Mechanisms for disappearance of factor premia -- how does it work for momentum?

Discuss all general (i.e. non-personal) investing questions and issues, investing news, and theory.
Post Reply
finvestor
Posts: 20
Joined: Thu Apr 20, 2017 1:22 pm

Mechanisms for disappearance of factor premia -- how does it work for momentum?

Post by finvestor » Sat Jun 16, 2018 6:00 am

It is often argued that once a trading strategy that has previously beaten the market is made public, it stops working as increasing number of investors start using the strategy. This may also apply to factor premia often discussed in this forum (small, value, momentum, quality, etc.). I would be interested in learning more about the mechanisms via which factor premia might disappear, or get reduced in magnitude, following an increased number of investors starting to use such strategies.

I find it rather straightforward to understand how a factor premium associated with a relatively "static" property of individual stocks might disappear as more investors start to flock into the factor. For instance, small cap stocks probably tend to stay small cap stocks for an extended period of time. Hence, when more and more people invest in such small cap stocks, they become more expensive and hence their expected future rate of return goes down, potentially leading to the disappearance of the size premium.

However, I find it more tricky to understand how the momentum premium would disappear when more people start investing in momentum stocks. Momentum stocks are stocks which have recently performed well, and if more investors start buying such recent winners, their price should go up even more, hence amplifying the momentum effect. Right?

At some point a stock that has been a winner for a while might of course become so expensive as a result of increased investor interest that it will underperform in the future. But as soon as that starts to happen, that particular stock is no longer a momentum stock.

Is there something obvious that I'm missing, or is there a case to be made that as a result of the difficulty to "kill" the momentum anomaly, it might survive in the future?

pkcrafter
Posts: 12830
Joined: Sun Mar 04, 2007 12:19 pm
Location: CA
Contact:

Re: Mechanisms for disappearance of factor premia -- how does it work for momentum?

Post by pkcrafter » Sat Jun 16, 2018 6:52 pm

finvestor wrote:
Sat Jun 16, 2018 6:00 am
It is often argued that once a trading strategy that has previously beaten the market is made public, it stops working as increasing number of investors start using the strategy. This may also apply to factor premia often discussed in this forum (small, value, momentum, quality, etc.). I would be interested in learning more about the mechanisms via which factor premia might disappear, or get reduced in magnitude, following an increased number of investors starting to use such strategies.

I find it rather straightforward to understand how a factor premium associated with a relatively "static" property of individual stocks might disappear as more investors start to flock into the factor. For instance, small cap stocks probably tend to stay small cap stocks for an extended period of time. Hence, when more and more people invest in such small cap stocks, they become more expensive and hence their expected future rate of return goes down, potentially leading to the disappearance of the size premium.

However, I find it more tricky to understand how the momentum premium would disappear when more people start investing in momentum stocks. Momentum stocks are stocks which have recently performed well, and if more investors start buying such recent winners, their price should go up even more, hence amplifying the momentum effect. Right?

At some point a stock that has been a winner for a while might of course become so expensive as a result of increased investor interest that it will underperform in the future. But as soon as that starts to happen, that particular stock is no longer a momentum stock.

Is there something obvious that I'm missing, or is there a case to be made that as a result of the difficulty to "kill" the momentum anomaly, it might survive in the future?
I don't think you're missing anything. Momentum is a play on behavior. I also suspect the pros play this game much better than individual investors. Pros will enhance excitement and then sell first. Might be the same for small caps.

Paul
When times are good, investors tend to forget about risk and focus on opportunity. When times are bad, investors tend to forget about opportunity and focus on risk.

grok87
Posts: 8123
Joined: Tue Feb 27, 2007 9:00 pm

Re: Mechanisms for disappearance of factor premia -- how does it work for momentum?

Post by grok87 » Sat Jun 16, 2018 7:36 pm

First of all WHAT momentum premium? Here is the Ken French data for the momentum factor by year with CAGRs for periods ending in 2017 shown in the last column.

Code: Select all

               year       return    cagr through 2017
                1988      -5.2%      2.1%   30 yr
                1989      27.9%      2.3%
                1990      17.6%      1.5%
                1991      14.7%      0.9%
                1992       3.2%      0.5%
                1993      23.7%      0.3%   25 yr
                1994       3.2%     -0.5%
                1995      17.4%     -0.7%
                1996       6.5%     -1.4%
                1997      11.9%     -1.8%
                1998      23.5%     -2.4%   20 yr
                1999      34.3%     -3.6%
                2000      15.4%     -5.4%
                2001       4.3%     -6.5%
                2002      25.7%     -7.1%
                2003     -24.2%     -9.0%   15 yr
                2004      -0.4%     -7.8%
                2005      14.9%     -8.3%
                2006      -7.8%    -10.0%
                2007      21.6%    -10.2%
                2008      13.2%    -12.9%  10 yr
                2009     -82.7%    -15.4%
                2010       6.0%      3.1%
                2011       7.3%      2.7%
                2012       1.6%      2.0%
                2013       7.9%      2.1%    5 yr
                2014       1.6%      0.6%
                2015      20.7%      0.3%   3 yr
                2016     -20.3%     -8.5%
                2017       5.0%      5.0%   1 yr
for periods ending in 2017 the CAGRs for the momentum premium have been predominantly "negative or small positive" for the past 30 years, and that is before transaction costs. Transaction costs are of course pretty material for a strategy like momentum that involves lots of trading.

To answer your question, I think people get excited about trends and start following them. Arguably this has intensified with social media etc.
And then something changes suddenly like it did in 2009 or more recently 2016 and all of the alleged gains evaporate in an instant.
Keep calm and Boglehead on. KCBO.

garlandwhizzer
Posts: 1883
Joined: Fri Aug 06, 2010 3:42 pm

Re: Mechanisms for disappearance of factor premia -- how does it work for momentum?

Post by garlandwhizzer » Sat Jun 16, 2018 9:28 pm

As grok87 points out MOM long/short portfolios can be lethal to portfolio value quickly when a bear market suddenly reverses. In 2009, the FF MOM factor lost 82.7% of its value in 12 months. It can take quite a long time to recover when your portfolio is only worth 17.3% of its value 12 months ago. Your MOM portfolio now needs to gain by a cumulative 578% to get back to where you were before the disaster happened. For practical purposes you're ruined. Also as he points out, MOM is a high trading frequency strategy. In the more illiquid SC space where bid ask spreads are wide, illiquidity is greater, and trading frictions are high, it is almost impossible for MOM to outperform after costs in the real world, one reason there are so few SC MOM funds. LC MOM is a space played in extensively by private equity and active fund managers, so it is a highly competitive arena.

Even though a factor like MOM is deeply grounded in human behavior (which never changes), that does not guarantee that it can be effectively harvested it in the real world, especially after costs. The FF MOM figures, again as grok87 points out, are cost-free and even without costs its results have been underwhelming for a long, long time. If you wish to invest in MOM the safest approach is long-only LC where trading costs are lower and stocks are liquid. MTUM has done well in this arena although it does not follow a pure MOM strategy but rather its own index which, oddly, may be its secret sauce: less turnover, less expense, and lower volatility than many MOM funds.

Garland Whizzer

finvestor
Posts: 20
Joined: Thu Apr 20, 2017 1:22 pm

Re: Mechanisms for disappearance of factor premia -- how does it work for momentum?

Post by finvestor » Sun Jun 17, 2018 9:50 am

Thank you for the responses.

Let me clarify my question a bit. First of all, it was not my intention to enter into a debate on whether a momentum premium exists, or is harvestable. Actually, I'm not necessarily referring to FF long-short MOM, the hypothetical excess return of a long-only momentum strategy over a standard cap-weighted index will do just fine.

What I wanted to ask was more like a theoretical, thought experiment type of question like this: First, assume that a momentum premium exists, such that following a certain momentum strategy tends to systematically lead to an excess return over the corresponding cap-weighted strategy. Let's forget about things like trading frictions if that makes the problem easier to analyse. Then, if an increasing number of investors would start investing according to this momentum strategy, how would this lead to a decreased magnitude or even complete disappearance of the premium?

golfCaddy
Posts: 591
Joined: Wed Jan 10, 2018 10:02 pm

Re: Mechanisms for disappearance of factor premia -- how does it work for momentum?

Post by golfCaddy » Sun Jun 17, 2018 10:53 am

This is easy to understand. Suppose on the first trading day of the month, everyone tried to buy high momentum stocks, based on the closing price in the previous calendar month, which they would then sell on the last trading day of the month. Everyone trying to buy the same stocks at the same time would drive up the price of those stocks. Everyone trying to sell the same stocks at the same time would drive down the price of those stocks. The average momentum buyer would buy higher and sell lower, until at some point the premium was zero. You could try to front run that strategy by buying and selling a day earlier than everyone else, but that's changing the definition of momentum.

User avatar
nisiprius
Advisory Board
Posts: 36036
Joined: Thu Jul 26, 2007 9:33 am
Location: The terrestrial, globular, planetary hunk of matter, flattened at the poles, is my abode.--O. Henry

Re: Mechanisms for disappearance of factor premia -- how does it work for momentum?

Post by nisiprius » Sun Jun 17, 2018 11:49 am

finvestor wrote:
Sat Jun 16, 2018 6:00 am
...I find it more tricky to understand how the momentum premium would disappear when more people start investing in momentum stocks. Momentum stocks are stocks which have recently performed well, and if more investors start buying such recent winners, their price should go up even more, hence amplifying the momentum effect. Right?

At some point a stock that has been a winner for a while might of course become so expensive as a result of increased investor interest that it will underperform in the future. But as soon as that starts to happen, that particular stock is no longer a momentum stock....
Well, I think it is. I think momentum is (said to be) bidirectional. I'm not sure how reliable an authority Investopedia is, but it says
If a trader wants to use a momentum-based strategy, he takes a long position in a stock or asset that has been trending up. If the stock is trending down, he takes a short position.
Similarly, Wikipedia says
Momentum in a stock is described as the tendency for the stock price to continue rising if it is going up and to continue declining if it is going down.
Annual income twenty pounds, annual expenditure nineteen nineteen and six, result happiness; Annual income twenty pounds, annual expenditure twenty pounds ought and six, result misery.

david1082b
Posts: 319
Joined: Fri Jun 09, 2017 12:35 pm

Re: Mechanisms for disappearance of factor premia -- how does it work for momentum?

Post by david1082b » Sun Jun 17, 2018 11:53 am

finvestor wrote:
Sun Jun 17, 2018 9:50 am
Thank you for the responses.

Let me clarify my question a bit. First of all, it was not my intention to enter into a debate on whether a momentum premium exists, or is harvestable. Actually, I'm not necessarily referring to FF long-short MOM, the hypothetical excess return of a long-only momentum strategy over a standard cap-weighted index will do just fine.

What I wanted to ask was more like a theoretical, thought experiment type of question like this: First, assume that a momentum premium exists, such that following a certain momentum strategy tends to systematically lead to an excess return over the corresponding cap-weighted strategy. Let's forget about things like trading frictions if that makes the problem easier to analyse. Then, if an increasing number of investors would start investing according to this momentum strategy, how would this lead to a decreased magnitude or even complete disappearance of the premium?
Going back to what you said earlier:
At some point a stock that has been a winner for a while might of course become so expensive as a result of increased investor interest that it will underperform in the future. But as soon as that starts to happen, that particular stock is no longer a momentum stock.
Presumably this stock will have to be discarded if it is no longer a mom stock. If no one wants to buy it except at a massively lower price, it will crash badly, leaving the momentum investors holding the bag all the way down (buy high, sell low). The more popular momentum investing is, the more crowded the trades can become. Strategies can get very trendy, leading to loads of people piling in (especially after big gains get advertised), but eventually they lose popularity. The "greater fool" theory can eventually leads to bad losses due to lack of new fools.

finvestor
Posts: 20
Joined: Thu Apr 20, 2017 1:22 pm

Re: Mechanisms for disappearance of factor premia -- how does it work for momentum?

Post by finvestor » Sun Jun 17, 2018 12:17 pm

golfCaddy wrote:
Sun Jun 17, 2018 10:53 am
This is easy to understand. Suppose on the first trading day of the month, everyone tried to buy high momentum stocks, based on the closing price in the previous calendar month, which they would then sell on the last trading day of the month. Everyone trying to buy the same stocks at the same time would drive up the price of those stocks. Everyone trying to sell the same stocks at the same time would drive down the price of those stocks. The average momentum buyer would buy higher and sell lower, until at some point the premium was zero. You could try to front run that strategy by buying and selling a day earlier than everyone else, but that's changing the definition of momentum.
This is actually a very clean and simple explanation -- thank you.

I wonder how some of the momentum ETFs like MTUM execute their trading. Do they rigidly follow their strategy (buying/selling during a given day), or do they to some extent front run their own strategy?

finvestor
Posts: 20
Joined: Thu Apr 20, 2017 1:22 pm

Re: Mechanisms for disappearance of factor premia -- how does it work for momentum?

Post by finvestor » Sun Jun 17, 2018 12:22 pm

nisiprius wrote:
Sun Jun 17, 2018 11:49 am
finvestor wrote:
Sat Jun 16, 2018 6:00 am
...I find it more tricky to understand how the momentum premium would disappear when more people start investing in momentum stocks. Momentum stocks are stocks which have recently performed well, and if more investors start buying such recent winners, their price should go up even more, hence amplifying the momentum effect. Right?

At some point a stock that has been a winner for a while might of course become so expensive as a result of increased investor interest that it will underperform in the future. But as soon as that starts to happen, that particular stock is no longer a momentum stock....
Well, I think it is. I think momentum is (said to be) bidirectional. I'm not sure how reliable an authority Investopedia is, but it says
If a trader wants to use a momentum-based strategy, he takes a long position in a stock or asset that has been trending up. If the stock is trending down, he takes a short position.
Well, what you describe is a long/short momentum strategy, but a momentum strategy can also be formulated as a long-only strategy. Buying recent winners, and forgetting about the rest... Right? This is what I think most of the momentum funds/ETFs available to retail investors are doing. It is of course true that this is not the pure FF momentum factor which indeed is normally understood as a long/short strategy, as far as I know.

finvestor
Posts: 20
Joined: Thu Apr 20, 2017 1:22 pm

Re: Mechanisms for disappearance of factor premia -- how does it work for momentum?

Post by finvestor » Sun Jun 17, 2018 12:24 pm

david1082b wrote:
Sun Jun 17, 2018 11:53 am
finvestor wrote:
Sun Jun 17, 2018 9:50 am
Thank you for the responses.

Let me clarify my question a bit. First of all, it was not my intention to enter into a debate on whether a momentum premium exists, or is harvestable. Actually, I'm not necessarily referring to FF long-short MOM, the hypothetical excess return of a long-only momentum strategy over a standard cap-weighted index will do just fine.

What I wanted to ask was more like a theoretical, thought experiment type of question like this: First, assume that a momentum premium exists, such that following a certain momentum strategy tends to systematically lead to an excess return over the corresponding cap-weighted strategy. Let's forget about things like trading frictions if that makes the problem easier to analyse. Then, if an increasing number of investors would start investing according to this momentum strategy, how would this lead to a decreased magnitude or even complete disappearance of the premium?
Going back to what you said earlier:
At some point a stock that has been a winner for a while might of course become so expensive as a result of increased investor interest that it will underperform in the future. But as soon as that starts to happen, that particular stock is no longer a momentum stock.
Presumably this stock will have to be discarded if it is no longer a mom stock. If no one wants to buy it except at a massively lower price, it will crash badly, leaving the momentum investors holding the bag all the way down (buy high, sell low). The more popular momentum investing is, the more crowded the trades can become. Strategies can get very trendy, leading to loads of people piling in (especially after big gains get advertised), but eventually they lose popularity. The "greater fool" theory can eventually leads to bad losses due to lack of new fools.
Indeed, and I guess that's why momentum is usually implemented in the liquid large cap space...

User avatar
Earl Lemongrab
Posts: 4712
Joined: Tue Jun 10, 2014 1:14 am

Re: Mechanisms for disappearance of factor premia -- how does it work for momentum?

Post by Earl Lemongrab » Sun Jun 17, 2018 12:32 pm

Depends on why the factor exists. With small and value, many believe that most or all of the factor is risk-related. If so, then that should not be arbitraged away regardless of how well known. If it is, then that in itself is an exploitable inefficiency. After all, the outperformance of stocks over bonds is well-known, but that's certainly not disappeared.
This week's fortune cookie: "Your financial life will be secure and beneficial." So I got that going for me, which is nice.

finvestor
Posts: 20
Joined: Thu Apr 20, 2017 1:22 pm

Re: Mechanisms for disappearance of factor premia -- how does it work for momentum?

Post by finvestor » Sun Jun 17, 2018 1:39 pm

Earl Lemongrab wrote:
Sun Jun 17, 2018 12:32 pm
Depends on why the factor exists. With small and value, many believe that most or all of the factor is risk-related. If so, then that should not be arbitraged away regardless of how well known. If it is, then that in itself is an exploitable inefficiency. After all, the outperformance of stocks over bonds is well-known, but that's certainly not disappeared.
This is an excellent point, of course.

When discussing the excess return associated with the momentum factor that may disappear after the strategy has become popular, one should of course speak about the risk-adjusted return.

User avatar
JoMoney
Posts: 5267
Joined: Tue Jul 23, 2013 5:31 am

Re: Mechanisms for disappearance of factor premia -- how does it work for momentum?

Post by JoMoney » Sun Jun 17, 2018 1:50 pm

Momentum investors lose when they get whipsawed too much in the process of trying to catch a trend.
They can also get caught on the downturn when their strategy fails to execute at the theoretical model exit point. If/when liquidity drys up it can be very difficult to guarantee any specific execution price and significant slippage can occur.

It feels good when you catch a trend and everything is growing fast, but when price gets too far ahead of fundamentals, or you run out of new money chasing the trend (higher prices on very thin volume), it can turn very quickly and the race begins to try to beat the other guy out the door first.
"To achieve satisfactory investment results is easier than most people realize; to achieve superior results is harder than it looks." - Benjamin Graham

jalbert
Posts: 3289
Joined: Fri Apr 10, 2015 12:29 am

Re: Mechanisms for disappearance of factor premia -- how does it work for momentum?

Post by jalbert » Sun Jun 17, 2018 2:13 pm

Earl Lemongrab wrote:
Sun Jun 17, 2018 12:32 pm
Depends on why the factor exists. With small and value, many believe that most or all of the factor is risk-related. If so, then that should not be arbitraged away regardless of how well known. If it is, then that in itself is an exploitable inefficiency. After all, the outperformance of stocks over bonds is well-known, but that's certainly not disappeared.
The equity risk premium likely got smaller once it was understood how to diversify away security-specific risk. Investors would be willing to pay more for a stock knowing this risk is diversified away.

It is possible that an analogous mechanism could be in play with any small or value premia.

Before the Fama-French study was published in 1992 the risk premiums for small-cap value stocks may have been discounted in at the individual stock granularity— capturing that the risk that the individual stock fails to deliver expected future revenue is higher than for the broad market, hence a larger risk premium is discounted in than for the broad market on average.

Once a small-cap value effect was articulated, investors might focus on the asset class in aggregate, diversifying away the additional risk of an individual stock not delivering its value or small risk premia, and accepting the returns of the small and value premia in aggregate.

Much as CAPM demonstrated that security-specific risk can be diversified away, leading investors to be willing to pay more for a stock because they know how to avoid that risk, investing in a portfolio that captures the small and value premia in aggregate may lead to a willingness to pay more for a small value portfolio than if you have to decide which small value stocks will deliver their risk premia.

I’m not saying this is what is happening, but what could be happening. If so, it would just mean that the risk is less and the premia are smaller, which is not necessarily a bad thing.
Index fund investor since 1987.

Post Reply