's performance. If the top sector gained +20%, then ties will be based on reaching 0.98 or 0.95 of that performance (19.6% or 19.0% respectively).
Based on the past year, these changes could further negatively impact performance. By making trades twice as often, this "2.5 week" experiment could take twice the negative impact of momentum. It's "doubling down" on something that has failed to beat the S&P 500. Further, in the past year the best sector was finance, which will be part of fewer ties with the new rule. 3 of the 4 small sectors performed in the bottom half of all sectors in the past year.
The fund will retain it's benchmark and history (currently -8% vs S&P 500 for first ~13 months). It will retain the same calculations of momentum (per MSCI white paper on momentum). Trades will happen twice as often, and ties will favor small sectors.
-
First Thursday purchase (4/13) with next purchase slated for Monday, May 1. But today is a good day to call "May day!" as well, watching financials dive -6% over the past month.. and seeing this experiment keep financials for next time.
Looks like a defensive posture as information tech was sold, and consumer staples and consumer discretionary replaced it. Here's the new fund allocation:
50% financials (still), VFH
0% info tech (was 50%)
25% consumer staples, VDC
25% consumer discretionary, VCR
An ETF tracking consumer goods named "VCR" still amuses me each time. Over the past 5 weeks the experimental 2.5W fund lost -4% while the S&P 500 took less than a -2% loss.
I did not update the spreadsheets to reflect the new way ties are determined, nor make rankings based solely on the top sector. So some work to do before the next purchase date. The "2.5W" stands for 2.5 weeks. The next purchase date is May 1st, 2017.
-
First Monday purchase (May 1 2017) of 2.5W fund experiment. Both S&P 500 and this experiment gained about +2.5% the past 2.5 weeks. That still leaves this experiment at -10% relative to the S&P 500 index (it's benchmark).
Performance momentum picked info tech (VGT), while price momentum (highest price in 48 of last 52 weeks) hit a 4 way tie. So here's the purchases for this round:
12.5% consumer discretionary (VCR), was 25%
12.5% health care (VHT), was 0%
12.5% raw materials (VAW), was 0%
62.5% information tech (VGT), was 0% (performance pick, price momentum tie)
In particular, VGT was the top of both criteria. Both VCR and VHT were within 1% of the top, and VAW was a small sector within 2% of the top. The top in this case is the highest price VGT reached in the past 52 weeks (ignoring the 4 most recent weeks). Current price divided by highest price is the "price momentum" of a sector.
Next purchase date is May 18th per the new 2.5 week schedule.
-
The fund lost less than it's benchmark, about -0.1% vs -0.7% for S&P 500. The 2.5W fund picked some defensive choices of utilities and consumer staples for the next 2.5 weeks.
Sold 1/5th of info tech and all of consumer discretionary to buy utilities and consumer staples. The new allocation for 2.5W fund is:
50% info tech (reduced from 62.5%)
12.5% financials (unchanged)
12.5% raw materials (unchanged)
12.5% utilities (new)
12.5% consumer staples (new)
All holdings except info tech declined, but the nearly +1% rise in info tech balanced out the losses elsewhere. Next purchase date is June 5th.
-
Another incremental win over it's benchmark, with 2.5W fund gaining +4.9% versus +3.5% for the S&P 500 in the past 2.5 weeks. Most of the gains are from the large allocation to tech.
Half the experiment tracks price momentum, which is the current stock price of a sector divided by it's highest price over the past 52 weeks. When the market hits new highs, price momentum sends strange signals like it did today: it signaled to buy 7 of the 11 sectors. Essentially it didn't buy energy, financial, real estate, and telecom. So half the fund has diversified into 7 sectors while half the fund remains invested in VGT (info tech).
57.14% VGT
7.14% VDC
7.14% VCR
7.14% VHT
7.14% VIS
7.14% VAW
7.14% VPU
The past 2.5 weeks it beat it's benchmark (by +1.4% before tax), but overall it still lags the S&P 500 (by over 5-6%, roughly) as measured from the start of the experiment (Feb 2016).
-
The 2.5W fund/experiment isn't so much mine as running on it's own... momentum. I don't make decisions at all, excepting rounding up or down with buying shares selected by algorithm. This 2.5 week period lost ground to the S&P 500. The S&P 500 only lost -0.3% while the 2.5W fund lost -1.1%.
Because of tech sector performance, the momentum measures stubbornly refuse to pick any other sector for performance momentum. So the fund remains half invested in VGT, the Vanguard Information Technology ETF.
Price momentum went from a 7-way tie to a 4-way tie. This half of the fund is now spit between the following 4 funds:
VIS (Industrials) now 25%
VAW (Raw Materials) now 25%
VPU (Utilities) now 25%
VHT (Health Care) now 25%
Yesterday was technically the day to make these purchases so this purchase was one day late, on Friday instead of Thursday. The benchmark S&P 500 is still well ahead (5-10% range) of the 2.5W fund/experiment since it's inception in Feb of 2016.
-
Do you have an IRA? Why not do this within your IRA and then not worry about tax consequences.
Anyone interested in momentum investing might want to check out this site:
http://mebfaber.com/timing-model/
-
runewell - Earlier in this thread I mentioned that the fund was moved into a retirement account.
-
Looks like an unpredictable world will continue to punish the momentum measures used in this experiment. The benchmark S&P 500 lost about -0.5% since 6/23, while the 2.5W fund experiment lost roughly -1.2%.
The fund has now held VGT (Info Tech) for more than 2 months, which is 4 updates where it has not been sold off. This is caused by "performance momentum" tracking whichever sector has had the best performance in the past 52 weeks (minus the most recent 4 weeks).
"price momentum" (how close a sector is to it's 52 week high) made the following changes:
VHT (health care) sold off, now 0%
VPU (utilities) sold off, now 0%
VFH (financials) new position, now 1/6th of fund
VAW (raw materials) increased position, now 1/6th of overall fund
VIS (industrials) increased position, also 1/6th overall
And VGT (info tech) remains the other 50%.
The volatility and losses continue... next update on 7/27.
-
"Hm, that's odd. Why is the VIS holding half the size of the others. Uh.. and why is the fund holding 8% in cash..."
Unlike a textbook where mistakes get edited out, this fund/experiment keeps the mistakes. Last time, the fund was supposed to raise VIS from 12% to 16%. Funny story, though... you'd be surprised how small that "buy" and "sell" selection box looks... So anyways, instead of a 4% buy order? Yeah, issued a 4% sell order and didn't notice for 2.5 weeks. So... the fund actually held 8% VIS and 8% cash for the past 2.5 weeks.
Back to momentum investing, information technology retains its crown for performance. Performance momentum is stuck on VGT, Vanguard's Information Technology fund.
For "price momentum", a new 3-way tie emerged. The, ahem, VIS (industrial) holdings were sold, as were VFH (financial). Both make room for VCR (consumer discretionary) and yet more VGT (info tech). As if the fund didn't have enough info tech, it's now 2/3rds in that sector. So the new holdings become:
66.66% VGT (information technology)
16.67% VCR (consumer discretionary)
16.67% VAW (raw materials)
The fund and S&P 500 were roughly tied in raw performance. The fund gained +2.5% while the S&P 500 gained +2.4%. But after taxes are calculated, that's +1.9% for the fund and +2.0% for the S&P 500.
Hopefully the next update won't discover 8% sitting in cash!
-
Have you looked into dual momentum strategy as a way to decrease trading costs?
I had a chance to talk with Meb Faber a few days ago and he seemed to think it might be OK to try to capture some of the "trend" premium. Of course he has his own EFT(GMOM)/robo/white paper on implementing some momentum into asset allocation. Just google trinity portfolio. I find the idea that we can add another asset class(momentum) to both decrease risk and add return but I haven't pulled the trigger on it yet.
-
Have you looked into dual momentum strategy as a way to decrease trading costs?
I had a chance to talk with Meb Faber a few days ago and he seemed to think it might be OK to try to capture some of the "trend" premium. Of course he has his own EFT(GMOM)/robo/white paper on implementing some momentum into asset allocation. Just google trinity portfolio. I find the idea that we can add another asset class(momentum) to both decrease risk and add return but I haven't pulled the trigger on it yet.
I really like what Meb Faber has to say, and some of the podcasts on his site are interesting. His site advocates being in one of five asset classes if the price is above the 10-month moving average, making changes monthly. From 1972 onwards, this method was barely better than the S&P. When you go this route, it takes you a bit longer to buy into stocks when they are increasing, but then you also stay out of the market once it tanks. It has the effect of steadying things so that your overall returns are good with limited upside and downside.
He suggests in his book that you could do this with as many asset classes as you like. My only is that I don't think an environment of increasing interest rates is favorable for bonds so my portfolio is usually 90%+ stocks. So while I don't adopt the strategy exactly (especially since stocks look quite bullish on the chart), I do plan to be very cautious when these puppies turn direction and cross the 200-day moving average. At that point it wouldn't surprise me to be following all five asset classes more closely.
There are other momentum strategies as well.
-
Have you looked into dual momentum strategy as a way to decrease trading costs?
I need a two part answer for that, but I'll start with trading costs. To avoid paying any commissions, I selected Vanguard sector ETFs. The bid-ask spread is often about 1/5,000th of the stock price, so that hasn't proved costly, either. The SEC also charges a few pennies when securities are sold. Overall trading costs, using Vanguard sector ETFs in a Vanguard account, have been very low.
I've read the book Dual Momentum, but I have trouble believing any system that neatly avoids all past crashes. The system predicts crashes and switches into bonds, and then buys back in right as the crash has ended. From what I saw, it doesn't actually profit off a crash - it just stays out of the market. Avoiding every crash looks like curve fitting the historical data, so I don't trust going 100% bonds to avoid stock market crashes.
I initially read about momentum in Larry Swedroe's books, where he shows various time frames and countries where it has provided a risk premium. In this context, momentum means the ability to buy and sell individual stocks - whatever stock has momentum. Typically academic studies also "short" the stocks with poor momentum, and overall do not reflect the cost of trading. I've also read a white paper (by MSCI?) collecting various studies in one place.
Now the problem: this fund may or may not track momentum. I decided to use Vanguard's 11 sector funds as a way of dividing up the market, but still paying $0/trade. But as a side effect, I can only capture the momentum of an entire sector - hundreds of stocks. The stocks with good and bad momentum get lumped together, which at a minimum dilutes the effect dramatically.
The ETF sectors are selected based on two (plain) momentum criteria: one for "price" momentum, or how close a sector has gotten to it's 52-week high. In the last update, 3 sectors were very close on price momentum - within 1% of their highest point in the past 52 weeks. The other momentum criteria is "performance", or how well the sector has gained in the past 52 weeks. The white paper recommends ignoring the past month, so I subtract out the past 4 weeks of performance. The result is about 11 months of performance used to compare every sector. In the last update (and many before), Information Technology (VGT) has beaten out other sectors.
Overall the experiment documented in this thread aims to see how "sector momentum" will perform against the S&P 500. In general the S&P 500 continues to beat it over the ~17 months I've been running this experiment. So if I had to make a recommendation I would say invest in the S&P 500 instead of sector momentum.
-
In general the S&P 500 continues to beat it over the ~17 months I've been running this experiment. So if I had to make a recommendation I would say invest in the S&P 500 instead of sector momentum.
In general the S&P has been going up for the last 17 months, and in any case 17 months really isn't enough time to determine anything meaningful.
-
Miles Dividend MD was using Dual Momentum and posted one update before going radio silent on his blog. It would have been interesting to see the results unfold.
-
runewell - Agreed, but that's all the data I've got. If you find 5+ years of sector momentum performance somewhere I'd be interested in that data.
Retire-Canada - Note this thread is not about Dual Momentum, and I'd like to avoid switching to that topic.
Mr Rich Moose - Not aware of a white paper or study using that method, but I'd be open to learning about it. Do you mean that 3 month performance gets counted 3x (in 3 mo, 6 mo and 12 mo data)?
-
I'll make another shot at explaining what I meant. Since the method you describe adds (3 month + 6 month + 12 month), let's say you do this at the end of 2016...
Oct - Dec : get counted in 3 mo, 6 mo, and 12 mo data.
Jul - Sep : get counted in 6 mo and 12 mo data.
Jan - Jun : only get counted in 12 mo data.
Is there a white paper or book describing that method?
-
Finally the "performance momentum" has shifted - from information technology to the financials sector. But still, over the past 2.5 weeks the fund managed to lose -1.5% compared to only -0.4% for the S&P 500. So for this period, the S&P 500 pulls ahead by over 1%.
The fund makes the following changes, in percentage terms:
VCR was 1/6th and is now 1/10th (16.67% to 10%), consumer discretionary
VGT was 2/3rds and is now 1/10th (66.67% to 10%), information technology (finally!)
VAW was 1/6th and is now 0% (16.67% to 0%), raw materials
VIS is a new holding of 10%, industrials
VPU is a new holding of 10%, utilities
VFH is a new holding of 60%, financials
Next purchase is last day of August: Aug 31st. Hopefully the fund switching most of it's assets from information technology to financials will prove interesting.
-
Over the past 2.5 weeks, this momentum experiment took another small beating: it lost -1.5% while the S&P 500 gained +1.2%. Taxes actually benefit this specific scenario, since the IRS shares in losses but also dips into profits. So after tax that would be roughly -1.1% versus +1.0%. Still a 2% gap in a few weeks.
Performance momentum got boring again, switching from VFH (financials) back to VGT (information technology). A lot of churning on price momentum, reflected below:
VFH now 0%
VCR now 0%
VIS now 0%
VGT from 10% to 62.5%
VPU from 10% to 12.5%
VAW from 0% to 12.5%
VHT from 0% to 12.5%
Next update will be Sept 18th.
-
The 2.5W fund kept pace with the S&P 500! Gained about +1.8% vs 1.5% for S&P 500, which shrinks after consideration for taxes: 25% bracket for 2.5W fund (assume short term since selling happens every 2.5 weeks) and 15% bracket for S&P 500 (assuming long-term buy and hold behavior). After tax, roughly 1.4% vs 1.28% which is noise. But overall that still leaves the S&P 500 well ahead.
Rather boring update otherwise: the fund adds the "Industrials" sector to the list, changing a 4-way tie into a 5-way tie. New allocations are:
VGT at 60%
VHT at 10%
VAW at 10%
VPU at 10%
VIS now 10% (newly added Industrials sector)
Note that when the fund doesn't make many changes, it resembles the original 5W fund that only updated every 5 weeks. The fund will assess momentum again on Oct 5th, and make any changes based on what happens.
-
Just wanted to say a big Thank You for keeping up with the updates! I've been enjoying following along :)
-
Xlar - Thanks for the vote of support. I think I'm reaching 20 months in a week or so.
To summarize my understanding, the S&P 500 is a benchmark. Anything has to prove itself against that benchmark, and hopefully this experiment gathers some evidence (for or against) about how sector ETFs and momentum compare against the S&P 500 benchmark.
Although the 2.5W fund did well (+1.3%), the S&P 500 did slightly better (+1.4%). Catching a benchmark is hard if a good month results in a tie. Performance momentum switched half the fund from VGT (info tech) to VFH (financials). The rest of the fund went from a 5-way tie to a 6-way tie. VPU not only performed poorly, it lacked momentum and dropped from the fund list (for now). Here's the changes:
VGT (info tech) was 60% now 8.33%
VFH (financials) was 0% now 58.33%
VHT was 10% now 8.33%
VIS was 10% now 8.33%
VAW was 10% now 8.33%
VPU was 10% now 0%
VCR was 0% now 8.33%
So the fund is mostly financials (VFH), and a tied group of VGT/VHT/VIS/VAW/VCR. At some point I plan to calculate how far 2.5W fund is behind it's S&P 500 benchmark (at least 5%, maybe 10%?). Early this year it was behind by at least 5%, and has rarely had a month beating the S&P 500, so I typically don't bother with the exact calculation. In theory it could be reconstructed from the data in this thread and a lot of patience.
Next update in about 2.5 weeks (this the 2.5W fund naming), on October 23rd.
-
spoiler/tldr - Both 2.5W fund and S&P 500 gained about +1% since 10/5/2017.
A 6-way tie wasn't enough - now it's a 7-way tie over price momentum. Too many funds are at their high point relative to the past 52 weeks, causing a lot of ties. Sadly, it's easier to describe what the fund does NOT hold than what it does.
The fund does not hold VDC (consumer staples), VDE (energy), VNQ (real estate) and VOX (telecom). Which leaves the fund holding the following...
57.14% financial (VFH)
7.14% consumer discretionary (VCR)
7.14% health care (VHT)
7.14% industrial (VIS)
7.14% information technology (VGT)
7.14% raw materials (VAW)
7.14% utilities (VPU)
Note the actual share prices and number of integral shares don't actually equal 7.14% each, but it's easier to represent them like this. It's interesting that despite the fund's concentration in financials (and before that info tech), it didn't pull ahead of the S&P 500. And overall in the 20 months of the fund (and this thread), it has trailed the S&P 500 by roughly 5-10% in total. When a dramatic change occurs, I plan to calculate the exact amount.
Another 2.5 weeks lands on Nov 9th for the next purchase.
-
One correction (I don't want to edit my prior posts, so they remain a record of what happened at that time): I started this thread on May 18, 2016 rather than mid Feb. The momentum experiment is 20 months old, but this thread is 17 months old.
I'd also like to summarize the current setup and mention changes from the original.
Originally the fund was in a taxable account with a "budget" for it's gains. Now the fund is in a retirement account with no concern for gains. Despite the actual location of the funds, a taxable gain of 25% is used to project how the fund would perform after taxes. The benchmark remains the S&P 500, which is presumably bought and held - so it's tax bracket is 15%.
The fund uses price and performance momentum, per a white paper by MSCI (that now requires name & email to download). Price momentum is current price divided by the highest price in the past 52 weeks. Everything at a 52 week high has 1.00 price momentum. Performance momentum is the gains over the past year divided by the asset's starting point. Because "reversion to the mean" is significant to momentum, the past 4 weeks are subtracted out and ignored when calculating momentum - see the white paper for a full description of these algorithms, data, and reasoning.
I picked Vanguard sector funds owing to their $0/trade and low expense ratios. They also cover the entire market, split into 11 funds. It would be better to rank the momentum of the entire market and keep only the highest momentum stocks - but that's too exhausting and costly. So this fund uses Vanguard sector funds to track momentum.
-
hi, cool experiment. Perhaps I missed it, but what are the rules for rotating in and out of a sector? I couldn't find it other than mentions of price momentum and performance momentum, do you take the average of the two?
and when you say "starting point", what do you mean exactly? price of jan 1?
Sorry if I missed something, Thanks.
-
Sure, I'll explain. The fund uses 52 weeks of data, so the "starting point" is the data from 52 weeks ago: the performance since that time, or the highest price since that time. Half the fund is purchased using price momentum, and half gets allocated based on performance momentum.
First performance momentum. To avoid "reversion to the mean" effects, the past 4 weeks are subtracted from the 52 week performance. As of right now, "financials" (VFH) has a 52-week performance of +35.94% and a 4-week performance of +4.46%. 35.94 - 4.46 = +31.48%. The next closest sector is information technology (VGT) with 32.76 - 4.40 = +28.36%. That's too far away for a tie, so "financials" is held to represent "performance momentum". As soon as another sector beats financials, the entire allocation gets switched to the new sector (unless they are so close there is a tie).
Price momentum can't be subtracted easily, so I compare the current price to the highest price in the past 52 weeks. For raw materials (VAW), that's $133.56/share divided by a highest price of $133.76, or 99.85% of it's 52-week high. There are several other funds over 99% of their 52-week high price, so a lot of ties with the #1 price momentum sector. All of those get purchased. A few smaller sectors only need 98% of their high to form part of a tie, which is why utilities (VPU) is also part of this tie between 7 sectors.
Every update, the fund sells everything that no longer matches it's momentum criteria, and buys everything that meets the criteria. The idea of momentum is that you hold an asset while it makes gains faster than other assets, and sell at a certain interval when something else is doing better. I'm not especially convinced that sector momentum is a useful strategy, but I'm running this experiment (for 1.7 years) to see where it goes. So far, it's trailed the S&P 500.
-
hmmm I see, thanks.
-
Another win for the S&P 500, which gained slightly (+0.12%) while the 2.5W fund lost ground (-1.5%). Momentum does badly in volatile times because the "signal" of rising prices is actually just rapid ups and downs of the market.
For while financials and information technology have taken the #1 spot on performance momentum - and now they're tied. So the "performance momentum" half of the fund's assets get split between VFH and VGT.
A new entry! For the first time, the fund bought VNQ ("real estate"). Five sectors are close enough to their 52 week peek price to form a 5-way tie: VNQ, VPU, VCR, VAW, and VGT. Because momentum can change quickly, if I had delayed my purchases one hour this would have only been a 3-way tie.
So the new fund allocation is as follows:
35% VGT (info tech), for both price and performance momentum
25% VFH (financials), for performance
10% VPU (utilities), for price
10% VCR (consumer staples), for price
10% VAW (raw materials), for price
10% VNQ (real estate), new entry, for price
If I had to make an educated guess, I think the S&P 500 is ahead by roughly +10% compared to the 2.5W fund. Some day I plan to make some cross-spreadsheet calculations that will make the full calculation easier.
Next purchase date is November 27.
-
The 2.5W fund gained 1.7% versus 0.6% for the S&P 500. The lead shrinks after taxes (+1.3% vs +0.5%), but it's still a rare (random?) win for the 2.5W fund.
I checked momentum 3 times today, and in 2 of 3 times VPU was part of a 5-way tie for price momentum. One time, it dipped low and made for a 4-way tie. So overall I kept it in (2 out of 3), but if my timing was different, it would be out.
I track all changes in value on a spreadsheet that I also use to calculate the profit of the fund. It would be better to have a separate place where the value of the stocks can be added to the cash holdings. Further, I'd like to roll up the old spreadsheet data into this, so I don't have to go back through time by hand to figure it out. And finally, the new spreadsheet should list the S&P 500 gains. Essentially I should automate the profit of both the fund and it's benchmark, going from the start of the fund to the present day.
A relatively boring update, otherwise: only performance momentum changed. The fund's 25% holding of VFH (financials) was sold and used to buy VGT (info tech). The fund now consists of:
60% VGT (info tech, was 35% + 25% VFH sold)
10% VCR (consumer discretionary)
10% VAW
10% VNQ
10% VPU
Both VGT and VCR made decent gains and are the main reason the 2.5W outpaced it's benchmark for 2.5 weeks.
Next update is on Dec 14th. (The update after that will be delayed by the New Year's Day closure of the NYSE, and will occur on Jan 2, 2018)
-
Ow, ow, ow. It's a decent summary of a benchmark (+2.6%) beating the 2.5W fund (-1.8%) by +4.4%. Taxes bring these two slightly closer together: the 2.6% gain is taxed at 15%, making the S&P 500 gain only +2.2% after taxes. And capital losses are shared by the IRS, so that turns the 2.5W fund's performance into a -1.4% after taxes. Since Nov 27, the 2.5W fund lost against it's benchmark by -3.6% after taxes.
My update is one day late, so I used data from yesterday's market close to make the fund purchases. The fund sells off utilities and 1/6th of information technology in order to make room for health care and industrials.
The fund now holds:
50% VGT (info tech) (was 60%)
0% VPU (utilities) (was 10%)
10% VCR (consumer discretionary)
10% VNQ (real estate)
10% VAW (raw materials)
10% VIS (industrials) (just added, was 0%)
10% VHT (health care) (just added, was 0%)
I also found overall performance numbers lurking in a spreadsheet I already set up: for the benchmark, from Feb 17 2016 until today the S&P 500 gained +37%. Over the same time period, the 2.5W fund gained +18%. With tax rates of 15% (long-term for S&P 500) and 25% (short-term for 2.5W fund) that comes to +31% after tax (S&P 500) and +13% for the 2.5W fund. Hopefully it's easy enough to keep this running that I reach the 2 year mark with this experiment.
The next update will also be one day late, but by necessity: the stock market will be closed on New Year's day. So next update on Jan 2nd.
-
Hi MAH,
Been digging around in the momentum studies since I found your thread, as it is interesting to me.
I don't mean to sidetrack your thread but I came across an AQR paper: a century of evidence on trend-following investing. Versions of 2014 and 2017 exist, regardless, they tell the same tale.
It is curious to me that per AQR paper trend-following (momentum) is the unequivocal winner in all situations (including out of sample tests in 2017 ver.), yet it appears less so in your experiment. I am wondering that if you have any comments, perhaps regarding the length of time-series or how the momentums were quantified in each experiment that could result in the discrepancies. You can PM me if you feel it is more suitable. Thanks.
ps. I am not affiliated with the AQR paper in any way or form.
-
I read some of the AQR paper just now. The 2.5W fund experiment goes 100% equity into some number of U.S. market sectors. The AQR paper includes commodities, international markets and probably other investments. It also uses volatility weighing to manage risk, and will short an entire market if that market has a negative trend. Lots of differences there, some of them interesting.
Going back to the setup of this 2.5W fund experiment, I didn't want to pay fees and wanted it simple. Using Vanguard sector funds accomplished both goals: the market is split into 11 (non-equal) sectors, each of which costs $0 to buy or sell. But a mutual fund tracking momentum would not be subject to these restrictions: both the complexity and fees can be higher, in order to capture more of the market's momentum.
Adding complexity and paying fees might capture more momentum. Buying individual stocks involves paying fees. I suspect "motif investing" would work best, since that allows up to 30 stocks to be purchased for a single $10 fee. On the complexity side, this would mean tracking the performance of hundreds (thousands?) of stocks and buying the "momentum winners" and selling any that fall out of favor.
Overall, I think the 2.5W fund experiment involves baskets of stocks with mixed momentum (U.S. stock sectors), and so can't capture momentum as directly as other methods. It also does not short any stocks, so it loses out on that traditional momentum measure as well.
Next update in a couple days...
-
Purchases and sales for the fund are normally recorded in two separate places, but the last purchase was only recorded in only one place. The cash position of the fund is nearly always below 1%, but the exact amount will require going over purchase records to determine (the fund is co-mingled with other assets, and tracked separately). To avoid problems, I made overly conservative purchases and left the fund in a roughly 1.5% cash position.
The S&P 500 lost a little bit since Dec 18, while the 2.5W fund had a gain. The accurate calculation of that gain awaits both updating the last set of purchases, and redoing the spreadsheets to include the cash position. Overall the fund probably had an insignificant gain and still trails the S&P 500 index.
The only change was to sell off the real estate sector (VNQ) which lacks enough momentum. The 5-way tie in price momentum becomes a 4-way tie. Information Technology remains the leader of performance momentum, and is retained.
Here's the fund's composition after today's purchases:
50% VGT (information technology)
12.5% VAW (raw materials)
12.5% VCR (consumer discretionary)
12.5% VIS (industrials)
12.5% VHT (health care)
00.0% VNQ (sold off real estate, which lacks momentum)
Hopefully I will get the records cleared up before the next purchase on Jan 18th. With some luck, I might even get a profit/gain calculation that includes the fund's cash position.
-
Spreadsheet fixed, and now I include the cash position of the fund. In the 22.5 months since this experiment started, the S&P 500 has gained +40% while this fund gained +20%. About the same after taxes: +36% (long-term tax rate of 15%) S&P 500, and +15% for the fund (short-term tax rate of 25%). A successful fund doesn't need to change much, but this fund has fallen 20% behind it's benchmark.
My mindset is so focused on indexing, that I added a restriction to this fund that doesn't have any inherent purpose. I selected 11 Vanguard sector ETFs that comprise the US stock market. But I don't need to cover exactly the stock market to track momentum - I can use any fund that exhibits greater performance than other funds.
Most of my updates involve changes to "price momentum", which compares the current price of an asset against it's highest price in the past 52 weeks. Every asset at it's 52-week highest price is a tie, so I handle ties by buying them all. But if I drop price momentum, and drop ties, the fund will be far easier to manage.
This experiment hits it's 2 year anniversary next month. I will keep the fund's existing policies for the purchases on Jan 18 and Feb 5. But I will make changes starting with the Feb 22 purchase date.
Starting on Feb 22 2018, the fund will still purchase "Vanguard ETFs", but no longer be limited to sector ETFs. The fund will track all Vanguard stock ETFs, and purchase whichever one exhibits the best performance (52 weeks minus last 4 weeks). This means the fund could buy US utilities one week and emerging markets the next week. It will not buy bond funds, only stock funds, and will only buy Vanguard ETFs.
Currently half the fund's assets are invested according to "price momentum" (how close the asset is to a new 52-week high water mark). Starting on Feb 22, 2018 the fund will sell off these assets and drop "price momentum". 100% of assets will be purchased based on "performance momentum", and with no ties. The fund may perform worse without "price momentum", but it will be far easier to manage the fund.
These changes probably invalidate the fund's benchmark. A fund that can invest in emerging markets does not have the same level of risk as the largest 500 publicly traded companies in the U.S. I don't like changing benchmarks, and I can't know in advance if the fund will even buy international stock. So while it's imperfect for the fund's new risk profile, I plan to keep the S&P 500 index as the benchmark until that no longer seems appropriate.
Overall, expect normal updates on Jan 18 and Feb 5, and then a big set of changes on the Feb 22 update.
-
In the past 2.5 weeks, the 2.5 fund gained about +6% while the S&P 500 gained +4% since the last update. Doing +2% better than a benchmark is great, but it needs to be put in context of the overall experiment (now ~23 months old): overall +26% for 2.5W fund, +44% for S&P 500.
For those wondering about stability, I suspect the policy changes I'm making in February (per prior post) will make the fund far more stable. Half of the fund tracks performance momentum, with no changes this update.
The other half of the fund tracks how close each sector is to it's 52 week high. That half of the fund went from a 4-way tie to a 7-way tie (out of 11 sectors). The constant changes in this half of the fund will go away on Feb 22 when the fund drops this momentum measure.
The funds holdings are now:
57.14% VGT (was 50%, added to existing holding), information technology
07.14% VCR (was 12.5%), consumer discretionary
07.14% VAW (was 12.5%), raw materials
07.14% VHT (was 12.5%), health care
07.14% VIS (was 12.5%), industrials
07.15% VDC (new holding), consumer staples
07.15% VFH (new holding), financials
Changing the 4-way tie to a 7-way tie required selling from 4 funds, and buying from 3 other funds. And trying to round up or down each fractional share to minimize the amount of idle cash. Apologies for not listing exact percentages above: once ties go away, the fund will be 100% in one fund at a time.
Note of all Vanguard ETFs, VGT (info tech) has the most price momentum. If the fund were using the new policies, it would be 100% in VGT. Those changes go into effect on the Feb 22nd update.
Next update on Feb 5th.
-
The momentum experiment is only using "sector momentum" one last time, today. On the next update, the fund will be allowed to follow momentum in any Vanguard ETF. And it will drop the use of "52 week high" momentum, which triggers when a sector/ETF gets close to it's highest price from the past 52 weeks.
The fund beat it's benchmark again by losing less. The S&P 500 dropped to a total gain of +41.9% while the momentum experiment's total gain is +24.3%. While the last 2 updates have chipped away at the benchmark's prior lead, I doubt most people would volunteer for losing -6.4% a year for two years. That's the fund's performance as measured against its benchmark over the experiment so far.
The fund dropped 3 ETFs as health care (VHT), raw materials (VAW) and consumer staples (VDC) all lost momentum. That results in the following allocation:
62.5% VGT, information technology
12.5% VCR, consumer discretionary
12.5% VFH, financials
12.5% VIS, industrials
Next month, using only performance momentum without ties will result in the fund holding only one ETF. If the fund were altered today to have those policies, it would be 100% in VGT, which is beating out both emerging markets and mega-cap stocks for the top performance spot. Those changes happen on the next update, scheduled for Feb 22.
-
Well, that was strange. On Jan 4 2018 I announced policy changes for this fund, which includes dropping "price momentum" and using only "performance momentum". And for the past 12+ months "price momentum" always had ties... until today, the very day I announced would end ties. So that's a bit strange, but at least it creates a neat transition from the old policy to the new one.
Over the 2 years of the experiment so far, here's how the fund has performed:
2.5W (formerly 5W) momentum experiment: +24.20% before tax, +18.15% after tax
benchmark S&P 500 index: +40.15% before tax, +34.13% after tax
(note momentum fund sells often so uses a short-term 25% tax rate. The benchmark is assumed to be a "buy and hold" strategy that uses long-term capital gains of 15%)
Roughly speaking, it's like the fund earned +9%/year while the S&P 500 earned +18%/year.
The fund no longer allows ties, and solely uses "performance momentum". Because of spreadsheet limitations, I can only track ETFs with a corresponding mutual fund. So Vanguard's suite of new "factor ETFs" will not be used by this experiment. So this momentum experiment will not be purchasing Vanguard's new momentum ETF.
Vanguard has 39 ETFs which also have a mutual fund. I'm tracking all of those on a spreadsheet for momentum. I take their 52 week performance, subtract their most recent 4 week performance, and the result is the momentum of that ETF. Here are the top 3 Vanguard ETFs ranked by momentum right now:
Information Technology (VGT) +32.76%
FTSE Emerging Markets (VWO) +27.20%
FTSE Europe (VGK) +27.15%
Since VGT has the most momentum, 100% of the fund's assets go into that fund:
100% Information Technology (VGT)
The next update is scheduled for March 12, but I'll check back to answer questions.
-
No change: Information Technology (VGT) has the highest 52 week (minus 4 week) performance of Vanguard's ETFs. The fund remains 100% invested in VGT.
In a refreshing change, the fund gained +5.84% since 2/22 versus it's benchmark +2.53%. That +3.3% gain against it's benchmark still leaves it well behind overall.
Next update is March 29th.
-
Unchanged. VGT, Vanguard Information Technology remains the 100% holding of the 2.5W fund experiment. Tracking the overall totals since the experiment began (Feb 2016):
S&P 500 index, +34.75% (pre-tax)... +29.5% after paying 15% LTCG tax
2.5W fund, +20.57% (pre-tax)... +15.4% after paying 25% short-term capital gains tax
Next update 4/16
-
Here's the top 3 Vanguard ETFs by performance (52 week - 4 week performance):
(1) VGT , 12 mo @ 31.75% , 1 mo @ -5.01% , 31.75 - -5.01 = 36.76 momentum
(2) VFH , 12 mo @ 20.62% , 1 mo @ -5.63% , 20.62 - -5.64 = 26.25 momentum
(3) MGK , 12 mo @ 20.82%, 1 mo @ -4.92%, 20.82 - -4.92 = 25.74 momentum
Since VGT is ahead by over +10 on momentum, the 2.5W fund experiment remains 100% invested in VGT (Technology).
It's really dramatic the change to the fund since I dropped the "52 week price" criteria for momentum. Any sector can approach it's own 52 week high point, and tie other sectors. But only one fund has the best performance, and if the gap is wide enough other ETFs find it difficult to catch up. At some point this "trend" or "momentum" will reverse, and a new ETF will be picked by the algorithm.
The fund currently invests everything in whichever Vanguard ETF has the best performance in the past year, ignoring the past month. Overall 2.5W fund profit since Feb 2016 is +24% vs S&P 500 (benchmark) profit of +37% (both before tax).
On May 3rd the fund will check to see if it remains 100% in VGT or picks another ETF.
-
Unchanged. The 2.5W momentum experiment remains 100% in VGT.
According to morningstar data, VGT grew +3.7% in the past month while VOO (S&P 500 index, the benchmark) gained +2.2%.
Here's the current momentum calculations for the top 3 Vanguard ETFs:
VGT, Vanguard Information Technology, 24 - 0 = 24 momentum
VWO, Vanguard Emerging Markets, 17 - (-2) = 19 momentum
VPL, Vanguard Pacific, 18 - 0 = 18 momentum
Vanguard announced that VGT (and other funds) have been going through a transition to a new index. As part of that transition, Facebook and Google will be removed from VGT and added to another sector fund. So while this experiment has held VGT, VGT itself has been changing contents.
Next update is May 21.
-
Still 100% invested in VGT.
As the last few months demonstrate, picking the ETF by performance is much more stable than picking the ETFs that are closest to their highest price. Information Technology still leads, with "finance" and "small cap growth" in the 2nd and 3rd place spots.
Next update on June 7.
-
Here's the 52 week performance of a couple interesting ETFs:
+30.8% Information Technology ETF (VGT)
+21.8% Finaicials ETF (VFH, the 2nd highest return of Vanguard ETFs I track)
+16.3% S&P 500 Index ETF (VOO, the benchmark of this experiment)
It took me two years to dump other criteria and solely focus on 52 week performance (minus 4 week performance). If I'd done that sooner, the 2.5W experiment might be ahead.
Instead, the overall raw performance for the past 2.3 years:
S&P 500 benchmark +43.2%
2.5W fund experiment +34.2%
So the raw performance has pulled back within 7% of it's benchmark. The other new twist is the fund doesn't make changes often. The fund would have held VGT for over 12 months, at which point tax on the sale uses long-term capital gains rates. So the changes made in Feb 2018 not only make the fund's raw performance better, but probably improves the after-tax performance as well.
Or as Ice Cube might say: "Dropped four funds and now I'm making all the dough"
-
I forgot to state it explicitly:
2.5W experiment remains 100% in VGT, Information Technology.
Next update on June 25th.
-
The fund remains 100% in Vanguard Information Technology (VGT).
For the past 13 weeks (1/4th year), VGT is +10.2% and it's benchmark S&P 500 is +7.0%. Overall the fund's performance since Feb 2016 has been -11% behind it's benchmark. The combination of the markets, and the policy changes in Feb 2018 have improved performance.
VGT is currently in transition. Vanguard is moving companies like Apple out of the index, and reclassifying those companies as consumer companies. The ideal approach would be to split out the individual stocks and measure their aggregate performance as two separate funds, but that's far too much effort for this. So I plan to just keep using the performance data of the existing Vanguard ETFs, and ignore the underlying stock changes.
Here are the top 3 ETFs at Vanguard, by momentum score. Momentum score is 52 week performance minus 4 week performance.
1) VGT @ +34.62% - (4.78) = +29.84 (info tech)
2) VDE @ +18.06% - (-3.68) = +21.74 (energy)
3) VBK @ +25.71% - (4.70) = +21.01 (small cap growth)
-
Been following this thread for a while, especially considering I'm thinking of running some of my own experiments.
I have one question though: why two and a half weeks?
Why not 2 (get an whole amount of periods in a year) or 4 (13 periods in a year) or monthly?
-
Originally, it was the "5W fund" experiment, which stood for 5 weeks.
I expected high turnover, and I wanted to avoid wash sales (selling at a loss within 30 days of buying). So that pushed the time out past 31 days. I originally wanted the fund to make purchases on the same day of the week, so I picked 35 days (5 weeks).
I moved the fund from taxable to a retirement account, and no longer needed to worry about wash sales (no taxes involved). I also wanted to see if checking for a change in momentum more often would help the fund better track momentum. So I cut the time in half, and the fund became the "2.5W fund" experiment.
Probably less useful than you hoped for your own experiment. Do you worry that certain companies announce quarterly earnings at the same time every quarter, which might bias which companies look good? Earnings tend to move company stock prices, so you might see the same company's stock moving more than average if you check the same time every quarter.
-
I figured it might be a US thing.
I live in the Netherlands, we have only a wealth and dividend tax, so trading frequency is not a factor for tax advantages.
I'm still working out the details, I have accounts with two, soon three brokers, each with different advantages in fees and available products, so I'm still looking to see what kind of momentum portfolio I can make with any of the three brokers and then decide on the implementation details like how often to rebalance. I was considering a non nice number to rotate trading days / cycle within earnings quarter. On the other hand I feel like I might be comparing apples to peaches if I keep changing the weekday and time relative to earnings, so there might be some drag/inaccuracy from that as well.
One thing I'm thinking of is what to do if the highest momentum score is negative, logically speaking the conclusion should be to pull out of the market and wait for a non negative momentum?
-
I originally learned a lot from reading an MSCI white paper on momentum:
https://www.msci.com/www/research-paper/introducing-momentum-a/0535854138
There's also academic research on stock markets that explores how momentum explains some of the total return of the stock market. It's a "factor" in factor investing. So you can read books or white papers and probably learn enough there.
From what I've read, you can pick "3 month momentum", which is just the highest performing asset according to 3 month performance. There might even be a "5 year momentum" effect I saw in one paper, but I settled on 12 month momentum. In this "2.5W fund" experiment, momentum is 52 week performance minus 4 week performance.
I originally made the mistake of trying to pick the market sector with the best momentum. I thought in terms of indexing, and slicing up the market. But that's just a limitation of my thinking. This year I switched to "any ETF at Vanguard", which means I can pick "small cap growth" if that's the best performer. So far, Information Technology has so much momentum that nothing has overtaken it, but I recommend not limiting your thinking to specific market sectors like I did initially.
Supposedly momentum can be negative as well. If you pick the worst performer, it's more likely to keep going down. So if you pick a fund or stock with negative momentum, there may be a tendency to lose money. Check books and white papers for how you want to handle that. More likely, it won't come up that often. You could also try "double momentum" which switches to bonds if bonds have better momentum than stocks. But delving into that is a topic for another thread, so I'll just leave that as a possible book for you to read.
Here's roughly how this 2.5W fund works:
spreadsheet shows performance for 52 weeks minus 4 weeks for every Vanguard ETF.
I use a sorting function to order the list by (52 week - 4 week performance), so it's easy to pick the best momentum ETF at Vanguard.
I use Vanguard since Vanguard ETFs cost $0 to trade at Vanguard. If your situation with fees in the Netherlands is more complex, you might need another solution.
Overall, the S&P 500 still beats this momentum experiment by over 10% in 2.3 years. So if you try something similar, use money you don't need to depend on.
-
The 2.5W fund remains invested in VGT.
None of Vanguard's other ETFs have come close in terms of performance momentum. Remaining invested in VGT has regained some of the ground lost to the S&P 500 benchmark, but the 2.5W fund remains -9.2% behind (since Feb 2016).
Vanguard is providing an alert concerning VGT:
"NOTE: In anticipation of upcoming changes that MSCI is making to select sector benchmarks, this fund will temporarily track a transition benchmark."
I wanted to see what I could learn with a few clicks, and wound up downloading MSCI's spreadsheet of which companies will move between sectors. Google and Facebook are in the list, and migrate from information technology (that's VGT) into communications services (at Vanguard, VOX).
VGT's Google holding ($964 million) is almost the same size as the entire VOX ETF ($988 million). Although GOOG is quickly becoming VOX's largest holding, it's not enough. So the transition index allows Vanguard to sell GOOG and FB shares - possibly to other Vanguard funds - before September arrives.
The question for the 2.5W fund is this: how does dropping GOOG and FB from VGT impact it's momentum? Here's 52 week performance minus 1 month performance:
VGT: +31.73% - (-0.28%) = 32.01 mtm
FB: +30.44% - 5.74% = 24.7 mtm
GOOG: +24.06% - 2.12% = 21.94 mtm
It looks like VGT has higher momentum than FB and GOOG. So when FB and GOOG migrate out of VGT, I'm estimating that VGT gains momentum. VGT is not relying on GOOG and FB for it's performance, so that lays one concern to rest.
Next update July 30th.
-
Despite today's -2.3% losses (so far) for VGT, the 2.5W momentum fund remains 100% in VGT.
The white paper I read long ago on momentum probably mentioned "reversion to the mean", or the tendency for stock movements to reverse. It's not predictable, so the momentum algorithm I use just waits it out. The prior 4 weeks are ignored, and the remaining performance over the past 52 weeks gets used. So for the next 4 weeks, today's price change won't be reflected in the momentum calculations for this 2.5W fund experiment.
Here's the top few Vanguard ETFs in terms of 12 month momentum (as of today):
#1 VGT, information technology
#2 VDE, energy sector
#3 VBK, small-cap growth
#4 VCR, consumer discretionary
#5 MGK, mega-cap growth
VGT still holds quite a lead, at +23.89 momentum versus others all in the +16 to +19 range. So today's performance, by itself, is not enough to shift momentum to another Vanguard ETF.
We'll see how volatile the market can be over the next 2.5 weeks....
Next update on Thursday, Aug 16.
-
VGT, at a new high, remains 100% of the fund's investment.
VGT recovered from the last drop and is now back to -9% behind the S&P 500 (overall, since Feb 2016). Some of the tech stocks have probably migrated out of VGT by now, although that won't be clear until Vanguard publishes an update.
An interesting observation: When I started this experiment, I decided on funds that cost $0/trade at Vanguard. At the time, that meant only buying Vanguard funds. But Vanguard plans to offer ~1800 ETFs for $0/trade some time in August 2018. So in the near future, the fund could buy from a much larger selection of ETFs for the same $0/trade.
A technical admission: the fund actually tracks 39 Vanguard mutual funds that also have an ETF share class. Google Sheets cannot track ETFs directly, only mutual funds. So I've used mutual funds as proxies for the ETF with the same name and assets. But that won't be possible for the ~1800 non-Vanguard ETFs.
I don't plan to change anything for the September 3rd update, but I might make changes 1-2 updates after that. Since I don't have to worry about people "front running" this fund, I'll reveal my probable source of information here:
http://etfdb.com/compare/highest-52-week-returns/no-leveraged/
That list shows the top 100 ETFs in etfdb's database ranked by 52-week performance. I will need a way to subtract their 4 week performance, but I'll probably do that by hand initially. So maybe late September or mid-October I'll start relying on that page to select ETFs.
In the meantime, the fund remains 100% invested in information technology (VGT).
Next update on September 3rd.
-
Two notes: "front running" in this case would mean someone sees my update, looks at the list I published, and then buys the top ETF listed there. They then wait for me to buy it, and sell on the "bump" I give the ETF due to a large purchase. But this fund is too small to make any market impact.
And second, I already own PSCH outside of this fund. So if you decide to buy PSCH because it's at the top of that list, you could be indirectly benefiting another holding of mine. If you watch financial news shows, you'll see people admitting if they own shares of whatever stock they discuss. Also, you can look at etfdb.com's other lists, rather than the one I mentioned above... they have 13 week performance, 5 year performance, and even leveraged ETFs.
http://etfdb.com/compare/
-
VGT, at a new high, remains 100% of the fund's investment.
VGT recovered from the last drop and is now back to -9% behind the S&P 500 (overall, since Feb 2016). Some of the tech stocks have probably migrated out of VGT by now, although that won't be clear until Vanguard publishes an update.
An interesting observation: When I started this experiment, I decided on funds that cost $0/trade at Vanguard. At the time, that meant only buying Vanguard funds. But Vanguard plans to offer ~1800 ETFs for $0/trade some time in August 2018. So in the near future, the fund could buy from a much larger selection of ETFs for the same $0/trade.
A technical admission: the fund actually tracks 39 Vanguard mutual funds that also have an ETF share class. Google Sheets cannot track ETFs directly, only mutual funds. So I've used mutual funds as proxies for the ETF with the same name and assets. But that won't be possible for the ~1800 non-Vanguard ETFs.
I don't plan to change anything for the September 3rd update, but I might make changes 1-2 updates after that. Since I don't have to worry about people "front running" this fund, I'll reveal my probable source of information here:
http://etfdb.com/compare/highest-52-week-returns/no-leveraged/
That list shows the top 100 ETFs in etfdb's database ranked by 52-week performance. I will need a way to subtract their 4 week performance, but I'll probably do that by hand initially. So maybe late September or mid-October I'll start relying on that page to select ETFs.
In the meantime, the fund remains 100% invested in information technology (VGT).
Next update on September 3rd.
I have plenty of Google Sheets tracking ETFs in them.
-
You misinterpreted my usage to match yours. If you only track today's price, you can use GOOGLEFINANCE() to lookup 1 day of data. But that same call fails to work for 52-week and 4-week returns. Try inserting the following into Google Sheets:
=GOOGLEFINANCE("VGT","return52")
The result will be "#N/A", with the Error explained as "Function GOOGLEFINANCE parameter 2 value is invalid for the symbol specified." So Google Sheets cannot provide the 52-week and 4-week tracking that I require. Google Sheets only supports "returns52" and "returns4" if the symbol is a mutual fund.
-
The earlier mentioned changes are being moved forward, to September 3rd. On that date I'll use a new input feed for buying ETFs based on momentum, using the same momentum criteria as before.
I've verified that Vanguard now charges $0 for many non-Vanguard ETFs. That means the 2.5W fund experiment doesn't need to stick to Vanguard ETFs.
I've double checked etfdb.com's top 52 week performers against mornignstar.com data, and they are a close match. So it looks like I can rely on etfdb.com rankings of ETFs.
So here's my new approach:
1. Every 2.5 weeks, visit the etfdb page for top 52 week returns (ignoring leveraged ETFs):
http://etfdb.com/compare/highest-52-week-returns/no-leveraged/
2. Check momentum of at least the top 5 ETFs. Pick the ETF with the highest 52-week performance minus 4-week performance, and switch the fund's assets to that ETF.
3. If Vanguard charges a fee for that ETF, go down the list until a $0 trade can be found.
As of right now, PSCH has a +10 momentum lead over the next ETF in that list. (I already own PSCH outside this experiment, as well). So if nothing changes in the next ~10 days, the fund would purchase PSCH on September 3rd.
Selecting the highest momentum from thousands of ETFs should give the 2.5W fund experiment greater momentum exposure than selecting from a few dozen Vanguard ETFs. Next update, using this new approach, on September 3rd.
-
The stock market was closed 9/3 for Labor Day, so my 2.5W fund purchases were delayed until today, 9/4.
My procedure is now to take the first 5 entries in the top 100 chart at etfdb.com - specifically the 52-week non-leveraged performance chart. Then, I take 1 year performance and subtract 4 week performance:
(1) PSCH = 64.50 - 7.61 = +56.89 mtm
(2) ULBR = 58.50 - (-2.34) = +60.84 mtm
(3) OILB = 55.69 - 3.59 = +52.10 mtm
(4) BNO = 54.45 - 5.54 = +48.91 mtm
(5) USO = 52.75 - 3.08 = +49.67 mtm
Based on that, ULBR has the highest momentum. But ULBR is NOT a commission free ETF (it's actually an ETN, an exchange traded note), so I can't use it. The whole reason I'm opening up the fund to purchase from this list is Vanguard's expansion of $0/trade for 1800+ ETFs. So Although ULBR has the highest momentum, it doesn't fit the funds "$0/trade" criteria.
PSCH is "Free" to trade, so I switched the fund's holding from Vanguard Information Technology (VGT) into Invesco S&P SmallCap Health Care ETF (PSCH).
Had the fund kept it's original criteria, it would also have switched now that VDE (Vanguard Energy) has higher momentum than VGT.
Speaking of changes in criteria, it might be worth reviewing the fund's progress over the past 2.5 years and what motivated each change.
Originally the experiment used two kinds of momentum, with assets split 50/50 between them. According to the white paper I read, "52-week highest price" momentum added additional value to "52-week performance" momentum. In practice, it was difficult to implement and hurt performance. Too many funds reach their 52-week highest price at the same time, so that approach tended to get diluted when applied to Vanguard sector funds.
The experiment also started out limited to Vanguard's 11 sector funds. Owing to my limited "index fund" thinking, I thought all investment approaches divided up the market and decided the weight to apply to each slice. But there's no reason to limit the experiment to sector funds - any Vanguard ETF could be used, provided it had the highest momentum.
So the fund lost it's "52-week highest price" momentum, and expanded to all Vanguard ETFs. I expected high turnover, and at first that was an accurate prediction. That's also the reason the fund was limited to Vanguard ETFs: they cost $0/trade. A high turnover fund paying commissions will feel an impact on performance.
Finally, last month, Vanguard announced ~1800 ETFs would be $0/trade at Vanguard. I took advantage of this change by expanding the fund experiment's possible ETFs to anything listed on etfdb.com's top 100 list. That list includes Vanguard ETFs - "VGT" is near the middle of the top 100 list. But now the fund can benefit from other ETFs with even higher momentum, which will hopefully create more exposure to the momentum factor. And hopefully, most of the time, more exposure to momentum translates to higher performance.
By a fun (fund?) co-incidence, the 2.5W fund is now 2.5 years old. Which reminds me of another criteria that changed: originally, the "5W fund" made purchases every 5 weeks. This ensured 35 days elapsed between purchases, while still occurring every week on the same day. The 31+ day criteria was needed to avoid wash sales in a taxable account. Since then I moved the fund to a retirement account, where there's no tax implication of selling at a loss in 30 days. So the fund was able to switch to "2.5 weeks" per update.
The fund is now 100% in small cap health care, PSCH, based on the data on etfdb.com's top 100 charts. Next update on Sept 20th.
-
The fund stays invested in PSCH (investco small cap healthcare).
I'm ignoring ETNs primarily because Vanguard charges a transaction fee for them, but not for most non-leveraged ETFs. Here's the top 5, ignoring ETNs:
#2) PSCH, +59 momentum
#4) DBO, +46 momentum
#5) BNO, +45 momentum
The top Vanguard ETF, VGT, has a momentum of about +30. Momentum for the purpose of this experiment is annual performance minus last month performance.
Overall, the fund profit is +40.3% versus the S&P 500 profit of +51.0%. But it's actually a larger gap, since the fund sells within a year, while the benchmark is assumed to be held for over a year before being sold. So that results in a better tax treatment for the S&P 500 benchmark.
Using the median income bracket of 22% and the equivalent long-term capital gains bracket of 15% gives:
40.3% less 22% taxes = 31.4% performance since Feb 2016
51.0% less 15% taxes = 43.3% performance since Feb 2016
Over 2.5 years, the gap widens from 9.7% before taxes to 11.9% after taxes. If you put that in annual terms, taxes cause the fund to lag it's benchmark an additional -0.9%.
I suppose if I wanted to keep people invested in a fund like this, I'd break down the fund's performance into before the Feb 2018 changes, and after those changes:
Feb 2016 - Feb 2018, fund gained +24.2% while benchmark gained +40.1%.
Feb 2018 - Sep 2018, fund gained +16.7% while benchmark gained +10.9%.
Hopefully relying on more ETFs, and using only performance momentum will eventually propel the fund past it's benchmark, but it's not looking likely for this year. But there's always the next update, which will be on Oct 8th.
-
At Vanguard most ETFs are $0/trade, but I believe ETNs are not. So I'm ignoring all ETNs on the chart of 52 week top performance chart at etfdb.
I looked at the top 10-20 ETFs on the chart, almost all of which were oil related. But momentum is calculated by subtracting recent performance, which works in PSCH's favor owing to it's recent drop. So PSCH retains the #1 slot on momentum:
#1) PSCH, 40.96 - (-7.26) = 40.96 + 7.26 = 48.22
#2) DBO, 56.98 - 10.65 = 46.33
Since momentum is calculated as 12 month minus 1 month performance, right now the negative recent performance helps PSCH. Once that -7% ages a month, it will subtract from PSCH's momentum. I think the momentum white paper I read discovered that was a useful change to simple 12 month performance, perhaps because reversion to the mean over 1 month shouldn't cause the fund to switch.
Another warning about etfdb: I've found mistakes in their data at times. For example, if you search for "etfdb vti", that first link leads to to the page for Vanguard Total Stock Market ETF. On etfdb.com, VTI is listed has having 1514 holdings. Both morningstar and Vanguard claim this ETF has over 3500 holdings. You could be fooled into thinking SCHB holds more stocks, when it holds ~1000 fewer stocks.
So with that in mind, here's the two ETFs above using morningstar's data:
#1) PSCH, 40.98 - (-7.80) = 40.98 + 7.80 = 47.78
#2) DBO, 56.98 - 9.67 = 47.31
It's a much closer comparison with those numbers, but still favors keeping PSCH.
Last month the performance gap narrowed to within 10% between the momentum experiment and the S&P 500. Unfortunately most of that -7% or -8% for PSCH showed up recently, and the fund lost -6.7% since the last update. So the performance gap has widened to +16.6%.
From 2/17/2016 until now, the raw performance is as follows:
S&P 500 benchmark: 1.4736x (not quite +50% profit), 1.4026x after tax
2.5W fund experiment: 1.3076 (not quite +33% profit), 1.2399x after tax
Next update on October 25.
-
Looking at ~25 or so of the top performing ETFs for the past 52 weeks, I whittled it down to just:
#1) DBO, 38.63%/yr - (-6.35%/mo) = +44.98
#2) PSCH, 30.13%/yr - (-14.31%/mo) = +44.44
So the fund sells PSCH and buys DBO (Investco DB Oil ETF). Apparently it tracks an oil index, and oil has been doing well. The new allocation is:
100% DBO
Looking at the past 8 months, both the experiment and it's benchmark are down. The S&P 500 is down -0.7% while the 2.5W momentum experiment is down -1.3%. With this much up and down, I'm surprised it's not worse. But given the new investment in an oil ETF, I'm probably speaking too soon...
Overall 2.5W fund profit +23% compared to S&P 500 +39% (since Feb 2016).
Next update on November 12.
-
DBO lost -9% in the 2.5 weeks it was held by the fund. And now PSCH beats DBO again, so the fund realizes that -9% loss and switches back into PSCH.
I will be avoiding commodities and ETNs in all future purchases. Momentum of equity ownership means only buying exchange traded funds with $0/trade that hold equities. ETNs are a promise (as from a bank) to pay based on a formula, and relies on the ability of the issuer to pay. Commodities are very volatile and a separate asset class than equities. So future purchases will not be oil funds, or gold, or any non-equity ETF.
But the -9% damage is done. The fund is now -30.4% behind it's benchmark overall (since Feb 2016), and if this was a real fund, I think it would be closing soon. Imagine losing over -10% per year against the S&P 500...
Performance momentum is 52 week performance minus 4 week performance:
#1) PSCH : 38.45 - (-3.73) = +42.18 mtm
#2) QAT : 36.91 - 5.48 = +31.43 mtm
#3) IHF : 35.67 - 0.55 = +35.67 mtm
#4) RTH : 32.29 - (-0.17) = +32.46 mtm
...
DBO : 15.55 - (-18.75) = +34.3 mtm
This is a dispiriting update. I feel like the rules used by the fund have to be stated before a purchase, so I couldn't avoid buying a commodity ETF when it topped the performance charts. I suppose it's useful to define the assets this fund can purchase: ETFs holding 95%+ stocks (they might hold cash).
But seeing "Qatar ETF" on the top chart, I think that should be updated to "ETFS holding 95%+ U.S. stocks". That excludes international allocations, but still allows small cap ETFs. So the S&P 500 isn't a perfect benchmark for that, but it's reasonable. (In theory a fund can tilt to small caps and beat the S&P 500 by taking on more risk than the S&P 500)
Time to try and forget the fund's recent performance, and see what happens between now and Nov 29th.
-
This time I'm ignoring non-US non-ETFs, which eliminates the need to look at the Qatar ETF or the ETN based on the LIBOR. Here's the top 3 from a much longer list:
#1) PSCH 31.70 - 5.47 = +26.23 mtm
#2) IHF 33.13 - 7.40 = +25.73 mtm
#3) XHS 27.39 - 6.73 = +20.66 mtm
I used etfdb.com for raw data, and then morningstar's performance data for the final 3. Relying on only etfdb data would have put IHF over PSCH, but I trust morningstar's data more than etfdb's. (For example, etfdb claims VTI holds 1515 stocks, when other sources - including Vanguard - all agree it's more than 3500+ stocks).
So the fund stays 100% invested in PSCH, and remains staggeringly far behind it's benchmark: -27% over ~2.8 years. So far, the best advice is "invest in the S&P 500", not this experiment.
Next update on Dec 15 2018.
-
Next update will be on Dec 17, not Dec 15.
I'd like to reduce the time I spend checking on this experiment, so I've decided to revert back to updates every 5 weeks starting on Dec 17, 2018. It's harder to keep motivated over an experiment that clearly lags the market. So here's the rough rules of the experiment:
1. use the etfdb.com charts to find ETFs with the highest non-leveraged performance in the past 52 weeks
2. calculate momentum using performance data from a more trusted website (morningstar.com):
52 week performance minus 4 week performance equals "one year momentum" (11 months, actually)
3. the fund switches 100% of assets into the U.S. equity ETF with the highest one year momentum
4. repeat every 5 weeks
Next update on Monday, December 17.
-
Make that Dec 18th, a day late.
I took the top 6 entries on the top 100 list from etfdb, and skipped over a lots of ETFs that weren't 95%+ U.S. Equities. Several ETNs (exchange traded note - a promise from a bank, no actual equities), various country ETFs, gas ETFs, etc. Then I took the top 3 from that list, and compared performance using morningstar data:
#1) PSCH (again), 13.57 - (-10.72) = +24.29 mtm
#2) IHI (another health care ETF), 13.93 - (-8.70) = +22.63 mtm
#3) XWEB, 14.89 - (-5.59) = +20.48 mtm
So the fund remains invested 100% in PSCH.
Thanks to the recent market declines (including the -10.72% in 1 month shown above), the 5W fund is now back where it started. It has a total profit of +0.55%, round that up to +0.6% for 2.8 yrs of investing. It's benchmark S&P 500 is up +31.88% during the same time period. Essentially, 10%/year versus 0%/year.
Next update on Jan 21, 2019. Wishing your assets a happy new year...
-
The U.S. equities markets won't open until Jan 22nd, so I'll place trades for this experiment at that time. Until then, the relative positions of ETFs can't really change, so I can make my calculation in advance. This time I'm going to show how I go about this in detail, so you see an example of the "algorithm" in action.
I visit the URL https://etfdb.com/compare/highest-52-week-returns/no-leveraged/ to see the current top 100 ETFs for the past 52 weeks. I select only ETFs (not ETNs), without leverage, with 95%+ equities (not natural gas). Here's the first ~30 from that list:
VIIX VelocityShares Daily Long VIX Short-Term ETN 37.61% not an ETF
VXX iPath S&P 500 VIX Short-Term Futures ETN 37.54% not an ETF
VIXY ProShares VIX Short-Term Futures ETF 36.30% not an ETF
VXXB iPath Series B S&P 500 VIX Short-Term Futures ETN 35.95% not an ETF
PALL Aberdeen Standard Physical Palladium Shares ETF 24.24% not U.S. equities
PSJ Invesco Dynamic Software ETF 21.81% roughly +4% momentum
XWEB SPDR S&P Internet ETF 21.72% roughly +3% momentum
QAT iShares MSCI Qatar ETF 21.21% not U.S. equities
KSA iShares MSCI Saudi Arabia ETF 19.89% not U.S. equities
VXZ iPath S&P 500 VIX Mid-Term Futures ETN 18.98% not an ETF
OLD Long-Term Care ETF 18.88% roughly +13% momentum
VXZB iPath Series B S&P 500® VIX Mid-Term Futures ETN 18.70% not an ETF
VIXM ProShares VIX Mid-Term Futures ETF 18.08% not U.S. equities
REZ iShares Residential Real Estate ETF 17.70% roughly +14% momentum
UNL United States 12 Month Natural Gas Fund 16.85% not U.S. equities
PUI Invesco DWA Utilities Momentum ETF 16.34% roughly +16% momentum
IHI iShares U.S. Medical Devices ETF 14.31% roughly +2% momentum
BTAL AGFiQ US Market Neutral Anti-Beta Fund 14.11% not U.S. equities
IGV iShares North American Tech-Software ETF 13.86% roughly -2% momentum
UNG United States Natural Gas Fund 12.78% not U.S. equities
XSW SPDR S&P Software & Services ETF 12.75% roughly -4% momentum
UTES Reaves Utilities ETF 12.46% roughly +13% momentum
FDN First Trust Dow Jones Internet Index 12.36% roughly -7% momentum
XITK SPDR FactSet Innovative Technology ETF 12.09% roughly -6% momentum
ARKG ARK Genomic Revolution Multi-Sector ETF 12.07% roughly -18% momentum
PSCH Invesco S&P SmallCap Health Care ETF 11.51% (current allocation) roughly -4% momentum
CLIX ProShares Long Online/Short Stores ETF 11.19% leveraged + not 95% U.S. equities
JHMU John Hancock Multi-Factor Utilities ETF 11.16% roughly +12% momentum
GULF WisdomTree Middle East Dividend Fund 11.15% not U.S. equities
SKYY First Trust ISE Cloud Computing Index Fund 11.13% roughly -3% momentum
XHE SPDR S&P Health Care Equipment ETF 11.09% roughly -3% momentum
BFIT Global X Health & Wellness Thematic ETF 11.01% roughly +1% momentum
Ignoring all the ETNs, non-U.S. funds, non-equity funds... and then ignoring momentum below +10%, I get the following list of 5 ETFs:
PUI Invesco DWA Utilities Momentum ETF 16.34% roughly +16% momentum
REZ iShares Residential Real Estate ETF 17.70% roughly +14% momentum
OLD Long-Term Care ETF 18.88% roughly +13% momentum
UTES Reaves Utilities ETF 12.46% roughly +13% momentum
JHMU John Hancock Multi-Factor Utilities ETF 11.16% roughly +12% momentum
In the past I've spotted problems with etfdb data (# holdings, % cash, etc) so I check morningstar's data for each of these ETFs to get my accurate momentum data. I will use the "performance" tab, and take 1 year minus 1 month performance:
PUI = 16.37 - (-.63) = +17 mtm
REZ = 17.76 - 2.00 = +15.76 mtm
OLD = 18.89 - 2.70 = +16.19 mtm
UTES = 12.54 - (-.57) = 13.11 mtm
JHMU = 12.21 - (-1.60) = 13.81 mtm
I could almost say "I'm getting OLD", but actually PUI has slightly higher momentum:
PUI is Invesco DWA Utilities Momentum ETF. When the markets open Jan 22 (Tues), I'll shift the 5W fund experiment from 100% PSCH into 100% PUI - from healthcare momentum to utilities momentum.
-
On Jan 22 I switched from 100% PSCH to 100% PUI (Investo DWA Utilities Momentum) per my prior update. I forgot to mention the next update, which will be on Feb 22, 2019.
This experiment is turning 3 years old next month. I've been using "Vanguard S&P 500 ETF" (VOO) as the benchmark. According to my records, when I started this experiment (on Feb 17, 2016) VOO was trading at $178.22 / share. It now trades at $242.13, for a share price growth of +35.86% (which fortunately matches my spreadsheet!). In that same time, the momentum experiment has gained +4.48% overall. The last quarter of 2018 did an immense amount of damage to performance, losing -28% in those 3 months, wiping out most of it's prior profit. It's actually impressive how fast it fell, and perhaps a warning.
On an amusing note, last year I downsized the fund a bit. The percentages are all the same, but I started a separate spreadsheet and copied a summary of prior performance into a new spreadsheet. So the fund was much larger when it gained +24%, and much smaller when it lost -28%. I couldn't have predicted that, but got lucky on the timing anyways. The fund remains a very tiny allocation relative to my portfolio.
To break those losses down into a bit more detail:
from 2018-09-03 to 2018-10-25 the fund held PSCH (health care) for a -15% loss
from 2018-10-25 to 2018-11-12 the fund held DBO (oil) for a -8% loss
from 2018-11-12 to 2019-01-22 the fund returned to PSCH (health care) to grab another -7% loss
(the fund gained +15% while holding VGT for most of 2018)
Overall, especially owing to the 4th quarter of 2018, this momentum experiment still badly lags the S&P 500.
Next update on Feb 22.
-
I said the next update is Feb 22, but my spreadsheet indicates I should make the purchase on Feb 25. I think the best compromise is to commit to what I purchase today, Feb 22, but then make the actual purchase on Monday, Feb 25.
I used etfdb to narrow the list of top performing ETFs, skipping any non-U.S. non-stock ETFs. That left me with 3 entries, with the following 1 year momentum (the latest month is subtracted to ignore reversion to the mean effects):
REZ) iShares Residential Real Estate Capped ETF, 30.61 - 4.63 = +26.0 mtm
OLD) The Long-Term Care ETF, 27.67 - 4.29 = + 23.4 mtm
PUI) Invesco DWA Utilities Momentum ETF, 25.82 - 5.64 = +20.2 mtm
So on Monday, Feb 25th, the experiment will switch from PUI (utilities) to REZ (real estate).
-
Done. Sold PUI when it was up +1% for the day, and REZ hadn't made any gains/losses. The 5W fund experiment is now 100% invested in REZ (iShares Residential Real Estate ETF). Something tells me switching investments every 5 weeks won't play out that well.
I wonder if there's a combination of checking often, but then not switching investments that often. For example, after a purchase the fund could always wait 2-3 months. And after the waiting period, check monthly. At any rate, the experiment has no such provision right now.
Happy Birthday 5W fund experiment - now 3 years old. Here's the performance during the entire experiment:
5W experiment: +10.3% total, or 3.3% / year for 3 years.
S&P 500 index: +44.9% total, or 13.1% / year for 3 years.
(sanity check: morningstar reports 15.1% / year for 3 years,
and etfdb shows +54.6% return over 3 years)
Note this experiment doesn't track distributions / dividends. Those would be annoying to track, and the dividends would only matter to the extent the S&P 500 had more or less than the fund experiment's holdings. So that probably explains most of the gap between other data sources and my S&P 500 performance. I use the GOOGLEFINANCE() function of Google Spreadsheets to track prices.
Back to the data, the experiment always sells during a year while a passive investor is assumed to hold the S&P 500 for years. So a short-term tax rate of 22% is used for the experiment and a 15% tax rate is used for the benchmark S&P 500:
5W experiment) 10.3% total x 0.78 = +8.0% after tax
S&P 500 index) 44.9% total x 0.85 = +38.2% after tax
So the 5W experiment trails the S&P 500 by about 10%/year after taxes (and about 10%/year before taxes, too).
Next update is April 1, and hopefully the markets have a good joke in store for the experiment...
-
I'm still reading along and following your progress. The old momentum threads were one of the forum's first serious attempts to promote market timing, and we spilled thousands of words arguing about it. Some die hard forum members remained convinced until their dying (banned) days that there was a persistent and exploitable signal in the historical stock market records that would tell them when to buy and when to sell assets of various types.
And the only way to really test that theory is to run the experiment, forward in time. Data mining the historical records will always find something that looks promising, but that doesn't mean the same something will work in the future. I applaud your dedication to accountability in this thread, and I hope other people will also find it instructive.
Trying to time the market is hard work, and hardly ever seems to work out for a profit.
-
It's been fun watching this unfold.
I bought MTUM (iShares momentum "smart etf") right around the time you started this thread. 2017 was a rip your face off good year (37.5% vs SPY = 21.83%). So far YTD I'm lagging 1.5% from SPY.
I did just notice the turnover is 104%. :|
Anyway enough about that fund. Again, cool to see you update this strategy.
-
According to morningstar data, "iShares Edge MSCI USA Momentum Factor ETF" (MTUM) as of 2/28/2019 had beaten the S&P 500 for the past 3 years: 18%/year vs 15%/year. Meanwhile, this experiment can't pick individual stocks - just entire ETFs with mixed momentum. MTUM's calculations impress me, since they incorporate standard deviation into their filters (which would be very painful for me to re-create for my experiment, so I didn't).
I tried creating a second experiment on this forum where I picked individual micro-cap stocks (under 50M in assets). I really, really should have considered standard deviation! It was too painful watching 5-10% daily swings, and I gave up the experiment. Although I kept invested through the 2008 crisis, I found my "stomach acid" limit for investing at that time.
I think it's unfortunate that I had to modify the rules of the experiment a few times. Early on, I wanted to limit turnover and bumbled for a way to do that. Later I realized I could pick any ETF not just sector ETFs, and then wound up picking from all ETFs since Vanguard began charging $0/trade for that. Where in theory performance should have been even better, it didn't turn out that way. I wish I hadn't bought the oil ETF, because it's such a bad fit for the experiment - but I also felt the rules can't be changed on the fly. So I had to limit to U.S. equity ETFs later.
I think it could be interesting to see how a "Buffet small cap" experiment would go. That would essentially test the Fama-French five factor model: small, value, quality, profitability (the 5th factor is beta, or the market itself). And then track that against the S&P 500.
-
Just caught up on a bunch of posts in this thread.
I appreciate your willingness to continue updating us.
Most people advocating for something unusual stop updating once they aren't outperforming. Having years of data is interesting, even if it's not showing what you've hoped. So thank you.
I hope it turns around for you, even with just your play money.
-
A year ago I moved the experiment from taxable to an IRA, and didn't mention that I reduced the amount of assets involved. So I actually got luckier than the numbers show, but that's not part of the experiment. I turned the prior performance of the experiment and it's benchmark into multipliers so that I could incorporate the prior 2 years, and still use a smaller allocation:
Feb 2016 - Feb 2018: +24.2% (larger allocation in taxable)
Feb 2018 - Feb 2019: -11.6% (smaller allocation in an IRA)
Another advantage of running an experiment like this: I can take my interest in experimenting and focus it here, instead of changing my core investments. Much like a surprising number of money managers, this experiment is active, but privately I'm indexing.
Next update on April 1, 2019.
-
@MustacheAndaHalf Where are you downloading your momentum numbers?
-
So here's my new approach:
1. Every 2.5 weeks, visit the etfdb page for top 52 week returns (ignoring leveraged ETFs):
http://etfdb.com/compare/highest-52-week-returns/no-leveraged/
2. Check momentum of at least the top 5 ETFs. Pick the ETF with the highest 52-week performance minus 4-week performance, and switch the fund's assets to that ETF.
3. If Vanguard charges a fee for that ETF, go down the list until a $0 trade can be found.
I switched to using etfdb in September 2018, and hand calculate momentum from their data. I look up the "performance" tab and use 52-week minus 4-week performance to give me the last year less the most recent month. Recently, I look at 10-25 entries by hand, and filter out those nowhere near the top few. I then look up those top few on morningstar, which I think has more accurate data than etfdb (for example, incorect # of stocks on etfdb). Both of those are free to use, which is another criteria for keeping the experiment's costs low.
-
I'll calculate the highest momentum today, then purchase it tomorrow.
I visited "etfdb.com" which has 52-week performance charts. I take the top ~10 entries, and look at etfdb.com performance (1 year minus 4 week). Anything too far from the leaders gets dropped, leaving me with 3-4 ETFs. This week I skipped an ETF named "MJ" because it's mostly foreign stocks. Here's the "1 year minus 1 month" calculations of momentum (made by hand):
PSJ ("Invesco Dynamic Software ETF") : 31.70 - 3.78 = +27.92 mtm
ARKG ("ARK Genomic Revolution Multi-Sector ETF") : 30.27 - 6.82 = 23.45 mtm
IHI ("iShares U.S. Medical Devices ETF") : 25.46 - 2.36 (etfdb listed this as -1%) = 23.1 mtm
Those numbers were much closer using etfdb data, which is why I double check here. The winner is PSJ ("dynamic software ETF"). On Monday I'll switch the 5W fund / experiment to 100% in PSJ.
When this experiment first started, the time it did well corresponded to a time when it kept the same holding (Vanguard Utilities). Which sometimes make me wonder if turnover hurts the fund's performance (much as turnover hurts personal performance in studies).
-
When this experiment first started, the time it did well corresponded to a time when it kept the same holding (Vanguard Utilities). Which sometimes make me wonder if turnover hurts the fund's performance (much as turnover hurts personal performance in studies).
So you're saying that buy and hold might be the key to not having momentum underperform? (https://i.imgur.com/xvMSoxY.gif)
-
While I should be a cheerleader for that viewpoint, my data really only applies to "picking an ETF with highest 12 month performance momentum", and before that "sector momentum". There's evidence that others might be doing this better - the largest momentum ETF ("MTUM" has $8.5 billion in assets) leads the S&P 500 by +3% in the past 3 years:
iShares Edge MSCI USA Momentum Factor ETF ("MTUM"), 3 year return: +16.86% / year
Vanguard S&P 500 ETF ("VOO"), 3 year return: +13.46% / year
5W fund experiment ("5W"? no real ticker), return since started (3.1 years): +4.32% / year
Maybe it's useful to delve into differences between my experiment and MTUM. After reading about momentum in academic papers and books, I looked at the description of how MTUM calculates momentum. They calculate on another level from me - they track performance divided by volatility. So a stock that has stable high performance gets favored. They calculate momentum for individual stocks over two time periods (12 mo and ... 6 mo?). And since they use an outside index, I get more visibility into what they're doing (sadly, Vanguard's momentum ETF provides no visibility - an internal index with no specifics that I've found).
The disadvantages of the 5W experiment are numerous. I had to buy entire ETFs at a time, with hundreds of stocks inside. The mixture of stocks has some momentum, but it's diffused by having higher and lower momentum stocks in the same fund. But it's $0/trade, unlike buying a number of stocks. Further, my data is 100% from public data on the web - I use etfdb data now, and used to use Google Sheet data on Vanguard's ETFs. Fairly crude, but I can only sustain the experiment by ensuring it's automated - otherwise the time drain is too much. I don't factor volatility in at all, so a "lucky" high volatility ETF can get picked over an ETF with a more significant out-performance relative to it's volatility.
My advice to anyone who can't help trying to beat the market as an individual investor:
1) benchmark yourself against the S&P 500. If you keep losing to the S&P 500, it might be a better investment.
2) read up on the arguments for investing in the S&P 500 (or total U.S. market)... Warren Buffet, A Random Walk Down Wall Street ... any book that uses a lot of stock market history, essentially.
3) make it worth your time - automate your investment approach. If you're spending hours each week, that's hundreds of hours per year. Maybe you need to invest less frequently, and automate the selection of stocks. Otherwise even if you beat the S&P 500 by a tiny amount, you're still out all that time spent.
-
The fund is 100% in PSJ (go software, go!). Next update on May 5th.
-
I suspect I need to go back to basics. The fund changes investments at nearly every update. Either my timing is perfect (ha!), or the fund is switching investments too often. Performance so far suggests the latter. I'll stick to the same approach for this update, but I think "sector momentum" might have been a better idea all along.
So for this update, I looked at the top 12 performing ETFs for the past 12 months, and examined every one of them on morningstar. Here are the top 5 performers, ranked by momentum:
1) ARKG, 33.45 - (-2.34) = 35.79 mtm (oops, 88% U.S.)
2) PSJ, 34.81 - 3.77 = 31.04 mtm (99% U.S.)
3) MJ, 27.27 - (-1.47) = 28.74
4) PXMG, 29.15 - 1.88 = 27.27 mtm
5) XSW, 26.22 - 2.23 = 23.99 mtm
Since ARKG isn't 95% in U.S. stocks, I'm skipping over it and leaving the fund invested 100% in PSJ. I setup the "95% U.S. equities" rule awhile back to avoid two things: ETFs that tracked commodities, and ETFs that weren't a good fit for an S&P 500 benchmark. So the fund stays in PSJ.
-
I'm bringing the 5W momentum experiment to an end after 3 years and 4 months. I'm not sure what change is needed, and the fund has already had two different approaches (sector momentum, and then any ETF with momentum).
Since the start of this experiment (Feb 17, 2016) the fund has gained just +17%. That's 5% annualized, which fell well below it's benchmark of the S&P 500. The S&P 500 gained +51% over 3.3 years, or about 15%/year. If there's one takeaway, it's that the S&P 500 beats sector momentum. Maybe it's also worth noting the 5W experiment beat the S&P 500 early on - a warning that results measured in months might not predict performance over several years.
This experiment doesn't represent all styles of momentum investing. If anything, it's unique in purchasing entire ETFs at a time, rather than individual stocks. While the fund did poorly, professionally run momentum funds have done fine - even beaten the S&P 500. About a year after I started the experiment, Vanguard even created a new momentum fund of it's own.
So the 5W momentum experiment is now closed / over, with a +17% overall performance over 3.3 years compared to the S&P 500 performance of +51%.
-
Thank you for running the experiment and taking the time to share it with the rest of us!
I have a suspicion that if there is an edge in momentum trading it is already being exploited by algorithms and there may not be room for individual investors to do much.
-
I really appreciate your continual updating and final results on this, despite the outcome. Thank you.
-
Glad it was appreciated. Stick to the S&P 500!
I doubt anyone is trying this exact approach - it's too scattered and unprofessional. It started by only making trades which cost $0 at Vanguard. There's no way professional traders or portfolio managers face that restriction. I also focused on sector momentum, where I bought a Vanguard sector ETF that holds every stock in a specific sector (REIT, utilities, tech, etc). A professional fund wouldn't need to buy every stock - they could pick the best sector, and buy the highest momentum stocks individually. Given the restrictions I imposed to keep my costs low, I doubt my approach is competing with professional funds. Plus, if they followed my approach, they'd probably have closed already with such bad performance relative to both the S&P 500 and other momentum funds.
-
Yes, thank you, I enjoyed following along and learning from your experiences.