Premium Market Analysis, Trader Education, Software, and Trading Strategies. Thirty Years Of Skin In The Game

Trend following

Trend-Following and Mean-Reversion are Complementary

Photo by Tima Miroshnichenko

Marketing frequently serves as the driving force behind efforts to promote one trading style over another. Mean reversion and trend following are complementary trading styles, and choosing one over the other is a false dichotomy.

Trend-following gets rid of losing trades quickly and rides the winners. Trend-following, also referred to as a divergent style of trading, aims at catching outlier trades that may stay open for months or even years. As a result, the trade profit/loss distribution has a positive skew, but at the expense of a win rate of less than 50% or even lower than 40%. Since the payoff ratio is much higher than 1, the right strategies are usually profitable over time.

Mean-reversion, also known as a convergent style of trading, closes winning positions quickly to secure profits. The win rate is high, and although the payoff ratio tends to be lower than 1, the right strategies are profitable. The profit/loss distribution has a negative skew. Usually, trend-following proponents argue that negative skew is a disadvantage, ignoring the fact that positive skew requires extreme discipline to ride long trends.

If, for a moment, we focus on the skew, as it turns out, there is an interesting fact: while the profit/loss distribution of trend-following strategies has a positive skew and the reverse is true for man-reversion strategies, the equity curve changes tend to have a negative skew, and the reverse is true for mean-reversion strategies.

More importantly, what is purported to be a major advantage of trend-following, i.e., the positive skew, can also be a disadvantage during long periods, such as the “lost decade,” when it cannot be realized. More on this below. See this article for some more details and examples.

Marketing is frequently the driving force behind trend-following proponents’ arguments against mean reversion based on negative skew. As it turns out, trend-following and mean-reversion are complementary styles, as we will demonstrate with a few examples below.


SG Trend Index

The SG CTA Trend Index is an ensemble of the performance of the 10 largest trend-following CTAs. Below is the buy-and-hold hypothetical performance from January 3, 2000, to September 25, 2023.


The chart marks the “lost decade,” from 2009 to 2018. During that period, the SG Trend Index was flat. Although all the CTA programs comprising the index had a high positive skew, the skew of the daily equity changes is -0.56. The “lost decade” is a period during which, on average, the top 10 CTAs had flat performance, although due to high dispersion, some had positive returns and some negative.

Trend-following Strategy

Next, we consider a specific strategy based on our PSI5 algorithm in divergent mode. The strategy trades 23 futures contracts. The backtest is from January 3, 2000, to May 16, 2024. This particular version of the strategy employs dynamic stops and trailing exits.


Although the P/L skew is highly positive (not shown), the daily equity change skew is -0.35. From mid-2015 to mid-2019, the strategy remained flat. Although the annualized return is more than 15% and the Sharpe is 0.82, extended periods of flat or even negative performance are always possible due to the market’s refusal to provide outlier trades.

Issues with trend-following

Although it may sound like a free lunch, trend-following is far from one. Besides the requirement for disciplined execution, long periods of underperformance are problematic. They can cause large reductions in the AUM of trend-following funds, especially when equities are in bull markets, as was the case during the “lost decade.” This is one reason the alternative space of trend-following has a combined AUM of about $300 billion, versus more than $15 trillion for passive equity index investing. Allocators hesitate to increase their exposure to trend-following because of the risk of long periods of flat performance. Many trend-followers blame the allocators for “not realizing” their “convexity” benefits, also known as “crisis alpha,” but underestimate their sophistication. Trend-followers have to face the reality of the markets and the fact that trends and trend-following are not equivalent terms; trend-following cannot capture outliers where there are none.

Some trend followers have responded by increasing diversification. They hope that investing in hundreds of markets will eventually catch outlier trades. Evidently, this is primarily a miscalculation, as an increase in the number of markets dilutes the gains from outliers due to reduced exposure, ceteris paribus. At the limit of exposure, these strategies turn out to be similar to capital preservation schemes that charge customers a fee. Trend-followers often use convoluted arguments and a few have used borderline-questionable backtests to convince allocators about the benefits of high diversification in vain. While it is clear that higher diversification and more capital can potentially yield higher returns, a reduction in AUM due to withdrawals, for instance, can trap the strategy with a large number of open trades, necessitating difficult decisions about closing positions or reducing exposure.


Although trend-following has a simpler interpretation, mean-reversion does not because it encompasses a wide range of strategies.

  1. Reversion to an indicator (for example, a moving average).
  2. Reversion to a price level.
  3. Reversion from overbought to oversold and the other way.
  4. Reversion due to negative autocorrelation (daily timeframe).
  5. Statistical arbitrage and pair trading.
  6. Long-term fundamental reversion.

Our algorithms attempt to exploit #4, i.e., reversion due to negative autocorrelation in daily returns. When traders refer to mean-reversion without explicitly defining what they mean, this often demonstrates confusion. Since negative autocorrelation in daily returns has been the predominant dynamics in equities since the late 1990s, when there was a regime change from high positive autocorrelation to negative and recently highly negative, we only apply our mean-reversion strategies to that market.

Long-only Dow-30 stocks with a mean reversion strategy

A backtest of the B2S2 mean-reversion strategy with Dow-30 stocks is shown below.


Note the high annualized return of 17.3% and a Sharpe ratio of 1 from 01/02/1993 to 05/16/2024. Also, note that the skew of the equity curve’s daily changes is positive (+0.50).

The problem with mean reversion strategies is the high turnover and the sensitivity to transaction costs and slippage. The impact on the annualized return in actual trading can vary from 2% to 5%, depending on slippage and trading costs. Still, there is potential for mean reversion to provide “trend-following crisis alpha” and convexity. The example below shows the application of the PSI5 algorithm in convergent mode to the SPY ETF from its inception to May 16, 2024.


Even after adjusting for slippage, mean reversion has the potential to improve trend-following returns. In this article, we discuss an example of combining the trend-following and the mean-reversion strategies based on the PSI5 algorithm in divergent and convergent modes, respectively. We also show that the two strategies have been uncorrelated on average.


Some authors and fund managers claim that trend-following provides “crisis alpha,” i.e., positive returns during periods of negative equity market performance. But what provides “crisis alpha” when trend-following has flat or even negative performance for extended periods of time? The solution is the use of mean reversion strategies. It is sensible, then, to attempt to combine these two predominant styles of trading and ignore the voices that attempt to enforce a false dichotomy for marketing purposes. An equal allocation to trend-following and mean-reversion has the potential to offer a high Sharpe ratio and leveraged alpha.

We also believe that adding equities to trend-following is the wrong approach, as the correlation with equity markets increases and convexity benefits may diminish. When you use classical trend-following with futures and equity market mean-reversion together, you can keep the convexity and have a low correlation to the performance of passive equity indexes.

Last but not least, any attempts to argue in favor of trend-following based on high skew show a misunderstanding of statistics and how they apply to trading strategies. A high positive skew at the trade profit/loss level could lead to long periods of flat performance or even underperformance. High skew is far from a panacea; it is both a blessing and a curse. Interestingly, at the equity curve change level, trend-following strategies tend to have a negative skew, and mean-reversion strategies tend to have a positive skew. If equity growth is the objective, as it should be, then combining the two strategies makes sense. However, both trend-following and mean-reversion strategies are hard to trade, and the risk of an uncle point or even a total loss is high. If you are not sure you understand the details, seeking help from an unbiased registered adviser could help. We wrote this article for informational purposes only.

Trend-Following Subscription

Access to premium articles on trend-following and the rules of a strategy with a Trend-Following subscription.



Disclaimer: No part of the analysis in this blog constitutes a trade recommendation. The past performance of any trading system or methodology is not necessarily indicative of future results. Read the full disclaimer here.

Charting and backtesting program: Amibroker. Data provider: Norgate Data

If you found this article interesting, you may follow this blog via RSS or Email, or on X.