Absolute Momentum and Stock Momentum Strategies: Friends, not enemies

/Absolute Momentum and Stock Momentum Strategies: Friends, not enemies

Absolute Momentum and Stock Momentum Strategies: Friends, not enemies

By | 2017-08-18T17:11:11+00:00 March 31st, 2015|Momentum Investing Research, $mtum|29 Comments

There is sometimes confusion associated with so-called “momentum” strategies–we want to clear the muddy waters. We break momentum into two categories to differentiate between the different approaches to momentum:

(1) Absolute, or time-series momentum: an asset classes’ own past return, considered independently from the returns of other asset classes, predicts its future performance. This could apply at the level of individual securities as well.

(2) Relative Strength, “Stock momentum,” or Cross-sectional momentum: an asset classes’ performance, relative to other asset classes, predicts its future relative performance. This could also apply at the individual security level, when performance is compared versus the performance of comparable securities; thus, the term is not exclusive to “asset classes.”

This post highlights two facts found in the data:

  1. Individual stock momentum has worked over the past 87 years. This is commonly labeled cross-sectional momentum and is often used in momentum investing funds and/or ETFs.
  2. Using a simple absolute return (i.e., time-series) signal, appears to limit drawdowns over the past 87 years. This is commonly referred to as time-series momentum and is often used as a risk-management overlay in tactical asset allocation systems.

We hope to educate everyone on the difference between the two ideas, and show that they are not competitors, but can be used in conjunction with one another.

Cross-sectional Momentum

Cross-sectional momentum, at the individual stock level, is a technique to sort stocks based on some measure of past return. Most momentum-based ETFs or mutual funds trade based on this general approach.

At a high level, for individual stocks, cross-sectional momentum results can be summarized as follows:

  • Short-term (1-month look-back measurement, 1-month holding period) shows reversals
  • Intermediate-term (6 to 12-month look-back measurement, 1 to 12 month holding periods) shows continuation
  • Long-term (36-month look-back measurement, 3-year holding period) shows reversal.

Below we  document the intermediate-term momentum effect using Ken French’s data. The momentum portfolios are formed monthly, by ranking all stocks on the past 12 months returns (ignoring last month — the academic 12_2 momentum variable). We look at the value-weight returns to the top decile of all firms ranked on their past 12_2 momentum, and compare this to the SP500, Long-term U.S. Bonds, and the risk-free rate.

Specifically, here are the four portfolios:

  1. MOM_10 = Value-weight returns to the top decile formed on 12_2 momentum. Data is found here.
  2. SP500 = Total return of the S&P 500 Index
  3. LTR = Merrill Lynch 7-10 year Government Bond Index (prior to 6/1982, Amit Goyal Data)
  4. RF = Total return to the risk-free asset (U.S. treasury bills).

The returns runs from 1/1/1928 to 12/31/2014. Results are gross of fees. All returns are total returns and include the reinvestment of distributions (e.g., dividends).

momentum funds (1)

The results are hypothetical results and are NOT an indicator of future results and do NOT represent returns that any investor actually attained. Indexes are unmanaged, do not reflect management or trading fees, and one cannot invest directly in an index. Additional information regarding the construction of these results is available upon request.

Takeaways:

  1. Cross-sectional momentum worked  well over the past 87 years. The strategy outperformed the index (SP500) by over 600 bps! Of course, the actual returns will be lower after transaction costs, which could be substantial due to the monthly rebalancing aspect of the strategy.
  2. Stocks were a better bet than Treasury bonds (LTR) and bills (RF) over the past 87 years.

Time-series Momentum

Time series momentum is a way to measure an asset classes’ own past return. Market participants use this measure to time market exposures. The measure is related to the simple moving average rule popularized by Meb Faber, which we point out in our post.

The time-series, or absolute momentum rule, popularized by Gary Antonnaci, is assessed monthly as follows:

  1. Excess return = total return over past 12 months less return of T-bill.
  2. If Excess return >0, go long risky assets. Otherwise, go alternative assets (T-Bills)

The basic premise behind the time-series momentum trading rule (TSMOM) is that if the trend (over the past 12 months) is positive, stay in the risk assets (“the trend is your friend”). Otherwise, if the trend is negative, invest in risk-free assets. This rule is very similar to the simple moving average rule, as discussed here. (there are other applications in futures that go long and short, but we are focused on equities here).

We use the time-series momentum (TSMOM) signal from the S&P 500 on both the S&P 500 and the cross-sectional momentum return series. This is in order to have the same “rule” applied to both return series. Comparing the performance of a cross-sectional momentum stock strategy against the S&P 500 with a TSMOM rule is like comparing apples to oranges. We want to run a proper horse race that highlights the benefits of both time-series–AND cross-sectional–momentum working together.

Here are the four portfolios we test:

  1. MOM_10 = Value-weight returns to the top decile formed on 12_2 momentum. Data is found here.
  2. MOM_10 TSMOM = Depending on the TSMOM rule (using SP500 TSMOM rule), the portfolio is either invested in MOM_10, or in the risk-free (RF) asset described above.
  3. SP500 = Total return of the S&P 500 Index.
  4. SP500 TSMOM = Depending on the TSMOM rule, the portfolio is either invested in the SP500, or in the risk-free (RF) asset described above.

The returns runs from 1/1/1928 to 12/31/2014. Results are gross of fees. All returns are total returns and include the reinvestment of distributions (e.g., dividends).

momentum funds (2)

The results are hypothetical results and are NOT an indicator of future results and do NOT represent returns that any investor actually attained. Indexes are unmanaged, do not reflect management or trading fees, and one cannot invest directly in an index. Additional information regarding the construction of these results is available upon request.

Takeaways:

  1. The Time-Series Momentum rule (TSMOM) helps to reduce volatility and increase the risk-adjusted returns as measured by the Sharpe ratio.
  2. The drawdowns are decreased for both the SP500 and the MOM_10 portfolios — the TSMOM rule helped to reduce drawdowns.
  3. While the CAGR increases when used on the SP500, the CAGR is higher for the MOM_10 portfolio compared to the MOM_10 TSMOM portfolio — timing the market is difficult!

Summary

Clearly, there are benefits–at least historically–to using both cross-sectional and time-series momentum. The 2 momentum effects are not competitors, but complements. Viewing them as competitors does a disservice to both types of momentum. Overall, we hope this post helped to clarify the difference between cross-sectional momentum and time-series momentum.

The main results are as follows:

  • Using individual stock momentum has worked over the past 87 years. This is commonly labeled cross-sectional momentum.
  • Using a simple trend following rule appears to limit drawdowns over the past 87 years. This is commonly called time-series momentum.
  • Combining cross-sectional momentum and time-series momentum has worked better than using either of the stand-alone elements.

Go momentum!


  • The views and opinions expressed herein are those of the author and do not necessarily reflect the views of Alpha Architect, its affiliates or its employees. Our full disclosures are available here. Definitions of common statistics used in our analysis are available here (towards the bottom).
  • Join thousands of other readers and subscribe to our blog.
  • This site provides NO information on our value ETFs or our momentum ETFs. Please refer to this site.

Print Friendly, PDF & Email

About the Author:

Jack Vogel
Jack Vogel, Ph.D., conducts research in empirical asset pricing and behavioral finance, and is a co-author of DIY FINANCIAL ADVISOR: A Simple Solution to Build and Protect Your Wealth. His dissertation investigates how behavioral biases affect the value anomaly. His academic background includes experience as an instructor and research assistant at Drexel University in both the Finance and Mathematics departments, as well as a Finance instructor at Villanova University. Dr. Vogel is currently a Managing Member of Alpha Architect, LLC, an SEC-Registered Investment Advisor, where he heads the research department and serves as the Chief Financial Officer. He has a PhD in Finance and a MS in Mathematics from Drexel University, and graduated summa cum laude with a BS in Mathematics and Education from The University of Scranton.
  • RT1C

    If your point is that momentum works–both cross-sectional and time-series–I’m in agreement. The academic studies show the effectiveness and persistence of the strategy–both types–at least in the past. (I dispute that Antonacci is the popularizer of absolute momentum; his book is entitled “Dual Momentum” and is about the combination of both cross-sectional and absolute momentum, much as you have done here in the second part; if anything, he is the popularizer of applying both strategies together).

    However, I want to challenge you to a higher standard of evidence if your goal is to provide a framework for an evidence-based approach to investing strategy selection. I believe the data you present to support your argument is biased and doesn’t make the point in a robust manner. Allow me to offer the following criticisms (and I speak as a layman, so correct me if I’m wrong):

    1. Your MOM_10 is based on the CRSP database of all U.S. stocks. That means it includes a large universe of small stocks compared to the S&P 500. Unless you correct for market cap, you are comparing apples and oranges if you compare all US stocks to the S&P 500.

    2. The problem is likely exacerbated by the greater volatility of small stocks, which means they are more likely to end up at the extreme deciles of the distribution and thus get selected by a cross-sectional momentum strategy.

    3. Your analysis doesn’t include trading costs, but especially if the population is biased towards small stocks, then trading costs will be relatively large.

    4. There is an argument in the literature that reversal strategies are biased towards small stocks for the reasons similar to those given above; this increases trading costs. (See de Groot, Huij, and Zhou, “Another look at trading costs and short-term reversal profits”). I believe intermediate momentum would suffer the same effect as reversal strategies, albeit attenuated due to averaging over the look-back period. Thus, again, without accounting for trading costs that are higher for small stocks, MOM_10 is excessively optimistic about performance.

    5. Asness et al. (“Fact, Fiction and Momentum Investing”) recently argued against momentum myths. They showed that both small and large stocks exhibit momentum effects, but small stock cross-sectional momentum significantly outperformed large stock cross-sectional momentum (see their Table 5). Again, that would be an argument against drawing conclusions from a CRSP vs. S&P500 comparison.

    6. There have been many attempts to explain momentum and its persistence. One explanation is that it reflects slow diffusion of information. This effect is greater in small stocks. This may explain increased momentum strategy profits in small stocks even after adjusting for the size factors, at least on paper before accounting for trading costs.

    In summary, I think for investors like myself to be effective users of momentum, we need to understand its true performance, with costs included, and with biases for other risk factors like market cap eliminated. Given the poor relative performance of several high profile cross-sectional momentum funds, we do need to get a solid grip on the problems that accompany some momentum strategies, and understand how to effectively use the strategy. This post is a contribution towards that, but because of the shortcomings discussed above, I think it leaves us wondering how effective each strategy component really is. For example, could it be that S&P500 TSMOM (via a low cost index fund) might be just as effective (similar or greater Sharpe Ratio) as MOM_10 TSMOM if adjusted for trading costs and normalized for the size factor? I hope you’ll not stop with this post but dig deeper into the reality and basis of momentum performance (both types).

  • Great comments and thanks for the thoughtful feedback. We try and limit the size of our posts so they don’t end up being full-blown research papers. So each post merely reflects some high-level summary points surrounding a basic research question.

    Also, Gary claims that using cross-sectional momentum to pick stocks should be avoided because these methods fails to curb risk compared to a broad index with time-series momentum. Plus, as a point of clarification, his dual momentum concept applies cross-sectional momentum to asset classes–not individual securities. The point of this post was to analyze cross-sectional stock selection AND time series momentum together and compare that to time-series momentum applied on a broad index.

    Some quick comments to your specific points

    1. We look at value-weight returns, which gives a lot more weight on larger stocks. If one conducts the analysis only on large caps, the results still hold.

    2. Again, the results are all robust to market cap.

    3. Yes, trading costs will definitely play a role. The extent of their importance we leave to our readers who face different facts and circumstances when it comes to trading and execution costs.

    4. Agree. Trading costs are definitely a factor to consider.

    5. We have done internal research on this subject as well. Like all anomalies–the effects are usually magnified in smaller names where limits of arbitrage are higher. But robust anomalies still have significant unexplained excess returns even after considering reasonable transaction costs.

    6. There is actually a large and well-established empirical and theoretical literature that would rival the literature on the “value anomaly” that helps explain why and how the momentum anomaly exists. There seem to be some clear behavioral finance elements involved: poor investor decision making and limits of arbitrage…the perfect ingredients for the long-term existence of anomalous asset pricing effects.

    Thanks for your comments on the momentum anomaly. We sit firmly on the side of the fence that momentum can be effective on a size-adjusted after-fee basis, but we respect those who think it is hocus-pocus.

    re “poor performance of high profile funds”: the products available in the retail/public marketplace are typically called “momentum” but they are really closet index funds focused on “capacity” at the expense of expected performance. MTUM is a great example. We can’t control how products are designed and it is unfortunate that there are few cross-sectional momentum products available that actually captures the benefits of momentum in a consistent and highly active manner.

    Good luck

  • Cheryl

    I think you are mistaken when you say Gary claims that cross-sectional momentum to pick stocks should be avoided because these methods fail to curb risk compared to a broad index with time-series momentum. Gary says that cross-sectional momentum applied to stocks is not as good as when applied to assets classes because of the high transaction costs associated with cross-sectional momentum when used with individual stocks. Gary doesn’t just compare cross-sectional stock momentum to time-series momentum. Gary compares the performance of the AQR cross-sectional stock momentum index to its benchmark Russell 1000 index, which has about the same Sharpe ratio as the AQR stock momentum index without accounting for high transaction costs that aren’t included in the AQR index.

    Also, it seems to me you could have easily compared your value-weighted momentum portfolio to a similar value-weighted portfolio of all stocks instead of the S&P 500. You also could have made some reasonable estimates of transaction costs, as is done in many academic papers.

  • dph

    “Few” or none? Where would one start to look for a decent one? Outside of your firm whatever fund families have well crafted anomaly capturing financial products?

    “”it is unfortunate that there are few cross-sectional momentum products available that actually captures the benefits of momentum in a consistent and highly active manner.””

  • Cheryl,

    Thanks for your comments.

    First, this is not a post saying Antonacci’s is nuts. In fact, we credit him at the very outset for bringing absolute momentum to the mainstream. That said, we do push a bit on the argument that stock-momentum is useless in an absolute momentum context–nothing personal here, simply looking at the evidence.

    Here is a post by Gary with a quote:
    http://www.dualmomentum.net/2014/11/individual-stock-momentum-that-dog-dont.html
    “In terms of both risk and return, momentum is more effective when it is used with asset classes or broad indexes, and when it incorporates trend-following absolute momentum, as described in my book.”

    As we highlight above, it isn’t more effective–there is a 500bps CAGR spread–and that is using generic 2-12 stock momentum. More sophisticated methods can generate 700-800bps+. I’m not sure transaction costs would eat that much away.

    As per the last comment, we mentioned that standard products out there are not pure momentum funds, but closet-index products–AQR momentum is a perfect example. Plus, the past 5 yrs have not been a good run for momentum–maybe it is dead, or maybe it is in a funk? Who knows. But there is a lot of evidence that over longer periods it is anomalous.

    Also, the value-weight portfolio of all stocks is 99.99% correlated with the S&P 500. You can run the analysis yourself by examining the VW_CRSP return series off of Ken French’s site if you’d like to investigate further.

    Thanks again and hopefully these comments enhance the discussion. We certainly don’t claim to have all the answers, but we do share our research to generate a conversation…

  • Steve

    The Dorsey-Wright based powershares etf’s are a little more concentrated (e.g. technical leaders holds 10% of eligible universe)

  • Cheryl

    Wes,

    With all due respect, Gary never says that stock-momentum is useless in an absolute momentum context. The quote you give above is of him comparing cross-sectional stock momentum to.cross-sectional momentum of asset classes or broad indexes. You never compare cross-sectional momentum of stocks, adjusted for transaction costs, to cross-sectional momentum of asset classes, so how can you say that Gary is incorrect? He says in general that cross-sectional momentum can be improved upon by adding absolute momentum to it, not as a substitute for it. That’s what his dual momentum strategy is all about.

    Also, i don’t think you are doing a fair comparison of cross-sectional stock momentum to a proper benchmark. This would be true even if you did use a broader universe than the S&P 500. Lesmond, Schill, and Zhou (2014) point out that the majority of momentum portfolio abnormal returns are from stocks that are “small, high beta, and off-NYSE.”

    http://papers.ssrn.com/sol3/papers.cfm?abstract_id=256926

    These should be your benchmark. Your CAGR on them should be substantially higher than with the S&P 500. Bid/ask spreads are also wider on these stocks. The paper looks at trading costs in light of this and concludes by saying, “…returns associated with relative strength strategies do not exceed trading costs.”

  • Thanks for your comments, Cheryl.

    1. Check out the Frazzini et al paper on transaction costs and review their conclusions.

    http://papers.ssrn.com/sol3/papers.cfm?abstract_id=2294498

    “We conclude that the main anomalies to standard asset pricing models are robust, implementable, and sizeable.”

    2. The paper you cite refers to long/short portfolios, not long-only portfolios. Short portfolios are obviously hazardous to your wealth and involve a lot of transaction costs.

    We respectively disagree on the transaction costs associated with running long-only momentum portfolios. All good–that’s what makes a marketplace.

    Best of luck,

    Wes

  • RT1C

    Good discussion… BTW, I know you don’t think Gary is nuts because you wrote a very generous forward in his book (and one that was unjustly self-deprecatory, I might add; you write well too)!

    Cheryl made the point I was going to. Just to augment her response, here is the quote from Ch. 2 of Gary’s book (after he reviewed momentum funds from AQR, DWA and BlackRock): “All of these publicly available products apply relative strength momentum to individual stocks. They therefore miss the potential risk-reducing benefits of cross-asset diversification. Using momentum with individual stocks also results in substantially higher transaction costs than applying momentum to broad asset classes and indexes. AQR, for example, estimates transaction costs of its U.S. momentum index to be 70 basis points per year.”

    Thus, while adding the absolute momentum overlay is a key part of improving performance (especially, reducing downside risk), Gary does seem to be making the case for asset classes vs. individual stocks on the basis of transaction costs, not diversification. His diversification argument is related to diversification between asset classes, not the diversification within a class. Of course, one could obtain cross-asset class diversification following an individual stock strategy (at least for equity classes, such as US Domestic vs. Global stocks, which is what Gary’s GEM does) too, but neither Gary nor you analyzed that. So, again, it seems to me his argument for indices/EFTs vs. individual stocks is transaction costs, not diversification.

    Then, later in the book, pp. 124-125, Gary does the comparison Cheryl mentions, i.e., AQR vs. Russell 1000, and notes the transaction costs would make AQR underperform Russell 1000. On. p. 131 Gary suggests sector momentum, which offers lower transaction costs than individual stocks but enhanced performance over total market momentum. That may be a more promising intermediate approach.

    Anyway, the point here isn’t to battle over quotes! The issue for me is simply that one should make a fair, unbiased comparison, and that includes using comparable stocks and including transaction costs. The S&P500 TSMOM had Sharpe Ratio of 0.54, not so far behind the comparative case of 0.66. What would the comparative SRs be if stock size and transaction costs were included? (Maybe one way to do this would be to regress the results vs. a 3-5 factor model as Gary did in his book, and compare alphas; that would let you normalize for market cap differences. One would want to look at risk too, though).

    Well, I’ve been to wordy here, so will address other issues in another post. However, I think it would be interesting to compare individual stock based momentum vs. index/ETF (inclg. sector strategies) momentum AFTER including transaction costs, and see which is best.

    Thanks for the discussion!

  • RT1C

    Are you sure on #1/#2? From the description of the data library you said you used, I’m not seeing anything about value weighting. http://mba.tuck.dartmouth.edu/pages/faculty/ken.french/Data_Library/det_10_port_form_pr_12_2.html
    Even if value weighted, isn’t that used only for determining the return of the decile, not to determine which stocks are IN the decile (which is the relevant point for my argument)?
    On your last point, I think of AQR in particular. I don’t understand why you call such funds “closet index funds” and would appreciate a fuller explanation. My understanding is that it uses t12-2 momentum balanced quarterly on the top third (not decile) of stocks. While that isn’t as focused as a decile, it is still a true momentum strategy by my understanding. (And incidentally, if that has 0.7% trading costs, I assume a fund more narrowly focused and rebalanced monthly instead of quarterly would be significantly higher).

  • Jack Vogel, PhD

    The returns are formed by value-weighting the stocks in the top decile on momentum. So smaller firms could be included in the portfolio, but they would have a lower weight. Does momentum work in larger stocks — yes. We have done research on this and will post results in the future.

  • RT1C

    Wes/Jack, to change the topic…have you ever looked at the “residual momentum technology” behind the Robeco fund? http://www.robeco.com/en/professionals/products/product-information.jsp?pdstchn=robecocom&pfndid=6170&plang=English

    Here’s a synopsis of the technology: http://www.robeco.com/en/professionals/insights/quantitative-investing/momentum/robecos-residual-momentum-less-risky-and-more-sustainable.jsp

    You can click a link at the bottom of the page to download a white paper which is very informative.
    I’d be interested in your reaction since you are well-read in the academic literature and I’m sure aware of the various commercial offerings. I would love to invest in a product like this. Unfortunately, the fund is not offered in the U.S., and Robeco said as yet they don’t have a U.S. partner to make it happen. (Is this something your firm might be able to do?)

  • Yes, that is a reasonable approach that moves beyond the basic 2-12 methodology. We think there are better methods, but this one is definitely an improvement over the generic momentum.

    The key issue with momentum strategies is the tax-management piece. We have solutions for these problems. We can talk offline at some point.

  • RT1C

    The Asness et al. paper makes the same point; momentum works in both small and large stocks. But it showed that returns were significantly higher with small stocks. So, given that the top decile does contain relatively more small stocks, I’ll grant that the value-weighting will diminish their impact, but vs. S&P500, such a portfolio should still be significantly skewed towards small stocks. Here’s my math:

    (data from http://www.nasdaq.com/screening/company-list.aspx)
    NASDAQ — 2972 stocks
    NYSE — 3284 stocks
    AMEX — 412 stocks

    Your dataset from French is based on that stock universe (of course, the number of companies has varied historically). Total currently is 6668 stocks.

    If we assume that cross-sectional momentum is independent of size (and I don’t), then the 500 stocks comprising the S&P500 represent only 7.5% of this stock universe. Thus, in a the top decile (if momentum is randomly distributed), they would only comprise about 7.5% of the stocks in the decile. S&P500 captures about 80% of US market cap (http://us.spindices.com/indices/equity/sp-500); thus, 20% of the decile would be due to smaller stocks on a cap-weighted basis. But if we assume the smallest S&P 500 stocks are more likely to get into the top decile too, then maybe this number grows to 25% (just a guess). And then if small (non-S&P) stocks disproportionately get selected into the top decile, maybe that number doubles (again, just a guess, it may be more or less). My point is simply that it doesn’t seem difficult for the top decile to have a large smaller stock bias, maybe 50%.

  • RT1C

    Tax management not an issue for me (IRA funds).
    I shot you a note on your webform and look forward to response.

  • Jack Vogel, PhD

    No argument here — using that dataset will include smaller stocks.

  • jonah789

    Wes – I don’t see how short positions can have much higher transaction costs than long position when both sides are rebalanced the same way. Besides, Lesmond et al say that the majority of trading strategy returns is generated by short positions. So long-only results may be even worse than what they show for both.

    All the writers of the Frazzini paper are associated with AQR, which manages stock momentum funds, so their research assumptions may or may not be unbiased. I would guess that all your posts about cross-sectional stock momentum mean that you plan to offer a stock momentum ETF. Correct?

  • Jonah,

    Please review the recent paper by Stambaugh et al. for some background into market structure. Shorting is more expensive than going long…this is a well established empirical fact and also a well known practitioner fact.

    Here is the paper: http://users.cla.umn.edu/~jianfeng/Anomalies_JFE_12.pdf

    I’m glad that you are providing a differing opinion on our findings. That’s great. Our blog is meant to inspire conversation and let people consider the evidence. We’ll leave readers to make their own conclusions.

    Best of luck

  • Paul Novell

    Wes/Jack, thanks for the always great work. Curious as to how momentum is working over time (cross sectional and absolute)? Are the excess returns to each staying the same or decreasing as market participants ‘catch-on’? Or is momentum such a fundamental reflection of human behavior that it’s effect have been consistent over time?

    Paul

  • Jack Vogel, PhD

    I added a picture showing the rolling 5-year max DDs for the strategies. It looks as though the absolute momentum rules have worked by limiting DDs over the entire time period.

  • Paul Novell

    Thanks Jack.

  • Govind

    In a more recent paper, Fisher, Shah, and Titman show that transaction costs are much closer to Lesmond’s higher estimates than to Frazzini’s.
    https://papers.ssrn.com/sol3/papers.cfm?abstract_id=2472936
    There is also the price impact issue with respect to scalability.

  • This relies on b/a spreads, which don’t represent actual transactions costs when trading these systems for professional investors (which are typically done via VWAP algos set to be not very aggressive).

    The AQR study is much more robust and thorough than the 1-liner in a footnote in the Fisher, Shah, and Titman paper.

    All that said, there is no doubt that momentum strategies that are managed for alpha (concentrated, higher turnover) have scalability issues. But those limits start coming into effect in the low $B AUM range. God bless us or any other investor who ends up with that problem.

  • Govind

    Novy-Marx and Velikov
    http://rnm.simon.rochester.edu/research/ToAatTC.pdf
    and Beck, Hsu, Kalesnick and Kostka
    http://www.cfapubs.org/doi/pdf/10.2469/faj.v72.n5.6
    using different approaches show that stock momentum has already reached capacity constraints where abnormal profits should no longer exist.

  • Yes, but this assumes the capital chasing the factors are permanent. They’re not. They are performance chasing flows, which actually increases the expected premium payoff.

  • Govind

    I haven’t seen any real time positive alpha from momentum that would cause performance chasing. Can you point me toward some?

  • Govind

    The paper shows in principle how that would work, but the data they use is from 1980 through 2006. There were no momentum funds before 2007, and momentum was not popular until recently.

  • Here is a shot from bb on AUM and performance with the momentum fund.
    https://uploads.disquscdn.com/images/f823d4822b236189ada349c668f83b282e01cf2d561aff6603f139d2a0767832.png
    Hsu has some interesting work on the behavior gap
    https://www.researchaffiliates.com/en_us/publications/articles/488_if_factor_returns_are_predictable_why_is_there_an_investor_return_gap.html
    They don’t pinpoint momentum specifically in that paper.
    My guess is that assuming that momentum fund investors are 20yr buy and holders is probably unrealistic. I’d argue that investors are actually more likely to be performance chasers in momentum funds versus other “factors” such as value.