Author Topic: Small Value Momentum Expeiment  (Read 4572 times)

MustacheAndaHalf

  • Walrus Stache
  • *******
  • Posts: 6633
Small Value Momentum Expeiment
« on: February 15, 2018, 09:16:58 PM »
I've decided on some specific rules and parameters for a new experiment that includes small cap, value and momentum factors.  Since the fund's assets are $4200, I find it appropriate to have "42 is the answer" as a theme for the fund.  Call it "the 42 small/value momentum fund", I suppose.

To avoid penny stocks and those that don't trade very often, two criteria will be used:
stock price over $4.20
volume over 4200 trades/day

The stock screening tool I use has a category "under $250M market cap", so I will use that to keep the experiment focused on small cap stocks.  These stocks are especially small, so "micro caps" is probably a better description.

Mutual funds cannot reveal this level of detail.  Other stock market participants who know what a large fund is about to do will "front run", or buy the stock first.  Then they sell those stocks at a profit to the fund that has revealed too much detail.  Early on, Vanguard's S&P 500 fund experienced this, which is probably why it actually has about 506 stocks.  Fortunately, a fund that only buys $1000 worth of stock that trades thousands of times per day isn't going to have much market impact.

The formula for the fund will mix value and momentum.  Value will be represented by price/book, with a lower number representing a "value" stock.  Momentum will be represented by 12 month performance.  Where possible, I will manually check (12 month minus last month) to decide if a few stocks are close in performance.

After screening for small size, I will screen for momentum until only 42 stocks remain.  The screen takes the performance of the S&P 500 over the past 52 weeks, and screens out stocks without a certain level of performance.  I might see too many stocks beating the S&P by +20%, and too few beating the S&P 500 by +50%, and use a setting between those numbers to get the 42 stocks with the most momentum.

From this list of 42 high performing stocks, I will pick the stock with the highest performance and the lowest price/book.  I will take the stock with the most momentum, and of the top 42 stocks, the stock with the most value.

Then I form a list of 42 value stocks using price/book as the criteria.  Maybe too many stocks have price/book under 2.0, and too few stocks have price/book under 0.5, so I will have a setting between those numbers.  From the list of 42 value stocks, I will pick the stock with the most performance and the stock with the lowest price/book.

Overall I will have 4 stocks.  One stock with momentum, one stock tracking value, and two stocks tracking a mixture of value and momentum.  Note 4 stocks is not a portfolio: it's very concentrated.  Although I avoid penny stocks ($4.20 minimum stock price), deep value micro cap stocks are very risky and could go under, costing the fund -25% of it's assets each time it happens.  I will avoid any stock that trades "on the pink sheets" or only on the OTC market.

Since this experiment tracks small cap, value and momentum I think a "smart beta" or "multi-factor" fund will need to be it's benchmark.  I will need to research that and come up with a benchmark by Monday.  I might also track the S&P 500, even though it's a totally inappropriate benchmark for a micro-cap value/momentum fund.  The level of risk is not comparable.

The fund will start with $4200 in assets, and each trade costs $2.  Every 21 trades costs 1% of the fund's assets, which is why I prefer 42 days between possible trades (6 weeks) rather than monthly or more often.  In addition, value stocks might not change as often, so that could also lower the expenses of the fund.  I plan on displaying exact amounts used in buying/selling as part of this experiment.

I plan on starting this experiment Monday, February 19.

MustacheAndaHalf

  • Walrus Stache
  • *******
  • Posts: 6633
Re: Small Value Momentum Expeiment
« Reply #1 on: February 16, 2018, 09:42:55 AM »
This is not my core portfolio, but an experiment to see what happens.  I should have stated I plan to list all trades in this thread, so people can have a record of what happened.  Note when someone claims to have always used momentum.. there's no record of it.  You might be right, or might have made mistakes.  With a record of transactions, others can replicate or analyze what happened.

I'd like to learn from it, and make actual transactions to see what happens.  I disagree that all small cap stocks are a trap.  Companies worth billions of dollars are considered small cap stocks - fairly difficult to manipulate that (and also illegal). Penny stocks are more known for that, and I'm avoiding those.

I'm not interested in putting a lot of time into this experiment, so I won't be doing research on these stocks.  I'm limiting it to $4200 rather than trying to make it an important part of my portfolio.  The goal is to try an approach and see what happens.

MustacheAndaHalf

  • Walrus Stache
  • *******
  • Posts: 6633
Re: Small Value Momentum Expeiment
« Reply #2 on: February 16, 2018, 09:53:53 AM »
I've decided on a benchmark.  This experiment essentially tries to tilt towards every factor in the "4 factor model" of the market:  market beta, small, value, and momentum.  (There's a 5 factor model that uses quality and profitability instead of momentum).  The surprising thing is that although all factor models of the market include small and value.... the multi factor funds don't.

I plugged over a dozen "multi factor" funds into morningstar.  I looked at the percentage of assets in small cap, and compared (value minus growth) boxes.  All the funds I looked at did terribly - many of them have 0% small cap.  Not only is that missing a key factor, but it makes it less useful as a benchmark for this experiment.

So I finally found "PNC Multi Factor Small Cap Value" (PMUIX).  According to morningstar, it's 94% small cap (split 50/50 between small and micro).  And when I subtract the growth tilt from the value tilt, there's about 26% more value than growth stocks (ignoring blend).  Doesn't seem like I have much choice on my benchmark: PMUIX will be the benchmark, as the only small value momentum fund I could find.

Note in theory value and momentum are opposites.  Value is the ignored stock, and momentum is the stock receiving everyone's attention.  So finding a stock with both value and momentum might not be a good idea.  In the literature I've read, value and momentum are negatively correlated (according to decades of prior data).

First purchase will be Monday, Feb 21.  I'll use the fund's $4200 to buy 4 stocks as described in the first post.

hodedofome

  • Handlebar Stache
  • *****
  • Posts: 1463
  • Age: 44
  • Location: Texas
Re: Small Value Momentum Expeiment
« Reply #3 on: February 16, 2018, 12:30:08 PM »
With a rules-based system like this you should be able to backtest it using www.portfolio123.com and see how it's performed for the last 20 years.

MustacheAndaHalf

  • Walrus Stache
  • *******
  • Posts: 6633
Re: Small Value Momentum Expeiment
« Reply #4 on: February 16, 2018, 09:49:16 PM »
hodedofome - That website costs $1000/year for using the screener with backtest.

Note screeners only let you sort with one criteria, and typically don't have "limit to 42" as an option.  So the screen I use needs to be manually modified and run a couple different ways.  Although I'm using rules, the screeners do not support the rules I'm using.

For example, the momentum screen.  I sort all micro-caps by momentum, screening out those with volume under 4200/day and under a price of $4.20.  But then I manually modify the screen until it holds only the top 42 stocks.  After that, I manually re-sort the list by value.  The use of two levels of sorting is probably not supported by most screeners, which is why I'm doing it by hand.


L.A.S. - Thanks for clarifying about the small cap companies and squandering assets.

This experiment emphasizes momentum, small cap, and value.  There's a number of funds that use 2 of the 3 factors I'm interested in:
VBR lacks momentum, is 67% small, and holds 33% more value than growth.
DWAS uses momentum, is 99% small, but it's a small/growth fund (-45% value tilt).
USVM uses momentum, is 85% small cap, but value tilts only 3%.

VBR both tracks larger funds (it's 37% mid caps), and does not use momentum, which is why I didn't pick it as a benchmark for this experiment.  But I can post it's price Monday along with the benchmark I picked, and maybe a couple others (VOO and USVM).  Later any of those can be compared against the experiment.

Nate79

  • Stubble
  • **
  • Posts: 146
Re: Small Value Momentum Expeiment
« Reply #5 on: February 16, 2018, 09:53:26 PM »
Only 4 stocks? Don't think you can draw any conclusions based on such small number. Why not something like 40? Can't you do free trades at something like Robinhood?

Sent from my SAMSUNG-SM-G930A using Tapatalk


MustacheAndaHalf

  • Walrus Stache
  • *******
  • Posts: 6633
Re: Small Value Momentum Expeiment
« Reply #6 on: February 16, 2018, 10:03:54 PM »
Actually the time the experiment is run is far more significant than the number of stocks.  Ages ago, the Motley Fool proposed "dogs of the dow" which was also only 4 stocks, and they did well until their theory collapsed.  The problem wasn't the number of stocks, it was the length of time it had been tested.  And unfortunately, that can take decades - and I don't plan on running the experiment that long.

If you read books related to modern portfolio theory, there are numerous studies showing that value, small, and momentum are important factors in explaining the market's performance.  All 3 show up in U.S. and foreign markets over various time frames, suggesting they are persistent.  But over time, they may be less significant as they get easier and cheaper to implement (or arbitrage against).  You can look at Larry Swedroe's "Your Complete Guide to Factor‑Based Investing" if you want some of the historical data on these factors, and why they've been established as useful before I began this experiment.

MustacheAndaHalf

  • Walrus Stache
  • *******
  • Posts: 6633
Re: Small Value Momentum Expeiment
« Reply #7 on: February 20, 2018, 09:19:57 AM »
==== First Purchase ====

Markets were closed 2/20, so I made the first purchase today.  This experiment starts with $4200 cash and pays $2/trade.  It looks for stocks with market caps under $250M, that have 4200 trades/day ("volume") or more, and that have a stock price of $4.20 or more (no penny stocks).

The following 4 stocks were purchased (sh = shares)
39 sh XOMA for $1038.18 ($26.62/sh)
192 sh VOXX for $1043.38 ($5.43/sh)
235 sh MFIN for $1057.45 ($4.50/sh)
48 sh AOI for $1050.96 ($21.90/sh)

For pure performance, several stocks had to be rejected since they only trade on "pink sheets", or off the market.  That includes ZYXI and VFFIF.  My stock screen showed the next 3 stocks as CERC, XOMA, and COOL.  Much of CERC's performance is from the prior month, which made XOMA the first choice.  Here are 12 month minus 1 month performance:
CERC = 405% - 55% = +350%
XOMA = 403% - -18% = +421% (best momentum of these 3)
COOL = 318% - -10% = +328%

With the top ~42 stocks sorted by performance, I re-sorted by price/book (p/b):
MFIN 0.40 p/b (lowest price/book, so value pick)
SORL 0.72 p/b
AOI 0.79 p/b

Here's the pure value picks, with momentum included:
VOXX 0.30 p/b, +40% momentum (tied first on value, highest momentum)
ANW 0.30 p/b, -51% momentum
GASS 0.31 p/b, +6% momentum
CGI 0.33 p/b, -22% momentum

And finally the top ~42 value picks, re-sorted by performance over the past 52 weeks:
AOI +61%, 0.79 p/b (highest performing value stock, selected)
ASFI +34%, 0.56 p/b
MFIN +31%, 0.40 p/b
GSVC +22%, 0.70 p/b

After spending $8 to purchase 4 stocks, the fund has $2.03 in cash (99.95% invested).

The experiments primary benchmark is PNC Multiple Factor Small Cap Value Fund Class I ("PMUIX").  As of the last day markets were open (Friday Feb 17), PMUIX price/share was $23.18.

Here's some ETFs that might be interesting for alternate benchmarks.  I took the bid and ask prices, averaged them, and rounded down the fraction of a penny.
USVM $50.10
VBR $131.16
VOO $249.97

VTI $139.63
MTUM $109.58
IMTM $31.09
VXUS $57.48

VGT $175.01
AAPL $173.62
GOOG $1103.56
GLD $126.87

Bitcoin $11455/BTC
Ethereum $936.65/ETH

The next update will be in ~42 days (6 weeks) on April 3rd.  Since the theme is 42, it's too bad the experiment can't trade on 4/2, but it's a Sunday.  So next trade date 4/3.

toganet

  • Stubble
  • **
  • Posts: 239
  • Location: Buffalo, NY
Re: Small Value Momentum Expeiment
« Reply #8 on: February 21, 2018, 09:01:16 AM »
I find this interesting and I always learn a lot from these experiments.

salmo trutta

  • 5 O'Clock Shadow
  • *
  • Posts: 11
Re: Small Value Momentum Expeiment
« Reply #9 on: March 02, 2018, 01:34:02 PM »
Some of this reading may interest you as well:
https://alphaarchitect.com/alpha-architect-white-papers/


yachi

  • Handlebar Stache
  • *****
  • Posts: 1141
Re: Small Value Momentum Expeiment
« Reply #10 on: March 20, 2018, 11:41:30 AM »


I plan on starting this experiment Monday, February 19.

You should start April 20th!  By chance are you excluding canabis stocks? because that would be ironic.

Telecaster

  • Magnum Stache
  • ******
  • Posts: 3551
  • Location: Seattle, WA
Re: Small Value Momentum Expeiment
« Reply #11 on: March 20, 2018, 12:19:58 PM »
Actually the time the experiment is run is far more significant than the number of stocks.  Ages ago, the Motley Fool proposed "dogs of the dow" which was also only 4 stocks, and they did well until their theory collapsed.  The problem wasn't the number of stocks, it was the length of time it had been tested.  And unfortunately, that can take decades - and I don't plan on running the experiment that long.

The Foolish Four (which was a variant of the Dogs of the Dow) was a classic case of curve fitting.  They came up with a theory and then backtested it back 25 five years or something using yearly starts--which was about as good as an individual investor could do at the time.   I don't recall the exact details, but the whole premise didn't make much sense.  Apparently they just started trying things and found a combination that showed out performance.  Actually, let me just Google it:

"You start with the 30 stocks that make up the Dow Jones Industrial Average . This acts as a prescreen to ensure that you are selecting from successful, large-cap U.S. stocks. Then you screen those 30 stocks for the 10 with the highest dividend yield. (Yield equals dividend per share divided by price per share.) You then buy the five lowest-priced stocks of the high-yield 10 and hold them for one year. At that point, you recalculate the list, then make any necessary changes. And so forth. The original Foolish Four (a.k.a. Foolish 4.0) modified the Beating the Dow list by dropping the lowest-priced stock and doubling the investment in the second-lowest-priced stock."

Really?  You drop the number one and double down on the number two?  How does that make a lick of sense?  There was no logic behind any of it that I could see.   Anyway, they got access to more data and did a longer backtest and found there was no out performance.  Sadly, it appears the TMF website has been scrubbed of most of the Foolish Four website. 

It might be cheaper for you to backtest your ideas using portfolio123.com or something similar as  hodedofome suggested. 

Also, you may wish to consider more stocks.  You only have one stock representing each of your factors.   We all know what can happen to any individual stock. 


MustacheAndaHalf

  • Walrus Stache
  • *******
  • Posts: 6633
Re: Small Value Momentum Expeiment
« Reply #12 on: March 20, 2018, 06:36:51 PM »
It might be cheaper for you to backtest your ideas using portfolio123.com or something similar as  hodedofome suggested.
I already replied right after hodedofome's suggestion, that website costs too much per year for access to it's stock screen back test. 

You're assuming my goal is to find something cheap.  My goal is to run an experiment.

Also, you may wish to consider more stocks.  You only have one stock representing each of your factors.   We all know what can happen to any individual stock.
I think you're missing the goal of this thread - I'm not asking for advice, but running an experiment. 

Each stock captures several factors, not just one.  All the stocks are exposed to both "beta" (the stock market) and "small" (all are micro cap stocks).  In addition:
1 is selected on momentum
2 are picked based on a mixture of value and momentum
1 is purely a value pick

MustacheAndaHalf

  • Walrus Stache
  • *******
  • Posts: 6633
Re: Small Value Momentum Expeiment
« Reply #13 on: March 20, 2018, 07:06:28 PM »
OP: Looks like you're treading water... Still a little soon to determine the signal to noise level on your strategy though.  Have you made any changes to the allocation?
Using only your data for a moment, you're saying that +2.5% in one month ($106.79/4189) is considered treading water?  Isn't that like saying +34% / year is treading water?  (1.025 ^^ 12 = 1.34).  I would classify it as beating VBR by +1.2% over the past month, again using your data.

Regarding performance:
* The fund's starting assets of $4200 should be used for it's performance
* The fund spent $4189.97 on stocks, paid $8 in commissions, and had $2.03 in cash.
So that gives me the following update to your spreadsheet data:
($4296.23 + $2.03) / $4200 = 4298.26 / 4200 = 1.0234 or +2.3% for the past month

My data from after the market close differs from data you captured earlier:
"42" fund experiment = +1.33%
benchmark (PMUIX) = +1.55%
VBR = +0.88%
S&P 500 = +0.02%

So according to the experiment's benchmark, it's behind -0.22% right now.  I won't change any stocks without making a post in this thread - the fund has the same holdings as before.  It turns out 4/2 is actually a day the stock market is open, so I will definitely try to make the next purchase for the 42 fund on 4/2.

MustacheAndaHalf

  • Walrus Stache
  • *******
  • Posts: 6633
Re: Small Value Momentum Expeiment
« Reply #14 on: March 20, 2018, 07:14:12 PM »
Some of this reading may interest you as well:
https://alphaarchitect.com/alpha-architect-white-papers/
Those were worth a look, thanks.

salmo trutta

  • 5 O'Clock Shadow
  • *
  • Posts: 11
Re: Small Value Momentum Expeiment
« Reply #15 on: March 21, 2018, 08:00:56 PM »
Some of this reading may interest you as well:
https://alphaarchitect.com/alpha-architect-white-papers/
Those were worth a look, thanks.

Are you familiar with any of their products? Very interesting. I bought $VMOT after doing a lot of reading on the 4 funds that make up its components. Unique strategies for sure.  https://alphaarchitect.com/etfsite/

Cheers,

Salmo

tardis

  • Handlebar Stache
  • *****
  • Posts: 2376
  • Location: Canada
Re: Small Value Momentum Expeiment
« Reply #16 on: March 22, 2018, 12:26:13 AM »
Posting to follow since this sounds interesting.  :)

MustacheAndaHalf

  • Walrus Stache
  • *******
  • Posts: 6633
Re: Small Value Momentum Expeiment
« Reply #17 on: April 02, 2018, 08:48:13 AM »
Hopefully buying on 4/2 (for the "42" fund) will have better results than the past purchase. 

The pure performance and pure value stocks did badly, losing -26% (XOMA) and -9% (VOXX).  Some might simply suggest avoiding stocks with "XO" in their names, but I think it makes more sense to focus the fund on the point of the experiment: mixing value and momentum. 

Stock prices are changing as I type,. so when I started this message MFIN was +1% and it's now -1%.  Consider it roughly breaking even.  Meanwhile AOI gained +21% (or +23% when I started this message).  Overall the fund lost -3% in 6 weeks. 

The fund's benchmark, PMUIX, lost -1%.  The fund under performed it's benchmark by -2%, although it's closer to -3% as I keep typing.  That compares against -2% for Vanguard Small/Value and -4% for the S&P 500.

In order to focus the fund on stocks with both momentum and value, I will be picking all four stocks using the mixed criteria today and going forward. 

When I pick the top performance stocks and sort by value, the top 2 entries are MFIN and AOI.  Same when I take value stocks and sort by performance.  So those 2 stocks stay where they are.

Among top performance (micro-cap) stocks, re-sorted by value, I'm picking RFIL:
1) PTSI  +110 mtm, 1.77 p/b
2) RFIL  +151 mtm, 1.87 p/b
3) JNP   +97 mtm, 2.65 p/b

Among value (micro-cap) stocks, re-sorted by performance, I'm picking ACY:
1) ACY +52 mtm, 0.52 p/b
2) GSVL +84 mtm, 0.79 p/b
...
6) TRT +40 mtm, 0.91 p/b

I'm using a new screener that is less flexible about entering custom numbers.  As a result, I couldn't narrow down to the top 42 stocks so I cast a wider net.  I'll keep trying new screeners until I find one where the data is consistent and the usage flexible.

In order to save on costs, I don't rebalance.  Selling any part of MFIN or AOI would incur a commission to sell.  Instead, I sell the stocks that no longer match the criteria, and use the proceeds to buy the new stock picks.  Here's today's sales:

XOMA sold for $763.55 out of $1040.18, -26.6% loss (-26.3% ignoring commissions)
VOXX sold for $946.64 out of $1045.39, -9.5% loss (-9.1% ignoring commissions)

$16.43 cash
$1043.40 MFIN
$1281.60 AOI
$851.90 ACY (new: 49 shares x $17.35/sh + $2 commission)
$843.89 RFIL (new: 186 shares x $4.53/sh + $2 commission)

Note execution quality matters.  When I bought ACY, the range was $17.05 to $17.35, so $17.20 would have been good execution - but I got $0.005 under the max.  That $0.15 worth of bad execution x 50 costs about $7.50.  That also meant when I bought RFIL, I had to pick the max stock price of $4.60, but instead I got better execution with $4.53, which is $13 less than the max.  That's also why the fund has $16 in cash instead of $3 in cash.

So this experiment now only picks micro-cap stocks with a mixture of both momentum and value.  Looks like the next purchase date is May 14th.

MustacheAndaHalf

  • Walrus Stache
  • *******
  • Posts: 6633
Re: Small Value Momentum Expeiment
« Reply #18 on: April 24, 2018, 10:01:39 AM »
Owing individual, volatile micro-cap stocks is even more painful than I expected, and I plan to end this experiment.  Seeing one of these stocks up +8% one day and -10% down the next is emotionally draining to watch.  And I worry it's making "picking stocks" easier by making the volatility of everything else look small by comparison.  In short, owning individual micro-cap stocks is not for me.

Here's the record from 2/20 to 4/24 (note markets are open, your numbers may vary):
"42" fund turned $4200 into $3898.78, and requires $2/trade x 4 trades to exit.  So
total loss is about -7.36% over 2 months (annualized -37%/year: 0.926 ^^ 6)

PNC Mutli Fact Sm Value, the benchmark, gained +1.5%
USAA Small value momentum gained +2.8%
Vanguard Small Value gained +1.0%
S&P 500 index lost -2.2%
Bitcoin (BTC) lost -18.4%

MustacheAndaHalf

  • Walrus Stache
  • *******
  • Posts: 6633
Re: Small Value Momentum Expeiment
« Reply #19 on: April 24, 2018, 10:13:50 AM »
After actual sells (bid-ask spread instead of "current price"):

$3851.02 after selling + $16.43 cash = $3867.45
3867.45/4200 = .9208 or -7.9% in 2 months

Selling details:
Each sale involves $2/trade commission plus a few pennies SEC transaction fee.

AOI net amount $1020.37 (48 sh x $21.30 = $1022.40 - $2.03 fees)
ACY net amount $728.32 (49 sh x $14.905 = $730.34 - $2.02 fees)
MFIN net amount $1041.11 (235 sh x $4.4389 = $1043.14 - $2.03 fees)
RFIL net amount $1063.25 (186 sh x $5.7164 = $1063.25 - $2.03 fees)