A look at the research and evidence on time-series MOM

Discuss all general (i.e. non-personal) investing questions and issues, investing news, and theory.
Topic Author
larryswedroe
Posts: 16022
Joined: Thu Feb 22, 2007 7:28 am
Location: St Louis MO

Re: A look at the research and evidence on time-series MOM

Post by larryswedroe »

Vesalius
My pleasure ,You too.
Random Walker
Posts: 5561
Joined: Fri Feb 23, 2007 7:21 pm

Re: A look at the research and evidence on time-series MOM

Post by Random Walker »

Hi Larry,
Can you straighten me out on this? Which is a better diversifier for an equity rich, value oriented portfolio, TS Momentum or CS Momentum? And why? I think I read a post of yours earlier that said CS Momentum negatively correlated with value, while TS Momentum most likely uncorrelated. Can you explain why we should expect different behavior from the different types of momentum? Also, would you say that TS and CS are "different definitions" of momentum, thus fulfilling your robustness criteria? Thanks,

Dave
Angst
Posts: 2968
Joined: Sat Jun 09, 2007 11:31 am

Re: A look at the research and evidence on time-series MOM

Post by Angst »

Thank you to both lack_ey and Larry for the very helpful responses above, on page 2 - lack_ey's here and Larry's following.

I have a much better handle now on why Momentum, TS in particular, has not mirrored the availability of small cap and value funds in the past, and probably won't anytime soon. I kick myself to some degree because I already know many of the points which you made, but late at night I sometimes forget these details and get diverted by the occasional pro-momentum hyperbole in posts made by people other than yourselves. And it's amazing sometimes how much my perspective can improve with a cup of coffee and in the light of morning, and especially with two new thoughtful and detailed posts by lack_ey and Larry to read through. Today I'm thankful for the likes of both of you! :happy What a forum - I really do appreciate your responses.
alphabeta01
Posts: 14
Joined: Tue Oct 18, 2016 10:07 am

Re: A look at the research and evidence on time-series MOM

Post by alphabeta01 »

-----
Last edited by alphabeta01 on Thu Sep 30, 2021 12:21 pm, edited 1 time in total.
lack_ey
Posts: 6701
Joined: Wed Nov 19, 2014 10:55 pm

Re: A look at the research and evidence on time-series MOM

Post by lack_ey »

Random Walker wrote:Hi Larry,
Can you straighten me out on this? Which is a better diversifier for an equity rich, value oriented portfolio, TS Momentum or CS Momentum? And why? I think I read a post of yours earlier that said CS Momentum negatively correlated with value, while TS Momentum most likely uncorrelated. Can you explain why we should expect different behavior from the different types of momentum? Also, would you say that TS and CS are "different definitions" of momentum, thus fulfilling your robustness criteria? Thanks,

Dave
I certainly don't have Larry's perspective here and don't want to answer, but generally I think you should be careful about clarifying which investment universe you're looking at and the implementation when talking about momentum. Time-series momentum traditionally brings to mind managed futures funds, CTAs, etc., which generally trade futures contracts (particularly for commodities, though more generally not limited to that) long and short. For example you may look at a AQR Managed Futures Strategy (AQMIX). But when people talk about cross-sectional momentum, they're frequently asking about funds like iShares Edge MSCI USA Momentum Factor (MTUM), which is a long-only US equity fund. Also be careful distinguishing between a factor and funds that may be exposed to a factor (along with others).

Generally long-only CS momentum in equities has different properties from CS momentum in equities, which is different from but correlated to CS momentum in other asset classes, which is related to but not the same as TS momentum.

In any case, both TS and CS momentum are covered in the factor investing book.

alphabeta01 wrote:AQR TS MOM fund QMHIX has extreme negative (0.91) Size but decent 0.19 value and 0.56 momentum load. If one is trying to maintain overall portfolio load of 0.2 size and 0.4 value and close to zero momentum, do we have to tilt more aggressively to Small Value? or Should we exclude this asset in overall portfolio load calculation?
This kind of relates in a way to the above response in terms of being careful with the investment universes and definitions: what are you actually regressing on? For example, don't expect to regress a time-series momentum fund across asset classes on US or global equity factors and get particularly meaningful results. What's R^2 look like, confidence intervals? Are those figures even significant?


Anyway, Happy Thanksgiving, all. I'm off to do some Thanksgiving things.
User avatar
Taylor Larimore
Posts: 32839
Joined: Tue Feb 27, 2007 7:09 pm
Location: Miami FL

Momentum

Post by Taylor Larimore »

alphabeta01 wrote:AQR TS MOM fund QMHIX has extreme negative (0.91) Size but decent 0.19 value and 0.56 momentum load. If one is trying to maintain overall portfolio load of 0.2 size and 0.4 value and close to zero momentum, do we have to tilt more aggressively to Small Value? or Should we exclude this asset in overall portfolio load calculation?
alphabeta01:

Be careful what you wish for: QMHIX
The enemy of a good plan is the dream of a perfect plan. -- Jack Bogle
Best wishes.
Taylor
"Simplicity is the master key to financial success." -- Jack Bogle
alphabeta01
Posts: 14
Joined: Tue Oct 18, 2016 10:07 am

Re: A look at the research and evidence on time-series MOM

Post by alphabeta01 »

-----
Last edited by alphabeta01 on Thu Sep 30, 2021 12:21 pm, edited 1 time in total.
Topic Author
larryswedroe
Posts: 16022
Joined: Thu Feb 22, 2007 7:28 am
Location: St Louis MO

Re: A look at the research and evidence on time-series MOM

Post by larryswedroe »

Dave
CS momentum is negatively correlated, with pretty simple explanation. How do you get to be value, have relatively poor recent performance. How do you get to be growth, have relatively good recent performance.
TS momentum you are in or out of equities, so changes your asset allocation.
You can do both across asset classes as there is evidence they work across asset classes, with TS momentum providing tail risk as it moves out of equities in persistent bear markets
Larry
Random Walker
Posts: 5561
Joined: Fri Feb 23, 2007 7:21 pm

Re: A look at the research and evidence on time-series MOM

Post by Random Walker »

I don't understand how TS Momentum results in shifting asset allocation. I thought long stocks going up and short stocks going down would eliminate stock risk and isolate TS Momentum premium. The long and short sides aren't equal? Thanks,

Dave
Angst
Posts: 2968
Joined: Sat Jun 09, 2007 11:31 am

Re: A look at the research and evidence on time-series MOM

Post by Angst »

Random Walker wrote:I don't understand how TS Momentum results in shifting asset allocation. I thought long stocks going up and short stocks going down would eliminate stock risk and isolate TS Momentum premium. The long and short sides aren't equal? Thanks,

Dave
Dave, I'm not sure if this is exactly right, but I'm guessing it has to do with TS moving between different asset classes besides equity, i.e. commodities, bonds, currencies... and I gather that longs and shorts in any particular asset class do not necessarily add up to zero with TS mom, and when equity is crashing, TS might not hold any long equity positions. Caveat - I'm still trying to get the basics of CS vs. TS mom straight in my mind.

Back in September, Larry posted this summary which I've gone back to a couple times for help. Actually the whole thread might be worth rereading for myself:
larryswedroe, in [url=https://www.bogleheads.org/forum/viewtopic.php?f=10&t=197606&newpost=3019797#p3019305]this earlier thread[/url] wrote:just to straighten it out

Cross Sectional MOM is RELATIVE momentum, going long the best performing (typically top 30%) and short the worst performing 30% in the asset class. If all going down, then it still is long the securities that have gone down the least and if all going up still going short those that are going up the least.
Times series MOM is trend following and is ABSOLUTE MOM, going long what has been going up and short what has been going down

MOM is SHORT TERM positive, positive MOM based on last 12 months (excluding the most recent month) ON AVERAGE tend to keep being positive MOM for short period, about 5 months.

We set relatively wide bands to allow for momentum to work in favor, but still want to control risks within reasonable limits, hence the need for bands.

Hope that helps
Larry
lack_ey
Posts: 6701
Joined: Wed Nov 19, 2014 10:55 pm

Re: A look at the research and evidence on time-series MOM

Post by lack_ey »

Random Walker wrote:I don't understand how TS Momentum results in shifting asset allocation. I thought long stocks going up and short stocks going down would eliminate stock risk and isolate TS Momentum premium. The long and short sides aren't equal? Thanks,

Dave
No, TS momentum is generally not defined to be market neutral in every asset class used (at any given point in time). In fact, most or at least many funds don't operate on as granular a level as individual stocks. It may short a stock market on a down trend and go long a stock market on an up trend. In other words, it's taking directional bets on markets (or whatever is considered in its investment universe), hoping for each trend to continue.

Anyway, be careful to check any given fund's investments and strategy.

Some have brought up AQR's fund. As noted in an annual report:
We expect that the correlation of this strategy to the equity markets will average close to zero over a full economic cycle. That said, it is the nature of the strategy to exhibit a positive beta when markets have been rising and a negative beta when the markets have been falling.
Though note that in plenty of situations, some stock markets will be trending up while others are trending down, so it could be significantly long and short equities simultaneously. Doesn't have to be that way, though.
Topic Author
larryswedroe
Posts: 16022
Joined: Thu Feb 22, 2007 7:28 am
Location: St Louis MO

Re: A look at the research and evidence on time-series MOM

Post by larryswedroe »

Dave
TS MOM could be called a tactical asset allocation strategy, shifting allocations between asset classes depending on the TS MOM signal. If long only would be long assets with + TS MOM and avoid those with - TS MOM. If long short, would be going long assets with + and short those with -.
Larry
Random Walker
Posts: 5561
Joined: Fri Feb 23, 2007 7:21 pm

Re: A look at the research and evidence on time-series MOM

Post by Random Walker »

Larry,
Is the AQR TS Momentum fund long/short? If so, would it have no exposure to market beta? Or can it be long and short to different extents at different times, thus changing its equity exposure over time? For example, doesn't QSPIX have no net equity exposure and give pure exposure to value factor? Thanks,

Dave
Topic Author
larryswedroe
Posts: 16022
Joined: Thu Feb 22, 2007 7:28 am
Location: St Louis MO

Re: A look at the research and evidence on time-series MOM

Post by larryswedroe »

Dave
Yes it is long short but NOT in way you think. It goes long assets that are positive TS MOM and short those with negative. So it could either be + beta or - depending on what markets are doing. That is why TS MOM can provide tail hedging benefits, it gets short beta in prolonged down markets like in 2000-02 and 2008 through early 09.

Now CS MOM is always neutral as it is always long and short, long the 30% that is positive and short the 30% that is negative. And it's relative remember, not absolute as is TS MOM.

So CS is always long and short in same asset class, while TS can be long or short, not both

Larry
lack_ey
Posts: 6701
Joined: Wed Nov 19, 2014 10:55 pm

Re: A look at the research and evidence on time-series MOM

Post by lack_ey »

I don't know if it needs restating as I think all the information was already there a few posts back, but
  • Cross-sectional momentum has long/short positions within a market, analogous to how value, size, and other factors are defined, and as such these offsetting positions largely cancel out the beta exposure
  • Time-series momentum goes long markets that are trending up and shorts markets that are trending down, so the beta is deliberately positive or negative at any point depending on what's happening (albeit perhaps averaging out to something around zero over a business cycle)
CS momentum is a bet on relative performance of some assets within a group vs. others, that recent relative overperformers beat recent relative underperformers. TS momentum is a bet on direction, that the trend of a given asset or group continues.
User avatar
Taylor Larimore
Posts: 32839
Joined: Tue Feb 27, 2007 7:09 pm
Location: Miami FL

A Private Message (PM) About Momentum Funds

Post by Taylor Larimore »

Bogleheads:

This morning I received a Private Message about momentum funds. I believe its message is worth considering (personal comments removed):
I don't post much any more. Thankfully, I GOT the message many years ago and have been pleased, and thankful.

This AQR stuff being promoted is a bad thing, in my view. This is way different than the DFA vs Vanguard debates, or small vs. large.

There was a time when it was generally realized (especially by the advisors and other strong voices on the forum) that very expensive and hard to acquire funds that are opaque in their holdings and precise methods (long/short, momentum exotic combos, and lord knows what else,) were considered bad vehicles—by their very nature. In my view, they are still bad ideas for almost everyone.

These AQR funds may be run by smart people (as are ANY active funds) but I fear their popularity rise is due more to having something cute to offer clients--better mousetraps, under the disguise of "science/evidence". Of course, what the evidence shows and how a firm chooses to interpret can vary tremendously.

This in turn feeds advisory firms who then use their access to these funds as lures to additional clients, and this trickles down to the forum, where less knowledgeable and vulnerable investors read all the fancy print about them.

It matters not that their strategies work or don't (at present they don't). Their existence seems to run contrary to the ideas put forward by Bogle and supported by many good people.

I see this as a sad thing. Unfortunately, there seems to be an insufficient number of power voices, other than your own, to offer expert counter-commentary. Would be nice if other advisors or experts actually spoke up, as I cannot believe all are in lockstep with this.
Last edited by Taylor Larimore on Fri Nov 25, 2016 12:49 pm, edited 1 time in total.
"Simplicity is the master key to financial success." -- Jack Bogle
Random Walker
Posts: 5561
Joined: Fri Feb 23, 2007 7:21 pm

Re: A look at the research and evidence on time-series MOM

Post by Random Walker »

Lack_ey,
Thanks. Your explanation of TS Momentum just above clarified things for me greatly. So my next question is how often does the AQR fund reconstitute based on the prior months 2-12?

Dave
Trader/Investor
Posts: 264
Joined: Wed Dec 02, 2015 8:35 pm

Re: A Private Message (PM) About Momentum Funds

Post by Trader/Investor »

Taylor Larimore wrote:Bogleheads:

This morning I received a Private Message about momentum funds. I believe its message is worth considering (personal comments removed):
I don't post much any more. Thankfully, I GOT the message many years ago and have been pleased, and thankful.

This AQR stuff being promoted is a bad thing, in my view. This is way different than the DFA vs Vanguard debates, or small vs. large.

There was a time when it was generally realized (especially by the advisors and other strong voices on the forum) that very expensive and hard to acquire funds that are opaque in their holdings and precise methods (long/short, momentum exotic combos, and lord knows what else,) were considered bad vehicles—by their very nature. In my view, they are still bad ideas for almost everyone.

These AQR funds may be run by smart people (as are ANY active funds) but I fear their popularity rise is due more to having something cute to offer clients--better mousetraps, under the disguise of "science/evidence". Of course, what the evidence shows and how a firm chooses to interpret can vary tremendously.

This in turn feeds advisory firms who then use their access to these funds as lures to additional clients, and this trickles down to the forum, where less knowledgeable and vulnerable investors read all the fancy print about them.

It matters not that their strategies work or don't (at present they don't). Their existence seems to run contrary to the ideas put forward by Bogle and supported by many good people.

I see this as a sad thing. Unfortunately, there seems to be an insufficient number of power voices, other than your own, to offer expert counter-commentary. Would be nice if other advisors or experts actually spoke up, as I cannot believe all are in lockstep with this.

R.
That was not my private message but couldn't agree more and have been railing against all these faddish funds. Please don't be a sucker and lemming and drink the kool-aid. Momentum has been popularized in one form or another since the Jesse Livermore days and hardly a new concept. I have scores of books from the 60s, 70s and 80s on the topic Just seems some here (with no real money documented track record whatsoever) want to dress it up with new lipstick.
Random Walker
Posts: 5561
Joined: Fri Feb 23, 2007 7:21 pm

Re: A look at the research and evidence on time-series MOM

Post by Random Walker »

For those not "buying the kool aid", do you believe the data? Do you think the benefit will be arbitraged away? Do you not trust it because it is a behavioral story rather than a risk story? Do you look at it in isolation rather than as a component of a multi-factor portfolio? Just looks too good to be true? These are not rhetorical questions, I'm truly interested. The fact that it has been widely known for a long time yet persists, increases my interest in it.

Dave
vesalius
Posts: 794
Joined: Tue Jul 13, 2010 7:00 pm
Location: Texas

Re: A look at the research and evidence on time-series MOM

Post by vesalius »

Random Walker wrote:For those not "buying the kool aid", do you believe the data? Do you think the benefit will be arbitraged away? Do you not trust it because it is a behavioral story rather than a risk story? Do you look at it in isolation rather than as a component of a multi-factor portfolio? Just looks too good to be true? These are not rhetorical questions, I'm truly interested. The fact that it has been widely known for a long time yet persists, increases my interest in it.

Dave
I believe in multifactor based investing, but I'm having trouble with this single factor TS momentum because I don't understand it well enough to be comfortable. This distinction between TS and CS momentum is new to me personally and I never make asset allocation or investment changes based on brand new or short term info. I let that marinate for a good while before I do anything.

I am fine with and want my funds to incorporate momentum the way AQR does with there mutltifactor funds, but will take what DFA and Bridgeway do as well.
Last edited by vesalius on Fri Nov 25, 2016 2:14 pm, edited 1 time in total.
Trader/Investor
Posts: 264
Joined: Wed Dec 02, 2015 8:35 pm

Re: A look at the research and evidence on time-series MOM

Post by Trader/Investor »

Random Walker wrote:For those not "buying the kool aid", do you believe the data? Do you think the benefit will be arbitraged away? Do you not trust it because it is a behavioral story rather than a risk story? Do you look at it in isolation rather than as a component of a multi-factor portfolio? Just looks too good to be true? These are not rhetorical questions, I'm truly interested. The fact that it has been widely known for a long time yet persists, increases my interest in it.

Dave
Look, I am no expert, yet have traded a small four figure account into a seven figure account using momentum/trend. So I am hardly anti-momentum. Successful momentum strategies have been around for years. Just read the profiles of many of the traders in Jack Schwager's Market Wizard books. Real traders, real money and documented results!! If this is news to you, you must be real new to the game. What bothers me are academics with no real money documented results dressing up momentum in new lipstick as if is something novel and amazing.
Topic Author
larryswedroe
Posts: 16022
Joined: Thu Feb 22, 2007 7:28 am
Location: St Louis MO

Re: A look at the research and evidence on time-series MOM

Post by larryswedroe »

trader/investor
momentum in new lipstick as if is something novel and amazing.
I really don't understand this statement. As you note momentum has been well documented for decades.
And even Fama threw in the towel in 2003 and DFA began incorporating momentum into construction rules, so what AQR is doing isn't really different from DFA, at least in CS MOM (though AQR in some funds will go long short while DFA only avoids the short).

There is nothing new here, in fact Carhart incorporated MOM into the FF three factor model to make it a four factor model back in 1997. So academics have been using it now for about 20 years.

And I would note Asness wrote his dissertation, with Fama as his thesis advisor, on momentum in early 90s.

Dave
Here's short explanation, they in fact use three signals, not just one, in their MF fund (which seeks to reduce risks in extreme markets), there is short term, long term and overextended. They weight the three signals, so heavier weight when signals agree and lesser when they disagree and they size positions based on a short term volatility forecast

‒Covers 67 instruments across 4 major asset classes
‒If the excess return of an instrument has been positive over the past n months: take a long position*
‒If the excess return of an instrument has been negative over the past n months: take a short position*
‒n = 1, 3, and 12 months
‒Portfolio is scaled to target 10% annualized volatility
‒Rebalances monthly

the fund also equal weights the four asset classes of stocks, bonds, commodities and currencies

the fund also uses algo trading to keep trading costs down, seeking to sell, not buy liquidity

Hope that helps more than adds confusion (complexity)
Random Walker
Posts: 5561
Joined: Fri Feb 23, 2007 7:21 pm

Re: A look at the research and evidence on time-series MOM

Post by Random Walker »

Larry,
Scaled to target 10% volatility; that means leveraged to different extents depending on recent volatility? Thanks,

Dave
lack_ey
Posts: 6701
Joined: Wed Nov 19, 2014 10:55 pm

Re: A look at the research and evidence on time-series MOM

Post by lack_ey »

Random Walker wrote:Larry,
Scaled to target 10% volatility; that means leveraged to different extents depending on recent volatility? Thanks,

Dave
Not Larry but I know the answer.

Yes, to target a given volatility level you need to be able to estimate volatility. Obviously the realized volatility you get in practice will be different from the targeted or forecasted level. But any reasonable model will give you a different volatility forecast depending on what's recently been happening, in the very least (there could be other factors too). This means scaled leverage based on recent volatility.

Note that 10% here is the long-term or strategic target and they don't actually attempt to target 10% exactly over every given period, just something reasonably close to it. If trend signals are conflicting (e.g. 3-month direction is down while 12-month direction is up) then following the weightings, this results in a smaller position sizing. With smaller positions you get lower vol, and they don't rescale everything all the time frantically to try to keep to the strategic target. They have similar controls in the other long/short funds.


Taylor Larimore wrote:Bogleheads:

This morning I received a Private Message about momentum funds. I believe its message is worth considering (personal comments removed):
I don't post much any more. Thankfully, I GOT the message many years ago and have been pleased, and thankful.

This AQR stuff being promoted is a bad thing, in my view. This is way different than the DFA vs Vanguard debates, or small vs. large.

There was a time when it was generally realized (especially by the advisors and other strong voices on the forum) that very expensive and hard to acquire funds that are opaque in their holdings and precise methods (long/short, momentum exotic combos, and lord knows what else,) were considered bad vehicles—by their very nature. In my view, they are still bad ideas for almost everyone.

These AQR funds may be run by smart people (as are ANY active funds) but I fear their popularity rise is due more to having something cute to offer clients--better mousetraps, under the disguise of "science/evidence". Of course, what the evidence shows and how a firm chooses to interpret can vary tremendously.

This in turn feeds advisory firms who then use their access to these funds as lures to additional clients, and this trickles down to the forum, where less knowledgeable and vulnerable investors read all the fancy print about them.

It matters not that their strategies work or don't (at present they don't). Their existence seems to run contrary to the ideas put forward by Bogle and supported by many good people.

I see this as a sad thing. Unfortunately, there seems to be an insufficient number of power voices, other than your own, to offer expert counter-commentary. Would be nice if other advisors or experts actually spoke up, as I cannot believe all are in lockstep with this.
Personally I would be very interested to hear more counterarguments on the merits or lack thereof. Unfortunately, many of the reservations voiced in many threads if any are not presenting relevant evidence or any serious analysis. I think to change minds and offer a different perspective you are best served engaging the substance of the issues.

Now, there are some posters here who have "done the homework" and concluded this thing or the other is not for them (but sometimes out of more personal belief or satisfaction with an alternative and already established plan rather than a strong attack against the underlying arguments), so I'm not implying there is some kind of consensus among those in the know.

In any case whether you subscribe to a certain belief or not, I think it is worth understanding the arguments, so I disagree with whoever it is that sent the PM. That is encouraging the maintenance of a bubble around core Bogleheads ideas—seeking reinforcement of principles for the sake of emphasizing clarity of conviction rather than promoting inquiry and knowledge. It also seems a bit paternalistic and condescending to try to stand up against certain ideas in an attempt to keep these dangerous things away from the less knowledgeable and vulnerable. I understand the view and know the intentions are in the right place, but I don't support it.

For what it's worth, I think I can guess who sent the PM but will keep that to myself (and I'm probably wrong, anyway).
Random Walker
Posts: 5561
Joined: Fri Feb 23, 2007 7:21 pm

Re: A look at the research and evidence on time-series MOM

Post by Random Walker »

Lack_ey,
I thought that was the answer. Thanks for a great explanation.

Dave
stlutz
Posts: 5585
Joined: Fri Jan 02, 2009 12:08 am

Re: A look at the research and evidence on time-series MOM

Post by stlutz »

Personally I would be very interested to hear more counterarguments on the merits or lack thereof. Unfortunately, many of the reservations voiced in many threads if any are not presenting relevant evidence or any serious analysis. I think to change minds and offer a different perspective you are best served engaging the substance of the issues.
The counterarguments get made consistently in thread after thread. However, when the constraint is that one can only argue about the data is imposed--that is, you need to prove the time series momentum didn't work otherwise it's a great thing--well, then yes, any counterargument would not be "relevant" or "serious".

Let me do a little bit of a different type of analysis. Let's rerun some thread history on this forum. When I first joined, the only factor approaches that were discussed were small and value. Then momentum started getting more consideration. Then it was quality. Low vol. has been discussed. Now we've moved into various long/short strategies. (And then there have been other "trendy" concepts like commodities futures).

I'd like to see someone put together a backtest to see how the investor who followed all of these trends performed. Someone following Larry Swedroe's various advice over the years had probably made about 4 major changes in investment strategy. He says that you need to stick with a strategy for 20+ years for the benefits to really pay off, but you can't do that when a new strategy is promoted every couple of years.

Look--Larry will be fine with pretty much any investment strategy. Whether he is 100% T-bills, 100% smallcap value or 100% in some kind of AQR hedge fund, odds are well over 90% that he'll still be richer than me in 20 years.

Most people who visit this forum are mostly worried about having enough in their 401Ks or IRAs to retire around age 65. Some are trying to retire at 55. Regardless, the keys to either are saving a lot, taking an appropriate amount of risk, keeping costs low, and staying the course. Behavioral finance research backs up that these "factors" are the real keys to successful investing as opposed to picking the right stocks.

There's a lot of people in these theory threads who do work in finance or who have spent a lot of time studying it. I'm one of them--I enjoy reading and participating because they are interesting. Anybody using them for actual investing decisions is probably making significant portfolio changes every year or two as "new" research is promoted. That's not helpful for long-term investing success.

I'm neither rabidly for or against various tilting strategies. I believe in them a lot less than I used to. That's mainly because I've found the arguments for them in this forum to actually be rather unpersuasive. That said, I stick with my tilts because I believe in sticking with a sensible plan more than I believe in following what I think is right at the moment.

For those who do factor-weight, I think Robert T is the real "model investor" for the forum. His approach is very data-driven. He adopted a tilting strategy and has consistently stuck with it for many years. And he's not breathlessly posting about a new investment idea every year.
User avatar
Robert T
Posts: 2803
Joined: Tue Feb 27, 2007 8:40 pm
Location: 1, 0.2, 0.4, 0.5
Contact:

Re: A look at the research and evidence on time-series MOM

Post by Robert T »

.
Here are some things to consider:

1. The earlier backtest (2000-15) viewtopic.php?p=3047311#p3047311 showed very little benefit in Sharpe ratio from adding (time series based) managed futures when the allocation was taken from bonds. There was a more observable benefit (higher Sharpe ratio) when the allocation was taken from stocks but this both lowered returns and standard deviation. Personally would prefer to keep my same expected return (rather than lower it).

2. Here’s a recent analysis showing that the results of the Moskowitz et al paper on time series momentum (they coined the term) http://docs.lhpedersen.com/TimeSeriesMomentum.pdf was driven mainly by volatility scaling than by time series momentum http://www.sciencedirect.com/science/ar ... 8116301379

3. I recall one of the longest threads on Bogleheads was on 200 day moving average which is a form of time series or longitudinal momentum (shifting allocations over time based on time series momentum). I raised some doubts about it in that thread “...back-tested results are sensitive to asset class used, time period, duration of moving averages, buy and sell bands around the MA, and taxes…”

4. Time series momentum tends to provide better results for growth (and market) portfolios than value portfolios. This may be due to the positive correlation with cross-sectional momentum. i.e. series with positive loads to UMD (cross-sectional momentum) perform better in time series momentum than series with negative loads to UMD.

If you simply go long when past 12 months returns are positive and to t-bills when past 12 month returns are negative reviewed on a monthly basis:

6/1927–10/2016: Annualized return (%) /annualized SD
  • FF LG static = 9.3 / 18.4
    FF LG with “momentum” timing = 9.8 / 13.4

    FF SG static = 8.5 / 26.1
    FF SG with “momentum” timing = 9.3 / 18.6
For the growth portfolios the simulated returns from “momentum” timing were higher than the static portfolios
  • FF LV static = 11.7 / 24.8
    FF LV with “momentum timing” = 10.8 / 17.0

    FF SV static = 14.5 / 28.3
    FF SV with “momentum timing” = 12.9 / 20.7
For the value portfolios the simulated returns from “momentum” timing were lower than the static portfolios. While the SD was also lower, as similar return/SD profile could be achieved by simply using a fixed value stock:bond allocation.

5. Here’s an extract from Bernstein’s 2000 book – The Intelligent Asset Allocator. “As much as it pains me to admit it, momentum exists.” … “How does this data affect the average investor? Only at the margins. Lest we get too carried away, the most impressive autocorrelations we’ve encountered are in the 0.2 range. That means that no more than 4% (0.2 squared, or R-squared) of tomorrow’s price change can be explained by today’s.” His practical suggest is ….”rebalance no more than once per year”.

We each have to make our own decisions on how we allocate our assets. To me, 'risk based' allocations (e.g. to market, size, value, term) are more embedded in fundamentals of the underlying assets and, by association, provide more confidence in expected return (at least to me) than other 'factors'. Would just note that adding a small allocation to cross-sectional momentum can help offset negative momentum loads of value funds and thereby reduce the negative momentum load on an overall value tilted portfolio (i.e. get it closer to zero).

Robert
.
Angst
Posts: 2968
Joined: Sat Jun 09, 2007 11:31 am

Re: A look at the research and evidence on time-series MOM

Post by Angst »

^Thankyou Robert T for posting all your considerations listed in your previous post above. I really value and get a lot out of your posts.
Robert T wrote: [Snip...] Would just note that adding a small allocation to cross-sectional momentum can help offset negative momentum loads of value funds and thereby reduce the negative momentum load on an overall value tilted portfolio (i.e. get it closer to zero).
One question, if you can comment, particularly in regard to your final statement, quoted above. I know you've looked at MTUM before: Would you consider it a reasonable damper to negative (value derived) momentum if one's value is all small cap? I.e, since MTUM is large cap, does that reduce it's effectiveness or desirability in this example?
User avatar
tarheel
Posts: 452
Joined: Fri Sep 17, 2010 8:33 am

Re: A Private Message (PM) About Momentum Funds

Post by tarheel »

Taylor Larimore wrote:Bogleheads:

This morning I received a Private Message about momentum funds. I believe its message is worth considering (personal comments removed):
I don't post much any more. Thankfully, I GOT the message many years ago and have been pleased, and thankful.

This AQR stuff being promoted is a bad thing, in my view. This is way different than the DFA vs Vanguard debates, or small vs. large.

There was a time when it was generally realized (especially by the advisors and other strong voices on the forum) that very expensive and hard to acquire funds that are opaque in their holdings and precise methods (long/short, momentum exotic combos, and lord knows what else,) were considered bad vehicles—by their very nature. In my view, they are still bad ideas for almost everyone.

These AQR funds may be run by smart people (as are ANY active funds) but I fear their popularity rise is due more to having something cute to offer clients--better mousetraps, under the disguise of "science/evidence". Of course, what the evidence shows and how a firm chooses to interpret can vary tremendously.

This in turn feeds advisory firms who then use their access to these funds as lures to additional clients, and this trickles down to the forum, where less knowledgeable and vulnerable investors read all the fancy print about them.

It matters not that their strategies work or don't (at present they don't). Their existence seems to run contrary to the ideas put forward by Bogle and supported by many good people.

I see this as a sad thing. Unfortunately, there seems to be an insufficient number of power voices, other than your own, to offer expert counter-commentary. Would be nice if other advisors or experts actually spoke up, as I cannot believe all are in lockstep with this.
Part of the reason I love the forums are the multitude of voices on both sides of every issue. But let's be honest. The people reading this thread are the same 10 or so factor investing dorks such as myself (I could virtually name all 10 people). We are well versed in the science of investing, and invest in funds like QMHIX in <10% of our portfolios. We love discussing these topics.

"Mom and pop" investors are not reading these threads. I think Lady Geek from time to time posts on threads like these and says "this not for your average Boglehead". I think that's a good idea - maybe we could even have a new thread category called "advanced theory" or something like that.

But with all due respect, it's not helpful when people put opinion-based warnings up like "beware of this mousetrap" or "don't invest in anything complicated". If you have evidence supporting your opinions, we'd love to hear them, as always. Those of us that do invest in such funds only do so as a consequence of the science behind the funds in a very quantitative manner. As Robert T would say, of course no guarantees.
User avatar
siamond
Posts: 6003
Joined: Mon May 28, 2012 5:50 am

Re: A look at the research and evidence on time-series MOM

Post by siamond »

stlutz wrote:Let me do a little bit of a different type of analysis. Let's rerun some thread history on this forum. When I first joined, the only factor approaches that were discussed were small and value. Then momentum started getting more consideration. Then it was quality. Low vol. has been discussed. Now we've moved into various long/short strategies. (And then there have been other "trendy" concepts like commodities futures). [...]

There's a lot of people in these theory threads who do work in finance or who have spent a lot of time studying it. I'm one of them--I enjoy reading and participating because they are interesting. Anybody using them for actual investing decisions is probably making significant portfolio changes every year or two as "new" research is promoted. That's not helpful for long-term investing success.

I'm neither rabidly for or against various tilting strategies. I believe in them a lot less than I used to. That's mainly because I've found the arguments for them in this forum to actually be rather unpersuasive. That said, I stick with my tilts because I believe in sticking with a sensible plan more than I believe in following what I think is right at the moment.
This was a GREAT post, stlutz. Intellectual curiosity is a good thing, and yet sticking with a sensible plan has to override it. Totally agreed.
User avatar
Robert T
Posts: 2803
Joined: Tue Feb 27, 2007 8:40 pm
Location: 1, 0.2, 0.4, 0.5
Contact:

Re: A look at the research and evidence on time-series MOM

Post by Robert T »

Angst wrote:One question, if you can comment, particularly in regard to your final statement, quoted above. I know you've looked at MTUM before: Would you consider it a reasonable damper to negative (value derived) momentum if one's value is all small cap? I.e, since MTUM is large cap, does that reduce it's effectiveness or desirability in this example?
It depends on the desired portfolio factor load targets. If in the 0.4 size and value range, this can still be achieved with the addition of MSCI USA Momentum (the index tracked by MTUM) - if characteristics of historical simulations of indicative of the future characteristics.

Here are two examples using data from 8/1996 to 10/2016.

(i) Russell Fundamental US Small Cap Index had a -0.09 momentum load. This can be eliminated by adding 25% MSCI USA Momentum. As the MSCI index had a -0.17 size load and -0.03 value load its addition results in an overall portfolio with a lower size and value load than Russell Fundamental US Small Cap (the latter had about a 0.5/0.6 size/value load). The resulting factor loads on the 75:25 Russell Fundamental US Small Cap:MSCI USA Momentum allocation are Mkt = 0.99, size = 0.34, value = 0.42, momentum = 0.00. i.e. achieving about a 0.4 portfolio value load with zero net momentum load. The lower value load is obviously no problem for those targeting a 0.4 value load.

(ii) S&P600 Pure Value Index had a -0.30 momentum load. You would need about a 50% MSCI USA Momentum allocation to get to a net zero portfolio momentum load. The resulting factor loads on the 50:50 (49:51 to be exact) S&P600 Pure Value Index:MSCI USA Momentum are Mkt = 1.02, size = 0.35, value = 0.51, momentum = 0.00. i.e. achieving about a 0.5 portfolio value load with zero net momentum load.

The lower the momentum load on value funds, the smaller the momentum allocation needed to offset this. Interestingly the historical negative momentum load on S&P600 Value (from 8/1996 to 10/2015) has been very small at -0.01, and has been comparable to DFA funds after they introduced their momentum screens in 2003. In this case, not really necessary to add a separate allocation to momentum to get to a momentum neutral portfolio (if all you are using is S&P600 value for your value tilt).

Personally, among small value funds, I like S&P600 Value (tracked by IJS) and Russell Fundamental US Small Cap (tracked by FNDA).

Obviously no guarantees.

Robert
.
User avatar
sperry8
Posts: 3065
Joined: Sat Mar 29, 2008 9:25 pm
Location: Miami FL

Re: A look at the research and evidence on time-series MOM

Post by sperry8 »

Robert T wrote:Would just note that adding a small allocation to cross-sectional momentum can help offset negative momentum loads of value funds and thereby reduce the negative momentum load on an overall value tilted portfolio (i.e. get it closer to zero).

Robert
.
When you say small allocation, what % would you suggest? And is the % you quote a % of the value tilted portion of my portfolio or the total equity portion? Note, I own VBR ,VOE & VTV for my value tilt.
BH Contests: 23 #89 of 607 | 22 #512 of 674 | 21 #66 of 636 |20 #253/664 |19 #233/645 |18 #150/493 |17 #516/647 |16 #121/610 |15 #18/552 |14 #225/503 |13 #383/433 |12 #366/410 |11 #113/369 |10 #53/282
User avatar
matjen
Posts: 2189
Joined: Sun Nov 20, 2011 10:30 pm

Re: A Private Message (PM) About Momentum Funds

Post by matjen »

tarheel wrote: But let's be honest. The people reading this thread are the same 10 or so factor investing dorks such as myself (I could virtually name all 10 people).
Who are you calling a dork!?! :D

Yay Vanguard & Bogle
Yay DFA & Fama/French
Yay AQR & Asness/Ilmanen

YAY Open mindedness, variety of perspective and open discussion!!! The marketplace of ideas.
A man is rich in proportion to the number of things he can afford to let alone.
User avatar
tarheel
Posts: 452
Joined: Fri Sep 17, 2010 8:33 am

Re: A Private Message (PM) About Momentum Funds

Post by tarheel »

matjen wrote:
tarheel wrote: But let's be honest. The people reading this thread are the same 10 or so factor investing dorks such as myself (I could virtually name all 10 people).
Who are you calling a dork!?! :D

Yay Vanguard & Bogle
Yay DFA & Fama/French
Yay AQR & Asness/Ilmanen

YAY Open mindedness, variety of perspective and open discussion!!! The marketplace of ideas.
You are indeed one of the 10, matjen :)
User avatar
grap0013
Posts: 1892
Joined: Thu Mar 18, 2010 1:24 pm

Re: A Private Message (PM) About Momentum Funds

Post by grap0013 »

Taylor Larimore wrote:Bogleheads:

This morning I received a Private Message about momentum funds. I believe its message is worth considering (personal comments removed):
I don't post much any more. Thankfully, I GOT the message many years ago and have been pleased, and thankful.

This AQR stuff being promoted is a bad thing, in my view. This is way different than the DFA vs Vanguard debates, or small vs. large.

There was a time when it was generally realized (especially by the advisors and other strong voices on the forum) that very expensive and hard to acquire funds that are opaque in their holdings and precise methods (long/short, momentum exotic combos, and lord knows what else,) were considered bad vehicles—by their very nature. In my view, they are still bad ideas for almost everyone.

These AQR funds may be run by smart people (as are ANY active funds) but I fear their popularity rise is due more to having something cute to offer clients--better mousetraps, under the disguise of "science/evidence". Of course, what the evidence shows and how a firm chooses to interpret can vary tremendously.

This in turn feeds advisory firms who then use their access to these funds as lures to additional clients, and this trickles down to the forum, where less knowledgeable and vulnerable investors read all the fancy print about them.

It matters not that their strategies work or don't (at present they don't). Their existence seems to run contrary to the ideas put forward by Bogle and supported by many good people.

I see this as a sad thing. Unfortunately, there seems to be an insufficient number of power voices, other than your own, to offer expert counter-commentary. Would be nice if other advisors or experts actually spoke up, as I cannot believe all are in lockstep with this.
With all due respect, this is like thinking the world is flat and trying to prevent others from thinking it could potentially be round. Theory and knowledge are in a continuous state of development. New ideas should be welcome here. Indexing was once an innovative idea as well. I don't see any new posters with 80% QMHIX and 20% Total Stock Market so I don't see the damage to the average investor who visits this website. Furthermore, we are all adults and responsible for our own actions. Every book in the library should be available to read even if everyone does not agree with it. Bogleheads should not promote censorship as long as no forum rules are violated. Skip the thread if you do not like it. I skip TSM threads and they do not bother me. I respect others approaches. Not trying to be harsh but I feel very strongly these types of threads do not violate any rules.
There are no guarantees, only probabilities.
User avatar
grap0013
Posts: 1892
Joined: Thu Mar 18, 2010 1:24 pm

Re: A look at the research and evidence on time-series MOM

Post by grap0013 »

Ok folks, after much deliberation and research I'm buying a 5% slug of TS MOM. The evidence is sound and I see the benefit. I have a much better understanding of this topic now. In vivo correlations looks pretty and I see the role in the overall portfolio.

I'm taking my 20% QSPIX and doing QSPIX:QMHIX 15:5.

This mix effectively produces in real % terms:

5% TS MOM
5% CS MOM (1/3 from QSPIX)
5% value (1/3 from QSPIX)
2.5% carry trade (1/6 from QSPIX)
2.5% defensive/profitability (1/6 from QSPIX)

That's some good return source diversification. Paring down EM value and SCV today with a planned buy order of QMHIX tomorrow 1/24/17.

I'm posting this for later when QMHIX does really well and I have proof I bought low. :-)
There are no guarantees, only probabilities.
Random Walker
Posts: 5561
Joined: Fri Feb 23, 2007 7:21 pm

Re: A look at the research and evidence on time-series MOM

Post by Random Walker »

Image

This "smile chart" makes TS Momentum very interesting! And it's performance in 8 of 10 prior bear markets gets one's attention as well

Dave
alphabeta01
Posts: 14
Joined: Tue Oct 18, 2016 10:07 am

Re: A look at the research and evidence on time-series MOM

Post by alphabeta01 »

-----
Last edited by alphabeta01 on Thu Sep 30, 2021 12:27 pm, edited 1 time in total.
matto
Posts: 163
Joined: Fri Jan 03, 2014 2:51 am

Re: A look at the research and evidence on time-series MOM

Post by matto »

grap0013 wrote:Ok folks, after much deliberation and research I'm buying a 5% slug of TS MOM. The evidence is sound and I see the benefit. I have a much better understanding of this topic now. In vivo correlations looks pretty and I see the role in the overall portfolio.

I'm taking my 20% QSPIX and doing QSPIX:QMHIX 15:5.

This mix effectively produces in real % terms:

5% TS MOM
5% CS MOM (1/3 from QSPIX)
5% value (1/3 from QSPIX)
2.5% carry trade (1/6 from QSPIX)
2.5% defensive/profitability (1/6 from QSPIX)

That's some good return source diversification. Paring down EM value and SCV today with a planned buy order of QMHIX tomorrow 1/24/17.

I'm posting this for later when QMHIX does really well and I have proof I bought low. :-)
You should just get a job at AQR. They may have lower fee employee versions of their funds (and in their 401k, too).
User avatar
grap0013
Posts: 1892
Joined: Thu Mar 18, 2010 1:24 pm

Re: A look at the research and evidence on time-series MOM

Post by grap0013 »

alphabeta01 wrote:
Its interesting the way you broke out the QSPIX. How does it work for QLEIX (Long/Short fund)? How do we calculate the Value/Size/Profitability premiums on QLEIX?
Not sure. I have not thought much about the long/short funds. I hedge my tilting implementation companies. I'm about 40% Vanguard funds, 35% AQR funds, and 25% Fundamental Index funds. Spread your bets around.
There are no guarantees, only probabilities.
SnowSkier
Posts: 268
Joined: Fri Mar 16, 2012 12:21 pm
Location: Austin

Re: A look at the research and evidence on time-series MOM

Post by SnowSkier »

alphabeta01 wrote:How does it work for QLEIX (Long/Short fund)? How do we calculate the Value/Size/Profitability premiums on QLEIX?
FYI, here's some great info on QLEIX / QLENX (and QMNIX / QMNNX ) from Sam Lee, formerly of Morningstar:

http://svrn.co/blog/2016/5/23/aqr-long- ... l-a-review
Post Reply