Author Topic: Dual Momentum Investing  (Read 202191 times)

ScroogeMcDutch

  • 5 O'Clock Shadow
  • *
  • Posts: 86
Re: Dual Momentum Investing
« Reply #450 on: May 02, 2015, 03:50:26 AM »
absolute momentum feels a bit like gambling/market timing.

It feels like market timing because is market timing, pure and simple.  There is no debate about that.

I'm sort of disappointed that the upshot of this thread, on this forum, is that so many people here are actually considering trying to time the market.  Good luck with that, noobs.

Or maybe do a little research to be reminded why this is a terrible idea.  Google "market timing" and the read the first 10 links that come up, and then kick yourself for ever thinking this was a good idea.  You'll feel like someone who just narrowly escaped a cult.

I would have to be convinced by both rigorous backtesting done by myself (as in, that the lookback period or trading frequency hardly have an effect on the results) and convinced by a theoretical (but not necessarily proven) basis for the strategy.

In and of itself, there is something intellectually dissatisfying about B&H. At the same time it's easy, and provides better results than 90% of the alternatives at a much better results/time spent ratio. That said, I'm pretty sure there's more under the sun and I'm curious to the different theories. As of yet, I am firmly in B&H territory, partly due to lack of time to experiment with own backtesting and asset classes.

I may be completely deluded here (and I'm not the best at actual argumentation) and I really want to understand more of the underlying theories, but why would a broad index of stocks (or any liquid asset) grow much faster than inflation? Shouldn't bankruptcies keep things in check? Where is this continual higher growth than inflation coming from? Is it just because some people/parties hold so incredibly many stocks and will keep on holding (e.g. Bill Gates, Buffett) that the price for which they're being traded is only the tip of the trading iceberg (as we measure stock value on price last traded, not on intrinsic value only)? Compounding also works on a larger scale and at some point stocks would represent basically all the value in the world?

The growth in a stock index comes from growth in the economy.  As the economy grows, corporate profits grow, which makes people willing to pay more for company stocks.  And growth in the economy comes from growth in either (1) population, or (2) per capita consumption rate, or (3) both.  B&H indexers are assuming that the economy will grow indefinitely, and that continued indefinite growth in market capitalization will follow.  But a quick look at index performance in countries where the population is no longer growing should give indexers pause.  See: Japan since the late 1980s bubble popped, France and Britain since the 2000 bubble popped.  Germany's market seems to have found a way to continue growing.  Although I haven't really looked into it, my guess would be that they found a way to tap the export markets (i.e., sell to economies that are still growing).

Thank you for the explanation. Basically this would mean the stagnant countries (i.e. almost all western countries) bank their economic growth mainly on the overseas growth markets and their ability to capture that growth. It still seems to me we should run into compounding problems here as well, or that inflation should start increasing seriously at some point in the near future (10 years).
« Last Edit: May 02, 2015, 03:55:06 AM by ScroogeMcDutch »

forummm

  • Walrus Stache
  • *******
  • Posts: 7358
  • Senior Mustachian
Re: Dual Momentum Investing
« Reply #451 on: May 02, 2015, 06:21:37 AM »
Reading hodedofome's giant book pictures, it struck me during the "diversity and simplicity" portion that it sounded a lot like it was describing buy-and-hold indexing. As such, the book seems to say it's not dependent on particular market conditions. I think that's a big part of what attracts people to BH. Not saying that some other approach can't be robust with lots of diversity and simplicity.

brooklynguy

  • Handlebar Stache
  • *****
  • Posts: 2098
  • Age: 37
Re: Dual Momentum Investing
« Reply #452 on: May 02, 2015, 07:21:04 AM »
Hoded, I like the ecology analogy.  It provides a nice framework for describing the apparent contradiction in the DM strategy that folks have been zeroing in on.  On the one hand, the data say that the strategy has thrived across all the various markets and time periods in which it has been examined.  This suggests that the strategy is highly environment-independent--like an unspecialized, generalist species, it is able to survive under a wide a variety of conditions (to use the terminology we've been using in this thread, it is "robust").  However, because the strategy's success is entirely dependent on the predictive stength of the specific lookback period employed, it should be highly environment-specific--it wouldn't take a meteor impact or seismic climate change to kill the strategy; a subtle change in temperature would be enough, as long as it caused directional price movements to no longer follow the precise pattern needed for the specific lookback period being used to be reliable.  So how can this contradiction be explained?

hodedofome

  • Handlebar Stache
  • *****
  • Posts: 1107
  • Age: 37
  • Location: Texas
Re: Dual Momentum Investing
« Reply #453 on: May 02, 2015, 09:19:31 AM »
I was afraid the detractors would take the passages I posted and make a judgement based on limited information, possibly taken out of context.

I think I've finally realized my frustration with some on this thread. Some are quick to judge something, but really have not much idea about how or why it works. They lack context. I would say, before you make a judgement about an active strategy, read all you can about it first, then judge. If you are basing your info from only stuff posted in this thread, I'd say your info is incomplete. At the minimum, read:

 

I'll add in there Trend Following by Michael Covel and Market Wizards by Jack Schwager.

Then come back with all your reasons for why it can't work. You'll have a good context for what real trend following/momentum is and where it fits in the world. You'll also see where dual momentum as described by Gary fits in the trend following world.


Sent from my iPhone using Tapatalk

gluskap

  • Stubble
  • **
  • Posts: 156
  • Age: 37
  • Location: Los Angeles, CA
    • Money Savvy Mommy
Re: Dual Momentum Investing
« Reply #454 on: May 02, 2015, 09:31:28 AM »
I made a long post and the battery on my iPad died so I lost it doh. Just wanted to add that as a fairly new investor this thread was really interesting. I think the question we should focus on is not so much why it works but rather does it actually work.

When predicting if the sun will rise tomorrow if we look at all of known past history and see that it does, it doesn't really matter if we understand the physics of why it works or whether we are just an ignorant caveman. Of course there could be some catastrophic event such as the sun blowing up or whatnot but for the most part betting that the sun will rise tomorrow is a good bet.

One of the basic rules of investing is that past performance is not indicative of future success. But I argue that this is true because most people only look at the recent past. Just because a fund did well in the last year or last 5-10 years doesn't mean it will continue to do well next year. However if the fund or strategy did well for the last hundred years of data, that would be much more convincing regardless if we truly understood why it did well.

Miles dividends MD did give a reason for why DM works...perhaps it isn't fleshed out and to the level of detail that some would prefer but that's because the idea is new and it's not fully understood exactly why it works. But that reason shouldn't be ignored as if it didn't exist. Someone asked if you would invest in a strategy of buying every 7th Tuesday if it worked. And my answer would depend on how much back testing they did. Most of these active strategies don't work because people cherry pick certain periods of time such as the last 20-30 years and find some random pattern that seems to work. But if someone truly went back hundreds of years and the data did fit, even if I didn't understand why it worked I would still have to give it some serious consideration if it was truly shown that it did indeed work. Maybe with more testing an actual real reason not currently known would be discovered later that logically explained why it works. If you look at some of the modern theories of physics some of it sounds just as strange as an every 7th Tuesday cycle but as long as the data truly fits, it's rational to accept that hypothesis until proven untrue.

So what I want to see is more back tests that show whether it truly works or not. I think the funds used and the look back period are both important things to consider when doing these back tests. Most of the evidence shown against DM so far, didn't choose the correct funds and thus was not really DM or didn't provide enough details about the time period tested and the look back period to decide if the criticism was valid or not.

I think it's fair to say that even if this theory does work, it's possible that it might change due to changes in human behavior. An example might be that with increased technology and the faster dissemination of information, the optimal look back period of 9 months might shorten to 3 months or whatever the case in the future. However I think this is possible of B&H strategy too. Just because it's worked well so far there could be a black swan event like the fall of US govt or some bigger crash than black Monday that wipes out all investor confidence. This is a small chance though and most are willing to take that risk. So why not accept that there are certain risks to DM as well but that they might be acceptable and still be a very valid investing strategy?

I think the biggest risk of DM is the whipsaw as previously mentioned. The market has a downward correction and if your look back period is too short then you get a false signal to sell and then the market rebounds and you buy in at higher prices. Or the opposite effect where your look back period is too long and you pull out of a bear market too late or enter a bull market too slow. But this risks can be mitigated by diversifying your look back periods as miles already suggested. But this is too complex and too much trading for me.

Someone asked why doesn't everybody do this if it works and is so easy? And I'd argue that it's hard to execute. There will be periods of underperformance and human nature is hard to follow the course. I mean there are plenty of people that still don't index invest. Human nature just isn't rational which is more reason for DM. I will still probably stick mostly to index investing cuz I am lazy but I might try to test this strategy out in my smaller Roth IRA.

Again this strategy doesn't have to work perfectly to make money. A well balance 80/20 stock/bond portfolio will underperform pure stocks but there is less volatility. I think there might be some years that DM might underperform buy and hold but it seems to lower volatility and max drawdown which is a big advantage.

So let's focus more on whether this strategy really works or not and let's see more back tests.

brooklynguy

  • Handlebar Stache
  • *****
  • Posts: 2098
  • Age: 37
Re: Dual Momentum Investing
« Reply #455 on: May 02, 2015, 09:58:04 AM »
I was afraid the detractors would take the passages I posted and make a judgement based on limited information, possibly taken out of context.

I assume this refers to my post, but I wasn't passing judgment on the strategy based on the passages you posted -- I was using them to reframe the same question that has been asked of the strategy but not yet answered.  Is reading all the literature a prerequisite to participation in this thread?  I want to understand if there are reasons to believe in the merit of the strategy before I take the time to read seven volumes dedicated to it, not after.

Is the apparent contradiction I identified really a contradiction?  If so, how can it be explained?

I'm not asking these questions rhetorically assuming the answers disprove DM's validity.  And I'm not offering reasons why it doesn't work.  I'm asking questions about why it does work, if it does.  I'm asking because I'm interested and want to understand. 

forummm

  • Walrus Stache
  • *******
  • Posts: 7358
  • Senior Mustachian
Re: Dual Momentum Investing
« Reply #456 on: May 02, 2015, 10:03:29 AM »
I was afraid the detractors would take the passages I posted and make a judgement based on limited information, possibly taken out of context.

I think I've finally realized my frustration with some on this thread. Some are quick to judge something, but really have not much idea about how or why it works. They lack context. I would say, before you make a judgement about an active strategy, read all you can about it first, then judge. If you are basing your info from only stuff posted in this thread, I'd say your info is incomplete. At the minimum, read:

<<<<
trend following with managed futures by Greyserman
way of the turtle by Faith
dual momentum investing by Antonacci
following the trend by Clenow
the ivy portfolio by Faber
>>>>
I'll add in there Trend Following by Michael Covel and Market Wizards by Jack Schwager.

Then come back with all your reasons for why it can't work. You'll have a good context for what real trend following/momentum is and where it fits in the world. You'll also see where dual momentum as described by Gary fits in the trend following world.

Thanks for the recommendations. I added them to my reading list. It will be interesting to learn more about these ideas. It seems like it takes more than a giant thread to dig through them and to try to understand why they behaved as they did in the past, and whether it's reasonable to bet that they will continue to behave that way going forward.

I didn't intend my post in response to your excerpts to be reflecting judgment on the content. Just an observation of how the language makes it sound a lot like they are trying to achieve the benefits of indexing (while reducing some of the downsides). In fact, it occurred to me that the popularity of indexing actually makes this kind of trend following approach much more feasible to implement. Without it, you'd have a lot of extra risks and transaction costs and it would be much more challenging to calculate the performance to see what action you wanted to take.

Aside: I think it's funny that one of the books is "by Faith".

forummm

  • Walrus Stache
  • *******
  • Posts: 7358
  • Senior Mustachian
Re: Dual Momentum Investing
« Reply #457 on: May 02, 2015, 10:10:45 AM »
I think the question we should focus on is not so much why it works but rather does it actually work.

...

So what I want to see is more back tests that show whether it truly works or not. I think the funds used and the look back period are both important things to consider when doing these back tests. Most of the evidence shown against DM so far, didn't choose the correct funds and thus was not really DM or didn't provide enough details about the time period tested and the look back period to decide if the criticism was valid or not.

...

So let's focus more on whether this strategy really works or not and let's see more back tests.

I think it will be hard to back test this in ways other than using the author's own analysis and just trusting it. It was hard for Shiller to come up with all his data just for the US S&P 500 (or equivalent) and 10-yr Treasury rates. I've thought about different analysis designs from data I'm aware of and I think some on this thread would object to them. I also have a lot of work in my day job so I haven't had time to really dig into this more. I think the reason the backtests shown here (and approved of as being sufficiently DM) are limited to the last 20 years is because that's the analysis that was easy for people to do. Someone would have to put in some effort to do something more substantial than that.

hodedofome

  • Handlebar Stache
  • *****
  • Posts: 1107
  • Age: 37
  • Location: Texas
Re: Dual Momentum Investing
« Reply #458 on: May 02, 2015, 08:24:47 PM »
I decided to backtest MDMD's strategy of using different slices of his portfolio for each lookback period, instead of averaging out the lookback periods like I've seen some do. So this test takes VTSMX (Vanguard US Total Stock Market), VGTSX (Vanguard Int'l Total Stock Market) and VBMFX (Vanguard Total Bond Index) and does 25% each for 3 month, 6 month, 9 month and 12 month lookback periods. So each lookback period is it's own 'slice' of the portfolio. This should prevent at least some of the curve fitting that some might have an issue with a specific lookback period. You just decide to take them all. Unfortunately ETFReplay will only allow you to do backtests to 2003.





Link to the entire test showing holdings for each monthly period https://drive.google.com/file/d/0BzyyTlvGE-T2UXFwMk05N2FPZmM/view?usp=sharing
« Last Edit: May 02, 2015, 08:27:57 PM by hodedofome »

milesdividendmd

  • Handlebar Stache
  • *****
  • Posts: 1913
  • Location: Portlandia
    • Miles Dividend MD
Re: Dual Momentum Investing
« Reply #459 on: May 03, 2015, 12:59:56 AM »
A new post on the why's of Dual momentum (and their irrelevance in the end.)

http://www.milesdividendmd.com/bedtime-stories/

Enjoy!
I can teach you to play the miles game in 30 days.  http://www.travelmiles101.com/travel-rewards-course-registration.  (A free course by mustachians for mustachians)

Monkey Uncle

  • Pencil Stache
  • ****
  • Posts: 952
  • Location: West-by-god-Virginia
Re: Dual Momentum Investing
« Reply #460 on: May 03, 2015, 04:33:51 AM »
A new post on the why's of Dual momentum (and their irrelevance in the end.)

http://www.milesdividendmd.com/bedtime-stories/

Enjoy!

Thanks for that, Miles.  I've been thinking for a while that there is more to the momentum story than just trader/investor psychology.  You expressed my thoughts exactly with this:

Quote
But there is reason to believe that future bear markets will continue to look enough like past bear markets that trendfollowing approaches will continue to almost always work at mitigating draw downs.

Why do I say this? Because when we are talking about large scale expansion and recession, we are talking about the business cycle. And when we are talking about the business cycle we’re talking about the movements of a large economies. And large economies are like battleships, not Jet Ski’s. They cannot turn on a dime.

It takes time for bubbles to inflate. And it takes time for risks to work their way through a system. And when an economy begins shrinking, it takes time to for those in power to recognize that it is in fact shrinking. And when second order actions occur, and interest rates are dropped by central banks, and stimulus bills are passed by governments, it takes time for the pain to work its way through the system, and for the corrective actions to have any effect at all.

Ultimately, asset prices reflect the earnings that those assets are generating (recognizing, of course, the existence of short-term anomalies like the late '90s tech bubble and the '87 crash, which are attributable largely to investor/trader psychology).  Earnings are tied to the economy, which generally moves in long, broad trends.  Hence the apparent inability of traders to arbitrage away momentum.  So perhaps the theories of momentum and the efficient market are compatible after all.

But I think your article under-emphasizes the importance of the story behind any investment strategy.  No, we would not invest in a good story if it was not validated by back-testing.  But we also (most of us, anyway) wouldn't invest in a system that back-tests well but has no rational explanation (see the 7th Tuesday strategy that was discussed ad nauseum up-thread).
« Last Edit: May 03, 2015, 04:52:34 AM by Monkey Uncle »
"Take this job and shove it" - David Allan Coe

milesdividendmd

  • Handlebar Stache
  • *****
  • Posts: 1913
  • Location: Portlandia
    • Miles Dividend MD
Re: Dual Momentum Investing
« Reply #461 on: May 03, 2015, 11:54:38 AM »
Fair point about my under emphasizing the importance of story.

In the end stories are very important. Not because they are determinative, but because they help us to individually stick with an approach through thick and thin.

To me empiricism is a more useful tool for determining probabilities, but weighing probabilities won't help us at all if we can't stick to our own systems. (The biggest risk in any approach.)

As I have mentioned many times in this thread, the momentum and trend following stories jibe with my way of seeing the world.

I'm just suspicious of myself, that the convincing aspect of the story has followed the empirical observation that these approaches work. Not the other way around.
I can teach you to play the miles game in 30 days.  http://www.travelmiles101.com/travel-rewards-course-registration.  (A free course by mustachians for mustachians)

sirdoug007

  • Pencil Stache
  • ****
  • Posts: 583
  • Age: 37
  • Location: Houston, TX
Re: Dual Momentum Investing
« Reply #462 on: May 06, 2015, 03:17:55 PM »
I think the question we should focus on is not so much why it works but rather does it actually work.

...

So what I want to see is more back tests that show whether it truly works or not. I think the funds used and the look back period are both important things to consider when doing these back tests. Most of the evidence shown against DM so far, didn't choose the correct funds and thus was not really DM or didn't provide enough details about the time period tested and the look back period to decide if the criticism was valid or not.

...

So let's focus more on whether this strategy really works or not and let's see more back tests.

I think it will be hard to back test this in ways other than using the author's own analysis and just trusting it. It was hard for Shiller to come up with all his data just for the US S&P 500 (or equivalent) and 10-yr Treasury rates. I've thought about different analysis designs from data I'm aware of and I think some on this thread would object to them. I also have a lot of work in my day job so I haven't had time to really dig into this more. I think the reason the backtests shown here (and approved of as being sufficiently DM) are limited to the last 20 years is because that's the analysis that was easy for people to do. Someone would have to put in some effort to do something more substantial than that.

Speaking of Shiller's data, I decided to do a simple absolute momentum backtest using his S&P500 data-set. 

Basically I look back 6 months and if the S&P500 return is positive, I take put the money in the S&P500.  If the last 6 months of the S&P500 is negative, I keep the value in cash (no appreciation at all).  I then started the backtest at the beginning of every decade starting in January 1880 (Shiller's data starts in 1871).

The spreadsheet is attached.  Let me know if anyone sees any errors.

The results show this absolute momentum strategy beats the S&P500 in 12 of the 14 decades.  The outperformance is more than 50% over 10 years in 5 of the 14 decades.  In two of the decades (the 1990s and 2010s so far), this strategy has underperformed the S&P500.  In the 90's it underperformed with a 130% gain for AM vs. 218% gain for the S&P500 and in the 2010s so far it has underpeformed by with a 37% gain vs. 72% gain for the S&P500.

The decade by decade chart shows that this strategy is good at limiting downside risk but does so at the expense of bull market gains.  Overall this is pretty damn impressive IMHO.  Like MDMD said, this can be considered a cowards strategy as the real performance advantage is getting a sell signal to avoid typical bear markets.  Will bear markets continue to be "typical"?  Who knows, 12 out of 14 ain't bad.

milesdividendmd

  • Handlebar Stache
  • *****
  • Posts: 1913
  • Location: Portlandia
    • Miles Dividend MD
Re: Dual Momentum Investing
« Reply #463 on: May 06, 2015, 05:48:54 PM »
Thanks for sharing this.

In my view, the data does speak for itself. absolute momentum reproducibly diminishes drawdowns, (as do moving average approaches by the way.)

In my view absolute momentum is the main event of dual momentum. But relative momentum is very complementary in that it tends to boost returns during bull markets, which addresses trend following approaches major weakness.


I can teach you to play the miles game in 30 days.  http://www.travelmiles101.com/travel-rewards-course-registration.  (A free course by mustachians for mustachians)

forummm

  • Walrus Stache
  • *******
  • Posts: 7358
  • Senior Mustachian
Re: Dual Momentum Investing
« Reply #464 on: May 06, 2015, 08:03:45 PM »
I think the question we should focus on is not so much why it works but rather does it actually work.

...

So what I want to see is more back tests that show whether it truly works or not. I think the funds used and the look back period are both important things to consider when doing these back tests. Most of the evidence shown against DM so far, didn't choose the correct funds and thus was not really DM or didn't provide enough details about the time period tested and the look back period to decide if the criticism was valid or not.

...

So let's focus more on whether this strategy really works or not and let's see more back tests.

I think it will be hard to back test this in ways other than using the author's own analysis and just trusting it. It was hard for Shiller to come up with all his data just for the US S&P 500 (or equivalent) and 10-yr Treasury rates. I've thought about different analysis designs from data I'm aware of and I think some on this thread would object to them. I also have a lot of work in my day job so I haven't had time to really dig into this more. I think the reason the backtests shown here (and approved of as being sufficiently DM) are limited to the last 20 years is because that's the analysis that was easy for people to do. Someone would have to put in some effort to do something more substantial than that.

Speaking of Shiller's data, I decided to do a simple absolute momentum backtest using his S&P500 data-set. 

Basically I look back 6 months and if the S&P500 return is positive, I take put the money in the S&P500.  If the last 6 months of the S&P500 is negative, I keep the value in cash (no appreciation at all).  I then started the backtest at the beginning of every decade starting in January 1880 (Shiller's data starts in 1871).

The spreadsheet is attached.  Let me know if anyone sees any errors.

The results show this absolute momentum strategy beats the S&P500 in 12 of the 14 decades.  The outperformance is more than 50% over 10 years in 5 of the 14 decades.  In two of the decades (the 1990s and 2010s so far), this strategy has underperformed the S&P500.  In the 90's it underperformed with a 130% gain for AM vs. 218% gain for the S&P500 and in the 2010s so far it has underpeformed by with a 37% gain vs. 72% gain for the S&P500.

The decade by decade chart shows that this strategy is good at limiting downside risk but does so at the expense of bull market gains.  Overall this is pretty damn impressive IMHO.  Like MDMD said, this can be considered a cowards strategy as the real performance advantage is getting a sell signal to avoid typical bear markets.  Will bear markets continue to be "typical"?  Who knows, 12 out of 14 ain't bad.

Thanks for doing this! I was actually going to do almost the same thing but haven't had time. I thought about using the Treasury rates as the yield during the cash months. Looks like you didn't do that, but also didn't include any transaction costs. So if you put your cash into treasuries, perhaps that would more than cover the transaction costs. I'll dig into this more over time.

hodedofome

  • Handlebar Stache
  • *****
  • Posts: 1107
  • Age: 37
  • Location: Texas
Re: Dual Momentum Investing
« Reply #465 on: May 06, 2015, 08:22:17 PM »
Market Wizards is a good book to read about trend following, especially the Larry Hite interview. In regards to why he thinks trend following works:

"I knew that if you traded across the board (meaning trading multiple/diverse/uncorrelated markets), controlled your risk, and went with the trend, it just had to work. I could see it absolutely clearly."

If you can't visualize having a diverse group of uncorrelated markets, buying the ones that are going up, and selling the ones that are going down, and seeing how that could work in the long run, then I just can't convince you otherwise.

More good quotes:

"It is incredible how rich you can get by not being perfect. We are not looking for the optimum method, we are looking for the hardiest method."

I'm summarizing a longer passage here but the rules Larry lives by as a money manager is:

1) Manage risk first. Risk is a no-fooling-around game

2) Always follow the trends and never deviate from our methods

3) Diversify - we trade as many markets as possible and we use lots of different systems ranging from short term to long term. Some of these systems may not be that good by themselves, but we really don't care. That is not what they are there for

4) Track volatility. When the volatility of a market becomes so great that it adversely skews the expected risk/return ratio, we will stop trading that market

sirdoug007

  • Pencil Stache
  • ****
  • Posts: 583
  • Age: 37
  • Location: Houston, TX
Re: Dual Momentum Investing
« Reply #466 on: May 07, 2015, 10:05:48 AM »
Thanks for doing this! I was actually going to do almost the same thing but haven't had time. I thought about using the Treasury rates as the yield during the cash months. Looks like you didn't do that, but also didn't include any transaction costs. So if you put your cash into treasuries, perhaps that would more than cover the transaction costs. I'll dig into this more over time.

I considered using Shiller's "Long Interest Rate GS10" as the safe alternative at first but then realized it would be much more complicated to account for changes in bond value with changing interest rates.  Obviously you can do better than sit in cash, but his shows even using absolute momentum and sitting in cash at times can produce very good results.

Financial.Velociraptor

  • Pencil Stache
  • ****
  • Posts: 931
  • Age: 45
  • Location: Houston TX
  • Devour your prey raptors!
    • Financial Velociraptor
Re: Dual Momentum Investing
« Reply #467 on: May 07, 2015, 10:12:32 AM »
CAPE is a weird bunny.  You use an intermediate length of time to smooth P/E but Shiller et.al. never defined how many years constituted a cycle.  So people have sort of latched on to 10 years because it is a round number.  Trouble is, things look ugly if you use the standard 10 year CAPE on the US right now but things are rosy if you assume 5 years rounds out a cycle.

It is supposed to be an objective measure but its application is very subjective.
Achieve Financial Escape Velocity - Financial Velociraptor

forummm

  • Walrus Stache
  • *******
  • Posts: 7358
  • Senior Mustachian
Re: Dual Momentum Investing
« Reply #468 on: May 07, 2015, 10:50:32 AM »
Market Wizards is a good book to read about trend following, especially the Larry Hite interview. In regards to why he thinks trend following works:

"I knew that if you traded across the board (meaning trading multiple/diverse/uncorrelated markets), controlled your risk, and went with the trend, it just had to work. I could see it absolutely clearly."

If you can't visualize having a diverse group of uncorrelated markets, buying the ones that are going up, and selling the ones that are going down, and seeing how that could work in the long run, then I just can't convince you otherwise.

It sounds a little like belief in a religion. You have to have faith. Logic doesn't really come into play. You've seen evidence in the past that could tell you that God exists and you're living your life like it's so.

4) Track volatility. When the volatility of a market becomes so great that it adversely skews the expected risk/return ratio, we will stop trading that market[/i]

Does stop trading mean get out? And not "stay put"?

forummm

  • Walrus Stache
  • *******
  • Posts: 7358
  • Senior Mustachian
Re: Dual Momentum Investing
« Reply #469 on: May 07, 2015, 10:51:43 AM »
Thanks for doing this! I was actually going to do almost the same thing but haven't had time. I thought about using the Treasury rates as the yield during the cash months. Looks like you didn't do that, but also didn't include any transaction costs. So if you put your cash into treasuries, perhaps that would more than cover the transaction costs. I'll dig into this more over time.

I considered using Shiller's "Long Interest Rate GS10" as the safe alternative at first but then realized it would be much more complicated to account for changes in bond value with changing interest rates.  Obviously you can do better than sit in cash, but his shows even using absolute momentum and sitting in cash at times can produce very good results.

Yes, I had the same thought. It kept me from having a better idea and doing it.

forummm

  • Walrus Stache
  • *******
  • Posts: 7358
  • Senior Mustachian
Re: Dual Momentum Investing
« Reply #470 on: May 07, 2015, 10:55:15 AM »
CAPE is a weird bunny.  You use an intermediate length of time to smooth P/E but Shiller et.al. never defined how many years constituted a cycle.  So people have sort of latched on to 10 years because it is a round number.  Trouble is, things look ugly if you use the standard 10 year CAPE on the US right now but things are rosy if you assume 5 years rounds out a cycle.

It is supposed to be an objective measure but its application is very subjective.

PEs are still pretty high historically. VTSAX is 21.4. VFIAX is 19.8. I think historical is around 15-16.

hodedofome

  • Handlebar Stache
  • *****
  • Posts: 1107
  • Age: 37
  • Location: Texas
Re: Dual Momentum Investing
« Reply #471 on: May 07, 2015, 11:20:49 AM »
It sounds a little like belief in a religion. You have to have faith. Logic doesn't really come into play. You've seen evidence in the past that could tell you that God exists and you're living your life like it's so.

The same could be said for buy and hold. No matter what strategy you choose, you are expecting the future to be somewhat like the past. It is only the 'true believers' that are disciplined enough to follow a strategy for the rest of their lives, and that includes the guys who bought and held equities for the past 100+ years. At any moment their strategy could have stopped working for good. All of us have to decide what we truly believe about the markets, and then devise a strategy around those beliefs.

4) Track volatility. When the volatility of a market becomes so great that it adversely skews the expected risk/return ratio, we will stop trading that market[/i]

Does stop trading mean get out? And not "stay put"?

This is specific to their strategy, but yes they have volatility limits on individual markets. If a market gets too volatile, they will get out completely. Even if the volatility is in the same direction as their position.

sirdoug007

  • Pencil Stache
  • ****
  • Posts: 583
  • Age: 37
  • Location: Houston, TX
Re: Dual Momentum Investing
« Reply #472 on: May 07, 2015, 01:07:40 PM »
I think the question we should focus on is not so much why it works but rather does it actually work.

...

So what I want to see is more back tests that show whether it truly works or not. I think the funds used and the look back period are both important things to consider when doing these back tests. Most of the evidence shown against DM so far, didn't choose the correct funds and thus was not really DM or didn't provide enough details about the time period tested and the look back period to decide if the criticism was valid or not.

...

So let's focus more on whether this strategy really works or not and let's see more back tests.

I think it will be hard to back test this in ways other than using the author's own analysis and just trusting it. It was hard for Shiller to come up with all his data just for the US S&P 500 (or equivalent) and 10-yr Treasury rates. I've thought about different analysis designs from data I'm aware of and I think some on this thread would object to them. I also have a lot of work in my day job so I haven't had time to really dig into this more. I think the reason the backtests shown here (and approved of as being sufficiently DM) are limited to the last 20 years is because that's the analysis that was easy for people to do. Someone would have to put in some effort to do something more substantial than that.

Speaking of Shiller's data, I decided to do a simple absolute momentum backtest using his S&P500 data-set. 

Basically I look back 6 months and if the S&P500 return is positive, I take put the money in the S&P500.  If the last 6 months of the S&P500 is negative, I keep the value in cash (no appreciation at all).  I then started the backtest at the beginning of every decade starting in January 1880 (Shiller's data starts in 1871).

The spreadsheet is attached.  Let me know if anyone sees any errors.

The results show this absolute momentum strategy beats the S&P500 in 12 of the 14 decades.  The outperformance is more than 50% over 10 years in 5 of the 14 decades.  In two of the decades (the 1990s and 2010s so far), this strategy has underperformed the S&P500.  In the 90's it underperformed with a 130% gain for AM vs. 218% gain for the S&P500 and in the 2010s so far it has underpeformed by with a 37% gain vs. 72% gain for the S&P500.

The decade by decade chart shows that this strategy is good at limiting downside risk but does so at the expense of bull market gains.  Overall this is pretty damn impressive IMHO.  Like MDMD said, this can be considered a cowards strategy as the real performance advantage is getting a sell signal to avoid typical bear markets.  Will bear markets continue to be "typical"?  Who knows, 12 out of 14 ain't bad.

I realized I missed a very important factor in this spreadsheet: DIVIDENDS!!!

Adding them in makes this picture look very different.  AM only beats B&H in the 1910s, 1930s, 1970s, and 2000s (e.g., when the stock market was absolutely awful!).

In bull markets AM significantly lags B&H.  For example, $100 invested in 1990 grows to $405 in the S&P500  in 1999 and only $230 with AM. 

Probably not coincidental that the 2000s happened to be the best decade for AM EVER.  Great time to write a book after killing an awful stock market in the 2000s.

It's starting to look like an expensive hedging strategy.  No free lunch here.
« Last Edit: May 07, 2015, 01:09:50 PM by sirdoug007 »

skyrefuge

  • Pencil Stache
  • ****
  • Posts: 999
  • Location: Suburban Chicago, IL
Re: Dual Momentum Investing
« Reply #473 on: May 07, 2015, 02:57:45 PM »
I realized I missed a very important factor in this spreadsheet: DIVIDENDS!!!

For the record, while you now have dividends in your returns, it looks like you still aren't including dividends for your decision points. It doesn't have a huge effect, but it's still an important point.  If you made your decisions based on 6-month total-return rather than price differential, it looks like the AM strategy spends 17% more time in stocks, which increases its averaged annualized return from 5.4% to 5.5%. 1880s are also added as a decade that beats all-stock, but for some of the stock-beating decades, the margin of victory is actually decreased (i.e., to the extent that included dividends made you stay in stocks, that was a "bad" decision).

forummm

  • Walrus Stache
  • *******
  • Posts: 7358
  • Senior Mustachian
Re: Dual Momentum Investing
« Reply #474 on: May 07, 2015, 05:53:59 PM »
If you include the dividends in decision making, over the ~130 year time period (instead of starting over each decade), which wins?

skyrefuge

  • Pencil Stache
  • ****
  • Posts: 999
  • Location: Suburban Chicago, IL
Re: Dual Momentum Investing
« Reply #475 on: May 07, 2015, 09:39:31 PM »
If you include the dividends in decision making, over the ~130 year time period (instead of starting over each decade), which wins?

Absolute Momentum: $90,254
Stocks Only: $1,247,873

And no, I didn't forget a digit in the AM number.

EDIT: I discovered that the above figure for AM is incorrect. See this post for my, uh, "correction"?
« Last Edit: May 08, 2015, 02:53:49 PM by skyrefuge »

sol

  • Walrus Stache
  • *******
  • Posts: 5345
  • Age: 40
  • Location: Pacific Northwest
Re: Dual Momentum Investing
« Reply #476 on: May 07, 2015, 10:25:47 PM »
Absolute Momentum: $90,254
Stocks Only: $1,247,873

And no, I didn't forget a digit in the AM number.

I'm not sure how any of these comparisons matter.  If it's not AM or DM, then there will be some other technical trading metric that outperforms an index investor.  With enough random strategies, some of them are guaranteed to outperform over any finite time period.

If you're using sky's comparison above to convince yourself that buy and hold is superior to technical trading because buy and hold has offered better returns, then you've already tacitly accepted the argument made by miles and others that performance chasing a winning strategy is the best way to invest.  As soon as you find some technical trading scheme that outperforms the index over your chosen time period, you should then logically adopt that instead.

And I'm not drinking that kool aid.  I recognize and accept that there will always be some arbitrarily complex method of timing the market that would have provided superior returns in the past, but that does not tell me anything about how that strategy will perform in the future.

arebelspy

  • Administrator
  • Senior Mustachian
  • *****
  • Posts: 25810
  • Age: -999
  • Location: Traveling the World
Re: Dual Momentum Investing
« Reply #477 on: May 07, 2015, 11:07:15 PM »
Absolute Momentum: $90,254
Stocks Only: $1,247,873

And no, I didn't forget a digit in the AM number.

I'm not sure how any of these comparisons matter.  If it's not AM or DM, then there will be some other technical trading metric that outperforms an index investor.  With enough random strategies, some of them are guaranteed to outperform over any finite time period.

If you're using sky's comparison above to convince yourself that buy and hold is superior to technical trading because buy and hold has offered better returns, then you've already tacitly accepted the argument made by miles and others that performance chasing a winning strategy is the best way to invest.  As soon as you find some technical trading scheme that outperforms the index over your chosen time period, you should then logically adopt that instead.

And I'm not drinking that kool aid.  I recognize and accept that there will always be some arbitrarily complex method of timing the market that would have provided superior returns in the past, but that does not tell me anything about how that strategy will perform in the future.

We are two former teachers who accumulated a bunch of real estate, retired at 29, and now travel the world full time with a kid.
If you want to know more about me, or how we did that, or see lots of pictures, this Business Insider profile tells our story pretty well.
We (occasionally) blog at AdventuringAlong.com.
You can also read my forum "Journal."

milesdividendmd

  • Handlebar Stache
  • *****
  • Posts: 1913
  • Location: Portlandia
    • Miles Dividend MD
Re: Dual Momentum Investing
« Reply #478 on: May 08, 2015, 01:13:02 AM »
If you include the dividends in decision making, over the ~130 year time period (instead of starting over each decade), which wins?

Absolute Momentum: $90,254
Stocks Only: $1,247,873

And no, I didn't forget a digit in the AM number.

Please share your work.  It's impossible to understand your results if you just throw up numbers like this. 

I'm extremely dubious of your claim based on total returns data going back to 1950, using 10 month MA:

https://www.portfoliovisualizer.com/test-market-timing-model?s=y&outOfMarketAssetType=1&timingPeriods[0]=5&timingWeights[0]=100&volatilityWindowSizeInDays=0&timingWeights[2]=0&riskControl=false&endYear=2014&volatilityPeriodWeight=0&movingAverageType=1&windowSize=10&riskWindowSizeInDays=0&timingUnits[0]=2&timingUnits[2]=2&timingModel=1&volatilityWindowSize=0&startYear=1950&assetsToHold=1&timingWeights[1]=0&windowSizeInDays=105&s=y&volatilityPeriodUnit=1&multipleTimingPeriods=false&timingUnits[1]=2&riskWindowSize=10&rebalancePeriod=1

I suspect your calculations are seriously flawed, (but have no way of checking your work.)

Absolute momentum should have similar returns (sometimes better, sometimes worse depending on the decade) with markedly diminished drawdowns, when using total returns both for lookback and for returns calculation.  AKA similar CAGR with a far superior Sharp/Sortino.
« Last Edit: May 08, 2015, 01:18:17 AM by milesdividendmd »
I can teach you to play the miles game in 30 days.  http://www.travelmiles101.com/travel-rewards-course-registration.  (A free course by mustachians for mustachians)

milesdividendmd

  • Handlebar Stache
  • *****
  • Posts: 1913
  • Location: Portlandia
    • Miles Dividend MD
Re: Dual Momentum Investing
« Reply #479 on: May 08, 2015, 01:17:37 AM »
Absolute Momentum: $90,254
Stocks Only: $1,247,873

And no, I didn't forget a digit in the AM number.

I'm not sure how any of these comparisons matter.  If it's not AM or DM, then there will be some other technical trading metric that outperforms an index investor.  With enough random strategies, some of them are guaranteed to outperform over any finite time period.

If you're using sky's comparison above to convince yourself that buy and hold is superior to technical trading because buy and hold has offered better returns, then you've already tacitly accepted the argument made by miles and others that performance chasing a winning strategy is the best way to invest.  As soon as you find some technical trading scheme that outperforms the index over your chosen time period, you should then logically adopt that instead.

And I'm not drinking that kool aid.  I recognize and accept that there will always be some arbitrarily complex method of timing the market that would have provided superior returns in the past, but that does not tell me anything about how that strategy will perform in the future.

This statement proves that in the end it is just a matter of faith.  And you are a true believer.

If the most important factor for investing success is sticking to your guns through thick and thin, (and I think that it is,)  then your complete disinterest in performance speaks very positively for your chances of doing very well indeed.  (Not ironically intended.)
I can teach you to play the miles game in 30 days.  http://www.travelmiles101.com/travel-rewards-course-registration.  (A free course by mustachians for mustachians)

milesdividendmd

  • Handlebar Stache
  • *****
  • Posts: 1913
  • Location: Portlandia
    • Miles Dividend MD
Re: Dual Momentum Investing
« Reply #480 on: May 08, 2015, 01:28:22 AM »
I think the question we should focus on is not so much why it works but rather does it actually work.

...

So what I want to see is more back tests that show whether it truly works or not. I think the funds used and the look back period are both important things to consider when doing these back tests. Most of the evidence shown against DM so far, didn't choose the correct funds and thus was not really DM or didn't provide enough details about the time period tested and the look back period to decide if the criticism was valid or not.

...

So let's focus more on whether this strategy really works or not and let's see more back tests.

I think it will be hard to back test this in ways other than using the author's own analysis and just trusting it. It was hard for Shiller to come up with all his data just for the US S&P 500 (or equivalent) and 10-yr Treasury rates. I've thought about different analysis designs from data I'm aware of and I think some on this thread would object to them. I also have a lot of work in my day job so I haven't had time to really dig into this more. I think the reason the backtests shown here (and approved of as being sufficiently DM) are limited to the last 20 years is because that's the analysis that was easy for people to do. Someone would have to put in some effort to do something more substantial than that.

Speaking of Shiller's data, I decided to do a simple absolute momentum backtest using his S&P500 data-set. 

Basically I look back 6 months and if the S&P500 return is positive, I take put the money in the S&P500.  If the last 6 months of the S&P500 is negative, I keep the value in cash (no appreciation at all).  I then started the backtest at the beginning of every decade starting in January 1880 (Shiller's data starts in 1871).

The spreadsheet is attached.  Let me know if anyone sees any errors.

The results show this absolute momentum strategy beats the S&P500 in 12 of the 14 decades.  The outperformance is more than 50% over 10 years in 5 of the 14 decades.  In two of the decades (the 1990s and 2010s so far), this strategy has underperformed the S&P500.  In the 90's it underperformed with a 130% gain for AM vs. 218% gain for the S&P500 and in the 2010s so far it has underpeformed by with a 37% gain vs. 72% gain for the S&P500.

The decade by decade chart shows that this strategy is good at limiting downside risk but does so at the expense of bull market gains.  Overall this is pretty damn impressive IMHO.  Like MDMD said, this can be considered a cowards strategy as the real performance advantage is getting a sell signal to avoid typical bear markets.  Will bear markets continue to be "typical"?  Who knows, 12 out of 14 ain't bad.

I realized I missed a very important factor in this spreadsheet: DIVIDENDS!!!

Adding them in makes this picture look very different.  AM only beats B&H in the 1910s, 1930s, 1970s, and 2000s (e.g., when the stock market was absolutely awful!).

In bull markets AM significantly lags B&H.  For example, $100 invested in 1990 grows to $405 in the S&P500  in 1999 and only $230 with AM. 

Probably not coincidental that the 2000s happened to be the best decade for AM EVER.  Great time to write a book after killing an awful stock market in the 2000s.

It's starting to look like an expensive hedging strategy.  No free lunch here.

For some reason your spreadsheet contains no interpretable data for me.  (perhaps its an apple thing?)

But in the body of your response please include a full-ish analysis including standard deviation, max drawdown, sharp/sortino ratios, in the sharing of your findings.

A priori, I would not expect absolute momentum to outperform in any decade without a significant bear market.  The goal of AM is absolutely not to beat a 100% stock buy and hold portfolio in terms of CAGR.  It is to decrease the downside risk (period full stop.)

Put another way, on a risk/reward basis, what percentage of bonds would you have to hold in order to decrease S&P drawdowns to the level of an AM portfolio?  What would your total returns be with such a portfolio?

Relative momentum increases upside without diminishing drawdowns or volatility.  this is why DM is so attractive, increased upside, decreased downside.
I can teach you to play the miles game in 30 days.  http://www.travelmiles101.com/travel-rewards-course-registration.  (A free course by mustachians for mustachians)

skyrefuge

  • Pencil Stache
  • ****
  • Posts: 999
  • Location: Suburban Chicago, IL
Re: Dual Momentum Investing
« Reply #481 on: May 08, 2015, 08:52:38 AM »
Please share your work.  It's impossible to understand your results if you just throw up numbers like this. 

It's not my work, it's sirdoug007's work, which he showed. All I did was change it to use total returns for the lookback decision. That change actually improved AM. Without it, AM ended with $77,836.

I also tried 3, 9, and 12-month lookback periods, and those were all worse for AM than the original 6-month.

But yeah, there certainly could be stuff in there that's completely wrong, so please, review away! I uploaded the spreadsheet as a Google Doc, maybe that will work better for you? (only the 'Data' sheet seemed to upload, but that's the important one).

In my view, the data does speak for itself.
I'm extremely dubious of your claim

More proof that this is a matter of faith!

I'm guessing the "extreme-ness" of the final result is mostly just an effect of exponential-growth over really long time periods. The CAGRs aren't shockingly different (4.9% vs. 6.8%), but that percentage difference is enough to result in a huge nominal difference when persisting over 144 years.

milesdividendmd

  • Handlebar Stache
  • *****
  • Posts: 1913
  • Location: Portlandia
    • Miles Dividend MD
Re: Dual Momentum Investing
« Reply #482 on: May 08, 2015, 10:25:00 AM »

Please share your work.  It's impossible to understand your results if you just throw up numbers like this. 

It's not my work, it's sirdoug007's work, which he showed. All I did was change it to use total returns for the lookback decision. That change actually improved AM. Without it, AM ended with $77,836.

I also tried 3, 9, and 12-month lookback periods, and those were all worse for AM than the original 6-month.

But yeah, there certainly could be stuff in there that's completely wrong, so please, review away! I uploaded the spreadsheet as a Google Doc, maybe that will work better for you? (only the 'Data' sheet seemed to upload, but that's the important one).

In my view, the data does speak for itself.
I'm extremely dubious of your claim

More proof that this is a matter of faith!

I'm guessing the "extreme-ness" of the final result is mostly just an effect of exponential-growth over really long time periods. The CAGRs aren't shockingly different (4.9% vs. 6.8%), but that percentage difference is enough to result in a huge nominal difference when persisting over 144 years.

Faith (or bias) is a factor, of course, with me and everyone else.

But my suspicion has to do largely with the disagreement of this data with the excellent total return data that is present for trendfollowing approaches (including DM) over long time frames.

It would be useful to do an apples to apples comparison of total returns data with disparate results. Don't you agree.
I can teach you to play the miles game in 30 days.  http://www.travelmiles101.com/travel-rewards-course-registration.  (A free course by mustachians for mustachians)

forummm

  • Walrus Stache
  • *******
  • Posts: 7358
  • Senior Mustachian
Re: Dual Momentum Investing
« Reply #483 on: May 08, 2015, 10:26:04 AM »
Absolute Momentum: $90,254
Stocks Only: $1,247,873

And no, I didn't forget a digit in the AM number.

I'm not sure how any of these comparisons matter.  If it's not AM or DM, then there will be some other technical trading metric that outperforms an index investor.  With enough random strategies, some of them are guaranteed to outperform over any finite time period.

If you're using sky's comparison above to convince yourself that buy and hold is superior to technical trading because buy and hold has offered better returns, then you've already tacitly accepted the argument made by miles and others that performance chasing a winning strategy is the best way to invest.  As soon as you find some technical trading scheme that outperforms the index over your chosen time period, you should then logically adopt that instead.

And I'm not drinking that kool aid.  I recognize and accept that there will always be some arbitrarily complex method of timing the market that would have provided superior returns in the past, but that does not tell me anything about how that strategy will perform in the future.

Other than to buy my house, I've never taken a penny out of the market. I have been 100% stock, 100% index funds, since 2000 when I started investing. I haven't been convinced by momentum. But I'm an academic to my core and am fascinated by the various strategies people have devised and find it very interesting to see why people like them, why they may have worked, why they may not have worked, why they may not have worked but people seem to think they have, how things have possibly changed, and what that might mean for the future. I want to be prepared for whatever might come. And I just find this all very interesting. I like making spreadsheets and thinking this stuff through. And I am very open to doing something different with my money in the future. But I would need some pretty good evidence before I did that.

I'm enjoying this conversation and all the others that we've had on this forum. It's a pretty great place where we can explore ideas together and think them through. And there are some great people here who have really interesting ideas and think about things in ways that I've learned from. It's great that we can all collaborate and have this discussion together. I don't know anyone in my non-Internet life that I could talk to about any of this stuff.

So, thanks everyone!

forummm

  • Walrus Stache
  • *******
  • Posts: 7358
  • Senior Mustachian
Re: Dual Momentum Investing
« Reply #484 on: May 08, 2015, 10:30:01 AM »
If you include the dividends in decision making, over the ~130 year time period (instead of starting over each decade), which wins?

Absolute Momentum: $90,254
Stocks Only: $1,247,873

And no, I didn't forget a digit in the AM number.

Please share your work.  It's impossible to understand your results if you just throw up numbers like this. 

I'm extremely dubious of your claim based on total returns data going back to 1950, using 10 month MA:

https://www.portfoliovisualizer.com/test-market-timing-model?s=y&outOfMarketAssetType=1&timingPeriods[0]=5&timingWeights[0]=100&volatilityWindowSizeInDays=0&timingWeights[2]=0&riskControl=false&endYear=2014&volatilityPeriodWeight=0&movingAverageType=1&windowSize=10&riskWindowSizeInDays=0&timingUnits[0]=2&timingUnits[2]=2&timingModel=1&volatilityWindowSize=0&startYear=1950&assetsToHold=1&timingWeights[1]=0&windowSizeInDays=105&s=y&volatilityPeriodUnit=1&multipleTimingPeriods=false&timingUnits[1]=2&riskWindowSize=10&rebalancePeriod=1

I suspect your calculations are seriously flawed, (but have no way of checking your work.)

Absolute momentum should have similar returns (sometimes better, sometimes worse depending on the decade) with markedly diminished drawdowns, when using total returns both for lookback and for returns calculation.  AKA similar CAGR with a far superior Sharp/Sortino.

When I posted Portfolio Visualizer examples in this thread before, I was told that it wasn't possible to look at momentum the way you were defining it.

brooklynguy

  • Handlebar Stache
  • *****
  • Posts: 2098
  • Age: 37
Re: Dual Momentum Investing
« Reply #485 on: May 08, 2015, 10:36:27 AM »
So, thanks everyone!

Hear, hear!

Tensions in this thread may have risen and fallen faster than prices in the stock market we're discussing, but the return on investment for participants and readers alike have been terrific.

Thank you all!

forummm

  • Walrus Stache
  • *******
  • Posts: 7358
  • Senior Mustachian
Re: Dual Momentum Investing
« Reply #486 on: May 08, 2015, 10:38:44 AM »
So, thanks everyone!

Hear, hear!

Tensions in this thread may have risen and fallen faster than prices in the stock market we're discussing, but the return on investment for participants and readers alike have been terrific.

Thank you all!

Judging from my arbitrary post lookback period, I'm buying this thread. Does that mean I have to allocate 100% of my posts here?

sirdoug007

  • Pencil Stache
  • ****
  • Posts: 583
  • Age: 37
  • Location: Houston, TX
Re: Dual Momentum Investing
« Reply #487 on: May 08, 2015, 11:07:32 AM »
For some reason your spreadsheet contains no interpretable data for me.  (perhaps its an apple thing?)

But in the body of your response please include a full-ish analysis including standard deviation, max drawdown, sharp/sortino ratios, in the sharing of your findings.

A priori, I would not expect absolute momentum to outperform in any decade without a significant bear market.  The goal of AM is absolutely not to beat a 100% stock buy and hold portfolio in terms of CAGR.  It is to decrease the downside risk (period full stop.)

Put another way, on a risk/reward basis, what percentage of bonds would you have to hold in order to decrease S&P drawdowns to the level of an AM portfolio?  What would your total returns be with such a portfolio?

Relative momentum increases upside without diminishing drawdowns or volatility.  this is why DM is so attractive, increased upside, decreased downside.

Honestly I have no idea how to or inclination to figure out max drawdowns and sharpe/sortino ratios in excel.  Download the spreadsheet and you can do the math.

I see what you are saying about AM only helping in bear markets.  No bear market = no benefit.  And intuitively AM gets crushed in serious bull markets like the 20s, 50s, and 90s.

Looking at rolling 1 yr periods, the average return for AM is 5.67% and for the S&P500 it's 8.57%.  Standard deviation is 13.32% for AM and 19.59% for stocks.  Not sure how to go about figuring out an equivalent stock/bond mix for that return/standard deviation.

Attached is a chart of AM vs. stocks plotted since 7/1871.
« Last Edit: May 08, 2015, 11:10:54 AM by sirdoug007 »

hodedofome

  • Handlebar Stache
  • *****
  • Posts: 1107
  • Age: 37
  • Location: Texas
Re: Dual Momentum Investing
« Reply #488 on: May 08, 2015, 11:16:41 AM »
You should not expect to ever outperform buy and hold using trend following/momentum on only 1 market.  At best you'll lower volatility and drawdowns but the total return will most likely suffer. You need to use multiple, uncorrelated markets to really see a benefit on the return side.


Sent from my iPhone using Tapatalk

skyrefuge

  • Pencil Stache
  • ****
  • Posts: 999
  • Location: Suburban Chicago, IL
Re: Dual Momentum Investing
« Reply #489 on: May 08, 2015, 02:52:17 PM »
If you include the dividends in decision making, over the ~130 year time period (instead of starting over each decade), which wins?

Absolute Momentum: $90,254
Stocks Only: $1,247,873

And no, I didn't forget a digit in the AM number.

Ok, nevermind, the spreadsheet had a major error. When calculating the growth, dividends were being included for Stocks Only, but they were mistakenly being left out for AM. When I include dividends for AM as well:

Absolute Momentum: $3,350,259
Stocks Only: $1,247,873

When I look at only 1950-2014:

Absolute Momentum: $14,459
Stocks Only: $10,895

When I look at only 1950-2014, change the lookback method from "simple 6-month total return" to "10-month rolling average based on price index", start with $10k instead of $100, and switch to nominal values instead of real values (in an attempt to match the portfoliovizualizer.com methodology):

Absolute Momentum: $8,160,627.77 (compared to PV's $10,721,680)
Stocks Only: $10,602,478.35 (compared to PV's $10,468,979)

And, all-of-the-previous except using real values rather than nominal:

Absolute Momentum: $1,795,613.77
Stocks Only: $1,089,485.08

So, I'm still  not very confident that I have everything correct. It's nice to see that one of my values when using PV's method was pretty close, but then the AM value wasn't. And then I'm not sure why the "winner" would flip when switching from nominal to real values.

I think the upshot is: spreadsheets Я hard!

brooklynguy

  • Handlebar Stache
  • *****
  • Posts: 2098
  • Age: 37
Re: Dual Momentum Investing
« Reply #490 on: May 08, 2015, 03:07:55 PM »
Ok, nevermind, the spreadsheet had a major error.

This makes a lot more sense - given that the AM strategy is the product of a data mining quest for a (historically) high-performing strategy, the data had better say that it actually worked well in the past!

Quote
And then I'm not sure why the "winner" would flip when switching from nominal to real values.

I had the same question when I saw your results before I got to this paragraph, and I don't think there is any possible explanation for how they could be flipped, so there must still be an error somewhere.

milesdividendmd

  • Handlebar Stache
  • *****
  • Posts: 1913
  • Location: Portlandia
    • Miles Dividend MD
Re: Dual Momentum Investing
« Reply #491 on: May 08, 2015, 03:15:38 PM »
If you include the dividends in decision making, over the ~130 year time period (instead of starting over each decade), which wins?

Absolute Momentum: $90,254
Stocks Only: $1,247,873

And no, I didn't forget a digit in the AM number.

Ok, nevermind, the spreadsheet had a major error. When calculating the growth, dividends were being included for Stocks Only, but they were mistakenly being left out for AM. When I include dividends for AM as well:

Absolute Momentum: $3,350,259
Stocks Only: $1,247,873

When I look at only 1950-2014:

Absolute Momentum: $14,459
Stocks Only: $10,895

When I look at only 1950-2014, change the lookback method from "simple 6-month total return" to "10-month rolling average based on price index", start with $10k instead of $100, and switch to nominal values instead of real values (in an attempt to match the portfoliovizualizer.com methodology):

Absolute Momentum: $8,160,627.77 (compared to PV's $10,721,680)
Stocks Only: $10,602,478.35 (compared to PV's $10,468,979)

And, all-of-the-previous except using real values rather than nominal:

Absolute Momentum: $1,795,613.77
Stocks Only: $1,089,485.08

So, I'm still  not very confident that I have everything correct. It's nice to see that one of my values when using PV's method was pretty close, but then the AM value wasn't. And then I'm not sure why the "winner" would flip when switching from nominal to real values.

I think the upshot is: spreadsheets Я hard!

Score one for the biases of data based faith!
I can teach you to play the miles game in 30 days.  http://www.travelmiles101.com/travel-rewards-course-registration.  (A free course by mustachians for mustachians)

milesdividendmd

  • Handlebar Stache
  • *****
  • Posts: 1913
  • Location: Portlandia
    • Miles Dividend MD
Re: Dual Momentum Investing
« Reply #492 on: May 08, 2015, 03:23:57 PM »
If you include the dividends in decision making, over the ~130 year time period (instead of starting over each decade), which wins?

Absolute Momentum: $90,254
Stocks Only: $1,247,873

And no, I didn't forget a digit in the AM number.

Please share your work.  It's impossible to understand your results if you just throw up numbers like this. 

I'm extremely dubious of your claim based on total returns data going back to 1950, using 10 month MA:

https://www.portfoliovisualizer.com/test-market-timing-model?s=y&outOfMarketAssetType=1&timingPeriods[0]=5&timingWeights[0]=100&volatilityWindowSizeInDays=0&timingWeights[2]=0&riskControl=false&endYear=2014&volatilityPeriodWeight=0&movingAverageType=1&windowSize=10&riskWindowSizeInDays=0&timingUnits[0]=2&timingUnits[2]=2&timingModel=1&volatilityWindowSize=0&startYear=1950&assetsToHold=1&timingWeights[1]=0&windowSizeInDays=105&s=y&volatilityPeriodUnit=1&multipleTimingPeriods=false&timingUnits[1]=2&riskWindowSize=10&rebalancePeriod=1

I suspect your calculations are seriously flawed, (but have no way of checking your work.)

Absolute momentum should have similar returns (sometimes better, sometimes worse depending on the decade) with markedly diminished drawdowns, when using total returns both for lookback and for returns calculation.  AKA similar CAGR with a far superior Sharp/Sortino.

When I posted Portfolio Visualizer examples in this thread before, I was told that it wasn't possible to look at momentum the way you were defining it.

Forummm, for the last time, do you really not understand the prior criticism of your "testing" of DM?

It had nothing to do with using portfolio visualizer and everything to do with the fact that you were not in fact testing dual momentum.  You were backtesting index fund relative momentum.

Also note that the above PV example is a backtest of a 10 month moving average approach to the S&P 500, not dual momentum.

Details matter, and the prior criticisms were in no way personal (and in every way accurate.)
I can teach you to play the miles game in 30 days.  http://www.travelmiles101.com/travel-rewards-course-registration.  (A free course by mustachians for mustachians)

sirdoug007

  • Pencil Stache
  • ****
  • Posts: 583
  • Age: 37
  • Location: Houston, TX
Re: Dual Momentum Investing
« Reply #493 on: May 08, 2015, 03:25:31 PM »
I think the upshot is: spreadsheets Я hard!

Thanks for checking the numbers and finding this.  Damn $ in the wrong spot!

This is obviously a simplistic version of the strategy we are talking about but it is interesting to see it working even with only absolute momentum.

Attached is the corrected chart and spreadsheet.

forummm

  • Walrus Stache
  • *******
  • Posts: 7358
  • Senior Mustachian
Re: Dual Momentum Investing
« Reply #494 on: May 08, 2015, 06:20:06 PM »
If you include the dividends in decision making, over the ~130 year time period (instead of starting over each decade), which wins?

Absolute Momentum: $90,254
Stocks Only: $1,247,873

And no, I didn't forget a digit in the AM number.

Please share your work.  It's impossible to understand your results if you just throw up numbers like this. 

I'm extremely dubious of your claim based on total returns data going back to 1950, using 10 month MA:

https://www.portfoliovisualizer.com/test-market-timing-model?s=y&outOfMarketAssetType=1&timingPeriods[0]=5&timingWeights[0]=100&volatilityWindowSizeInDays=0&timingWeights[2]=0&riskControl=false&endYear=2014&volatilityPeriodWeight=0&movingAverageType=1&windowSize=10&riskWindowSizeInDays=0&timingUnits[0]=2&timingUnits[2]=2&timingModel=1&volatilityWindowSize=0&startYear=1950&assetsToHold=1&timingWeights[1]=0&windowSizeInDays=105&s=y&volatilityPeriodUnit=1&multipleTimingPeriods=false&timingUnits[1]=2&riskWindowSize=10&rebalancePeriod=1

I suspect your calculations are seriously flawed, (but have no way of checking your work.)

Absolute momentum should have similar returns (sometimes better, sometimes worse depending on the decade) with markedly diminished drawdowns, when using total returns both for lookback and for returns calculation.  AKA similar CAGR with a far superior Sharp/Sortino.

When I posted Portfolio Visualizer examples in this thread before, I was told that it wasn't possible to look at momentum the way you were defining it.

Forummm, for the last time, do you really not understand the prior criticism of your "testing" of DM?

It had nothing to do with using portfolio visualizer and everything to do with the fact that you were not in fact testing dual momentum.  You were backtesting index fund relative momentum.

Also note that the above PV example is a backtest of a 10 month moving average approach to the S&P 500, not dual momentum.

Details matter, and the prior criticisms were in no way personal (and in every way accurate.)

No, someone asked 'is it possible to use PV to backtest this' and the response was no. Not about my specific attempts. Not a big deal.

hodedofome

  • Handlebar Stache
  • *****
  • Posts: 1107
  • Age: 37
  • Location: Texas
Re: Dual Momentum Investing
« Reply #495 on: May 13, 2015, 05:05:30 PM »
Posting this here because it's a good summary if the advantages/disadvantages of momentum investing http://abnormalreturns.com/2015/05/11/qa-with-wes-gray-on-value-and-momentum-part-one/


Sent from my iPhone using Tapatalk

forummm

  • Walrus Stache
  • *******
  • Posts: 7358
  • Senior Mustachian
Re: Dual Momentum Investing
« Reply #496 on: May 14, 2015, 07:59:19 PM »
I think the upshot is: spreadsheets Я hard!

Thanks for checking the numbers and finding this.  Damn $ in the wrong spot!

This is obviously a simplistic version of the strategy we are talking about but it is interesting to see it working even with only absolute momentum.

Attached is the corrected chart and spreadsheet.

Note: This spreadsheet is showing calculations in real terms. The results may have been a little different if using nominal terms since, other than in a deflationary situation, you would have stayed in stocks longer. Since most people are not adjusting today's recent returns for inflation, they are unlikely to do any market timing based on real returns.

With this spreadsheet, the maximum drawdowns are 50.9% (AM 1933) and 76.8% (B&H 1932). The Stdevs are 14% (AM) and 18% (B&H).

hodedofome

  • Handlebar Stache
  • *****
  • Posts: 1107
  • Age: 37
  • Location: Texas
Re: Dual Momentum Investing
« Reply #497 on: May 14, 2015, 09:34:43 PM »
No, someone asked 'is it possible to use PV to backtest this' and the response was no. Not about my specific attempts. Not a big deal.

It's because Portfolio Visualizer uses a moving average to determine the trend of the asset and not past returns. DM as defined by Gary (we're talking about his implementation of DM in this thread, but you are free to use whatever you want to determine the trend) uses past returns to determine if an asset is going up or down. So PV would need to give you this option in order for it to be accurate.

Monkey Uncle

  • Pencil Stache
  • ****
  • Posts: 952
  • Location: West-by-god-Virginia
Re: Dual Momentum Investing
« Reply #498 on: May 15, 2015, 04:25:59 AM »
No, someone asked 'is it possible to use PV to backtest this' and the response was no. Not about my specific attempts. Not a big deal.

It's because Portfolio Visualizer uses a moving average to determine the trend of the asset and not past returns. DM as defined by Gary (we're talking about his implementation of DM in this thread, but you are free to use whatever you want to determine the trend) uses past returns to determine if an asset is going up or down. So PV would need to give you this option in order for it to be accurate.

PV's market timing module includes a relative strength model that lets you "Invest in best performing assets out of a ranked list of assets based on recent relative performance."  You can use a moving average as a risk control, but you don't have to select one.  Seems like this lets you replicate DM exactly, provided you include short term treasuries as one of your assets.  Am I missing something?
"Take this job and shove it" - David Allan Coe

hodedofome

  • Handlebar Stache
  • *****
  • Posts: 1107
  • Age: 37
  • Location: Texas
Re: Dual Momentum Investing
« Reply #499 on: May 15, 2015, 08:19:57 AM »
PV's market timing module includes a relative strength model that lets you "Invest in best performing assets out of a ranked list of assets based on recent relative performance."  You can use a moving average as a risk control, but you don't have to select one.  Seems like this lets you replicate DM exactly, provided you include short term treasuries as one of your assets.  Am I missing something?

It's kissing cousins but using past returns (absolute momentum) vs a moving average will be a little different in their signals.

For 2003-2015, on ETFReplay I put in VTSMX (US Stocks) and VGTSX (Int'l Stocks) using 6 month returns for relative strength. For the cash asset I use VBMFX (Bonds). If I just put those 3 assets into 1 basket and use relative strength to pick the top asset each month, I get:

CAGR: 14.1%
Volatility: 12.5%
Max DD: -16.6%
Sharpe Ratio: 0.73

vs a 100% VTSMX of:

CAGR: 10.2%
Volatility: 19.7%
Max DD: -55.4%
Sharpe Ratio: 0.34

Using a moving average of 6 months and switching to VBMFX if VTSMX and VGTSX are below their 6 month moving average I get:

CAGR: 16.3%
Volatility: 13.1%
Max DD: -16.6%
Sharpe Ratio: 0.83

Using absolute momentum and switching to VBMFX if VTSMX and VGTSX are below their 6 month returns I get:

CAGR: 14.1%
Volatility: 12.5%
Max DD: -16.6%
Sharpe Ratio: 0.73

It just so happened that in this time period the performance was about the same. However, I've seen noticeable differences in other tests with other asset classes and other time periods. The idea is the same though. Pick the best performing asset classes that are actually in an uptrend. It's up to you how you define what is an uptrend and what is performing the best.
« Last Edit: May 15, 2015, 08:25:34 AM by hodedofome »