Last Updated on 10 February, 2024 by Trading System

Originally a statistical term, mean reversion is now widely used in the financial trading world to indicate a specific approach to trading. In this post, we want to discuss why mean reversion works and which markets are best for mean reversion trading strategies, and then present 3 mean-reversion trading strategies/ but first, let’s find out what mean reversion means.

What does mean reversion in trading?

Mean reversion means the tendency of a dataset to revert toward the mean after moving significantly away from it. In trading, it is the tendency of the price of an asset to gravitate over and under a moving average. After a significant rise, the price tends to revert down and after a significant decline, it reverts upward.

The chart below shows a good example of mean reversion:

The blue line is simply a trend line draw randomly from the bottom to the upper right corner to represent the ascending mean price. As you can see, the price of SPY moves in waves as it ascends and never goes too long in either direction before it “reverts”.

Also, during the bear market of 2008/09, there were sharp reversals, even though the trend is down. To put it simply, the market never goes in one direction without breaks; it’s like ebb and flow.:

Some mean reversion trading strategies

While many trading strategies can be labeled mean reversion, the most popular ones are based on the following:

• Overbought and oversold indicators, such as RSI, MACD, Stochastic, Bollinger Band, etc.
• Reversal patterns
• Price pullbacks

Mean reversion vs. trend following

The mean reversion strategy is the opposite of trend-following strategies. While mean reversion normally has a high win ratio with many small winners and a few big losers, trend following gives many losers and a few big winners.

Thus, the trade distribution is different: mean reversion has more left tail losers, while trend following has more right tail winners.

Which markets show mean reversion, and why does mean reversion work?

From our experience, mean reversion strategies work best in stocks. It doesn’t perform as well in other markets, such as the commodity. But why does that happen? We know that mean reversion in the stock market didn’t work so well before the 1990s but has been performing since the 2000s. One of the reasons could be the rise in futures trading with the associated arbitrage between stocks and the futures contract.

Another reason could be the effects of profit-taking and short-sellers who sell strength and value investors who buy stocks that have fallen significantly. For instance, when a stock goes up in value, many are tempted to sell to realize some gains while others might want to short, creating selling pressure.

On the other hand, when a stock falls in value, more buyers are willing to buy than sellers are willing to sell, leading to a bullish price reversal. Moreover, those who are short also try to cover their shorts —this might be the reason for many of the rallies we see in bear markets.

Mean reversion work best in a bear market

From experience, we have learned that mean reversion in the stock market works best during a bear market. This might sound illogical, but the reason is simple: increased volatility. Even more surprising is the fact that long positions work better than short ones!

For example, in 2008/09 the market fell over 50%, but we made the most money on the long side from our day trading using a mean-reversion strategy. The fast decline in price during a bear market creates opportunities, unlike in a bull market that rises slowly over time and spends significantly more time above the 200-day moving average.

Another point is that bear-market rallies are very explosive. For example, even though the market lost about 50% of its value from May 2008 until early March 2009, there were significant rallies and up days. Here are the numbers:

• About 99 up days and 104 down days
• The average up day was 1.79%, which is not far below the average down day at minus 2.32%.
• There were 51 days with a rise >1%, 30 days with a rise >2%, 76 days with losses >1%, and 45 days with losses >2%.

Thus, while the market fell, we still had many explosive days on the upside. Shorting was difficult, but there were opportunities for buying weakness and selling strength, which was very profitable.

Of course, short-term trading needs prey, and it comes in the form of volatility. See the graph below, which shows the 25-day moving average of the absolute values in the daily changes from close to close:

Now, let’s take a look at 3 mean-reversion strategies:

1. Deviation from a recent high in XLP

We use this strategy is in the S&P 500 consumer stable sector ETF, which has the ticker code XLP. Here are the criteria for the strategy:

1. Find the average of the H-L over the last 25 days.
2. Get the (C-L)/(H-L) ratio every day (IBS).
3. With the average from point number 1, calculate a band 2.25 times lower than the high over the last 25 days.
4. Go long at the close, if XLP closes under the band in number 3 and point 2 (IBS) has a lower value than 0.6.
5. Exit when the close is higher than yesterday’s high.

The equity curve below shows the returns from 100 000 invested at inception in 2002 and compounded until the summer of 2021 — the graph on the bottom is the drawdown:

These are the result:

• 453 trades
• 37% average gain per trade
• CAGR of 7.7% (buy and hold 7.7%)
• 32% of the time was spent in the market
• 20% max drawdown
• A profit factor of 1.8

2. An IBS short strategy in FXI

The Chinese ETF with the ticker code FXI has shown strong mean reversion tendencies over the last decade. We used the strategy below on it, and it worked pretty well. These are the parameters of the strategy:

1. Go short at the close if today’s IBS (C-L)/(H-L) is higher than 0.9.
2. Exit at the close when the IBS is 0.25 or lower.

Below is the equity curve for 100 000 shorted in 2010 and reinvested until the spring of 2021 (the graph on the bottom is the drawdown):

The result is as follows:

• 224 trades made
• The average gain per trade was 0.63%
• CAGR was 12.4% compared to buy and hold that gave 3.3%
• Time spent in the market was only 34%
• The maximum drawdown was 12%
• The profit factor was 1.75

3. 5-day low in the S&P 500

This strategy was implemented on the S&P 500 Index. The criteria are as follows:

1. Go long at the close if today’s close is below yesterday’s five-day low.
2. Sell at the close when the two-day RSI closes above 50.
3. There is a time stop of five days if the sell criterium is not triggered.

The equity curve below is for 100 000 compounded from 1993 until July 2021 has produced this equity curve — the graph on the bottom is the drawdown:

The trading results are as follows:

• 448 trades
• 52% average gain per trade
• 8% CAGR (buy and hold 10.4%)
• Only 16% of the time spent in the market
• 20% max drawdown
• Profit factor of 1.8

Read more similar articles here on The Robust Trader or on Quantified Strategies