Does Trend Following Work on Stocks? Part II

Eric Crittenden emailed some new research today titled “Does Trend Following Work on Stocks? Part II (PDF).” Take a look.

---
You might like my 2017 epic release: Trend Following: How to Make a Fortune in Bull, Bear and Black Swan Markets (Fifth Edition). Revised and extended with twice as much content.

22 thoughts on “Does Trend Following Work on Stocks? Part II

  1. This looks like it was a long only system. If that was the case you may as well have just bought and held the SPY since this trend following system barely beat it. And when you consider commissions, fees and taxes then buy and hold would have proabbly resulted in greater capital appreciation than the trend system. Of course, if they went long and short it would been a different story. This is where trend following annihilates buy and hold.

  2. Jamal what do you mean? The trendfollowing system averages 10.2% yr with a 27% max drawdown. The S&P is only 8.4% and has a 45% max drawdown. You get 25% more gain and 40% less drawdown with trendfollowing. That blows away S&P on a risk-adjusted basis by a mile. On a risk-adjusted basis you are getting more than twice the gain. Commissions and slippage are included as mentioned in the Appendix. I do wish they said more about how many positions they held at once and how often they added positions though.

  3. Outperforming a benchmark by less than 2% a year does not warrant be giving them my money and them charging me 2 and 20. Actually Buffett was the first to make this point. He said most funds barely beat a benchmark so why bother givng them your money? When you deduct the fees they charge you their is little point to the fund. Of course if you found a fund that completely blew away the benchmark (of which their are plenty of trend followers such as John Henry’s) then it would be worth it.

  4. On a risk-adjusted basis you are beating the benchmark by almost 8%, not 2%. You are getting twice the return of the benchmark. I’ll gladly pay 2 and 20 to get double the return. The best investors think in terms of risk-adjusted returns.

  5. The point about fees is valid. I wouldn’t pay 2 and 20 for these results either. But I would pay 1.5 and 0 to outpace the stock market without the massive drawdowns and volatility. The point about taxes is valid for the short-term systems. But, the longer-term systems are actually quite tax efficient. The point about commissions is not valid. Realistic transaction costs are reflected in the results. In fact, the transaction costs we assumed are more than double what we’ve experienced in real life trading.

  6. While the results appear impressive for a very simple system, I have these questions:
    1) What is the after-tax return of the system versus the after-tax return of the benchmark? The turnover generates substantial capital gains taxes which further eat into the returns versus the passive index.
    2) They used 3,000 stocks (and not the S&P 500). That means they equal-weighted some number of tiny stocks which may well have risen dramatically, but which may also be illiquid. What would the results look like if they limited themselves to just S&P500 stocks? They used 0.5% roundtrip slippage on the universe of trading. This is fine for largecap stocks, but is it realistic for the smallest 1000?
    3) The period 2002-2006 accounts for almost all of the outperformance. Most other years, the system was inferior. So 5 years out of 20, the system worked. Is this statistically significant? Are investors/traders patient enough to wait 14 years for 5 years of excess returns?
    4) There were no fees charged. If the system charged 2&20, the results would look much worse. Admittedly the risk-adjusted return is better.
    5) What would the returns look like if they used a stop-and-reverse and went short the stocks they sold?

  7. Jamal, you obviously don’t understand the difference between absolute returns and risk adjusted returns. That’s how people get sucked into mutual funds, sitting out 50% drawdowns.

  8. Peter and Ken, if you guys want to pay 2 and 20 for returns that barely beat a pretty bad benchmark then go ahead. I would not be happy doing it.

  9. If you properly use leverage then you’re not barely beating it, you’re beating it by a wide margin. And if you trade your own system then you dont have to pay the 2 and 20, so you beat it by even more.

  10. Leverage in stocks is rather limited compared to futures. And if you use leverage your drawdowns will be seriously greater with worse risk adjusted returns too. Overall, leverage would not improve this study.

  11. You have no idea what you’re talking about. Not only do you not understand risk adjusted returns, you don’t know how to use them to maximize your profits.

  12. Eric,

    What is the performance difference for 3-year highs and all-time highs, assuming the same exit strategy and money management principles?

    Thank you,
    Robert

  13. Ken, how would use leverage on some of these stocks? The stocks were from a list of 3.000. Do you know how many non marginable securities that list probably featured? You couldn’t have used leverage even if you wanted to.

  14. Rocky,

    1) The longer term settings were quite tax efficient. Winners are held while losers are liquidated = long term capital gains & short term losses. The shorter term settings (which we tend to avoid in real life) were not very tax efficient.

    2) It does not mean we equal weighted tiny stocks. We “equal risk” weighted them. Tiny stocks tend to have higher volatility and subsequently smaller position sizes. Had I eliminated the smallest 1000 you’d hardly notice a difference. The transaction costs assumed are more than double what we’ve experienced in real life over the last 6 years.

    3) Incorrect, most of the outperformance came in 2000, 2001, and 2008. That’s when the risk management paid off. It’s not our fault the market waits awhile between blow ups. Are homeowners patient enough to keep buying fire and flood insurance even though disaster hasn’t struck in 5 years?

    4) There were no fees charged. If the system charged 2&20, the results would look much worse. Admittedly the risk-adjusted return is better.

    4) 2 & 20 is obscene in almost all cases. 1.5 and 0 seems, to me to be a fair fee structure for this kind of performance. But, these are example systems; designed to be as simple as possible with no optimization or cute stuff…to show that trend following is a viable approach to stock market participation.

    5) Don’t know, can’t model that accurately since there is no way to know what stocks would have been shortable in the past. Brokers tend to run out of inventory at regulators make short selling illegal at the most inconvenient times. In real life we do short, just not with individual stocks.

  15. Robert,

    I don’t understand your question:

    “What is the performance difference for 3-year highs and all-time highs, assuming the same exit strategy and money management principles?”

    There were no 3 year or all time highs in this paper. If you are referencing our previous paper, where we used all time highs and a 10 ATR stop…in that case the difference between 3 years and all time was very little. You wouldn’t be able to tell one equity curve from the other on a chart. Things don’t start to deteriorate untile you go faster than 9 month highs. Then the turnover goes up, transaction costs go up, volatility goes up…you get the point…all the bad stuff goes up. Anything slower than 1 year highs worked about equally well.

  16. Jamal,

    “Ken, how would use leverage on some of these stocks? The stocks were from a list of 3.000. Do you know how many non marginable securities that list probably featured? You couldn’t have used leverage even if you wanted to.”

    Every one of them was/is marginable. They are all members of the Russell 3000. We trade them every day and have for years.

    Also, with a portfolio margin account, available from any broker worth trading at, your min margin on a diversified portfolio of stocks is approx 15%. Not that we’d ever even come close to using that kind of leverage, it is doable.

  17. Eric,

    Thank you for the quick reply.

    I believe you did an outstanding job presenting your analysis in both of your trend following papers! It gets to the essence of trend following and should be a must read for anyone serious in trading stocks.

    A few questions:
    1) Which exit do you consider better (based on the higher overall expectancy of the system), yearly lows with 10 ATR trailing stop or yearly lows with 18 month low exit?
    2) How would 3-year highs or all-time highs compare to the other entries in this paper?
    3) Would it make sense to set any market cap filters? I.e. AAPL is now setting a new all-time high, but it’s almost 200Bn market cap and is unlikely to be a ten bagger from these levels?
    3) How is your performance YTD? 2009 has been a tough year for us trend followers…

    Thank you,
    Robert

  18. Robert,

    For the most part, one exit is not better than another; they simply have different winning percentages and payoff ratios and consider risk from different perspectives. The ATR stop treats day over day volatility as risk. The 18 month low treats aggregate directional movement as risk. Different market conditions will favor one method over the other.

    For the most part, one entry is not better than other; faster entries get in earlier but suffer from higher whipsaw risk and often opportunity cost (by diluting other positions). Again, different market conditions will reward one more than the other.

    Intuitively the market cap filter makes sense. But it’s dangerous. How would you have felt about Cisco, Microsoft, GE, etc. in 1995? Also, there is no way to know what kind of paradigm awaits us in the future. It might be that ONLY the mega-cap stocks work. There’s no law against that.

    No, 2009 has not been a profitable year for trend following on stocks…like Kansas we’ve been flat and boring.

  19. Jamal,

    “So were you already accounting for leverage within the study then?”

    The answer to your question is on the first page of the paper. You also might try reading the last paragraph of the appendix.

  20. Hi Eric – Great work you’ve done on the two papers. Its very generous of you to make these publicly available. I’m in the process of using them as a starting point to develop my own long-only stock TF system. What I did notice in the details of the first paper is that the equity curves in your simulation reflect the application of your proprietary portfolio management system and you go on to enumerate the characteristics. So the real value of your work to me is that you establish the expectation of a long-only stock TF system, some reasonable entry and exit parameters, and stock selection criteria. That certainly eliminates millions of possibilities that simply don’t work.

Comments are closed.