By far, the most popular way to trade mean reversion is using basic technical tools and building a rules-based system around them.
The signals given by each system are pretty correlated, so it doesn’t matter too much which tool you choose to use, but instead how you apply it and manage risk.
A linear regression line will fit the best line between both the starting and ending points.
Without getting into the math behind it, think of it like a trendline that cuts prices. It’s the trajectory a security took to get from point A to point B.
See the chart of Goodyear (GT) below as an example:
Above is a 90-day linear regression line. The line represents the best fit between point A (90 periods from today) and point B (today), which represents basically a sideways line.
This just tells us that on balance, Goodyear has been range-bound over the last 90 days of trading.
A typical mean reversion trading tactic is to plot standard deviations around the linear regression line. The security is considered overbought when near the upper range and oversold when near the lower range. See the below chart as an example:
Trading bands like Keltner Channels and Bollinger Bands are some of the most commonly applied technical indicators for mean reversion strategies.
These are generally useful when the price action inside the bands is considered ‘normal’, and action outside the bands is considered ‘abnormal.’
Just as when using linear regression, significant deviations from the trading bands present fading opportunities to take the opposite side of the trend.
Here’s an example chart of Starbucks (SBUX) with Keltner Channels. Shorter-term traders are likely to find these more useful because they’re quick to adapt to recent prices.
ConnorsRSI is a modified short-term RSI developed by Connors Research. It’s a combination of three different indicators:
- A 3-period RSI
- An RSI applied to the current up/down streak of the market (i.e., 5 consecutive positive closes)
- Rate of change
Each of these factors is equally weighted to form a sort of adjusted-RSI that not only identifies short-term overbought and overbought levels but also weighs the levels against factors that point the odds further in your favor.
The indicator’s calculation gives readings stronger conviction when the rate of change and up/down streaks are at extremes. These factors provide short-term mean reversion trades a higher expectancy.
Here’s the formula for calculating the indicator, taken from the Connors RSI Guidebook:
ConnorsRSI(3,2,100) = [ RSI(Close,3) + RSI(Streak,2) + PercentRank(100) ] / 3
The basic rules outlined by Connors Research are as follows:
- Buy when the Connors RSI < 20 (or 15, 10, etc.)
- Close the position when the stock closes above the 5-day moving average or when the Connors RSI is above 65.
Here’s an example of what a long setup for ConnorsRSI strategies might be:
There’s been a wealth of quantitative work done on this indicator, much of which can be publicly found on sites like Quantopian and the like. Below is a graph showing the 5-day returns of different ConnorsRSI readings.
As expected, the extreme oversold conditions present the most favorable opportunities, likely because the proverbial ‘rubber-band’ is stretched so far that, even if the market continues to decline, it still has to bounce in the opposite direction, at least in the short-term:
Mean reversion trading is psychologically difficult.
You’re buying falling knives, which may continue to fall precipitously, without a bounce, while your profit target is small. Worse, most mean reversion traders don’t typically use stop losses because they negatively affect trade expectancy.
By all accounts, learning trend-following or momentum is a much more comfortable and psychologically pleasing way to trade. You’re on the right side of the trend, often have a reasonably tight stop, and your winners tend to be much larger than your losing trades.
However, mean reversion trading can be highly rewarding precisely because it’s difficult and a less “crowded trade.”