Mean Reversion Trading | Tips & Strategy
We share a lot of trade ideas and strategy here at The Trade Risk but it’s primarily from the point of view of momentum, breakouts, and following recent price movement.
Mean reversion trading is a style that relies on price action to look much different (in most cases the exact opposite) as momentum, and it’s something we haven’t written a whole lot about on this site.
Instead of digging into this topic myself, I thought it would be best to have a friend of mine, with over 15 years of experience building and trading strategies with a focus on mean reversion, help us with some tips and best practices.
I chat with Cesar Alvarez, a trader who spent 9 years as a professional market researcher for Larry Connors at Connors Research and then went on to start his own business, Alvarez Quant Trading, to share ideas and build strategies.
Please enjoy the following Q&A with Cesar on mean reversion trading:
How long have you been trading mean reversion strategies?
I have been trading mean reversion strategies since 2003. The primary focus has been on equities and ETFs. I would say on average I have been trading two to three mean reversion strategies at the same time. The reason for multiple is that they trade different universes and look for different amounts of oversold/overbought. Right now, I am trading a long S&P100 strategy with shallow pullbacks, a long Russell 3000 strategy with deep pullbacks, and a short ETF strategy.
How do you define mean reversion trading? What does it look like? Why does it work?
Looking at a chart, a mean reversion setup is a stock in an uptrend and has had a recent sell-off.
Normally I define an uptrend as an upsloping 200-day moving average and the stock being above it. Then I want to see the stock sell-off towards the 200-day moving average. Later, I will cover some of the technical indicators that I use to determine a sell-off. Entry is either on the next open or waiting for a further sell-off intraday, thus entering on a limit order. The limit order is normally between 2%-10% below the previous close. Once you get in, we wait for a bounce before exiting.
Below is a recent trade of mine. We can see the MA200 rising and the stock above it. The large vertical line is where I got in. A few days later it had a large bounce, which is when I got out.
Here, we have another recent trade. We have another stock above the MA200 and then a strong sell-off. Entry is the up green arrow. Unfortunately, this one continued to sell off, which is the main danger of mean reversion trading.
The reasoning behind mean reversion trades is that people have overreacted and pushed the stock price too far. I want to get in before the correction to the overreaction happens. This can be hard to do but done correctly the win rate for mean reversion trades is usually between 65% and 70%.
What are some of your favorite indicators to help determine when a stock is “stretched” and likely to retrace?
I find that there is a very high correlation between a lot of the indicators used to find a mean reversion trades. The most common ones that I have used are:
- Relative Strength Indicator (RSI)
- %b (Bollinger Bands)
- Moving average stretch
- Rate of change
- The number of days down
The ones I tend to use the most are RSI, rate of change, and down days. Combining two of them together will lead to very good setups.
As to specific values to use, it really depends on your trading universe and how frequently you want signals. For me, using a 2-day RSI with a value under 5, and the stock is down 3+ days, produces good setups. Another good combination is stock being down 10%+ over the last 3 days and the stock down 3+ days.
It is easy for some indicators to signal a trade because of one large down day. That is why I like using two mean reversion indicators.
One reason why breakout trading or trend trading has always made so much sense to me is that there’s always a very clear exit point. If the trend reverses, you exit. If the breakout fails, you exit.
With mean reversion, it seems harder to know when you should throw in the towel. If a stock becomes stretched and we enter for reversion, but it continues to stretch away, it’s presumably even more ripe for a reaction to the mean.
How do you draw the line? Are traditional stop losses based on price levels appropriate or are there other techniques to manage risk while keeping you in the trade?
This is the hardest part of mean reversion trading. There is no clear exit on the downside if a bounce does not happen. A stock that keeps selling off is likely an even better trade. I have written and spoken extensively that adding stop losses to mean reversion trades tends to have significant degradation on your trade return. There is a multitude of ways of dealing with a stock that keeps selling off.
The first is before trade entry.
For example, only look to enter a stock on the day RSI2 crosses under 5, instead of simply looking for any value under 5. Or look for the 5-day return to be less than 10% but not greater than 20% down.
This will help prevent you from getting into trades that the rubber band has broken. See this post for more about the broken rubber band, Mean Reversion and the Broken Rubber Band.
What if we are already in the trade? The best thing from a pure trade statistical point of view is to wait for the bounce. The problem is that it can be hard to wait for that bounce. So if adding a stop loss does not help and waiting for a bounce is hard, then what?
A reasonable tradeoff is to include a rule that gets us out of losing trades after 10 days of being involved with it. This time-based stop happens without consideration of where the price is. Even though this does reduce the returns, it does so less than a price stop.
The danger is that because there are no stop losses, especially when trading smaller cap stocks, you will one day wake up to a stock that has gapped down 50% or more.
Because of this, I control my risk through position sizing.
I size all positions the same size, no matter the volatility. If I am trading $50,000 portfolio with a maximum of 10 positions, then I trade $5,000 per position. I then ask myself, could I handle a 50% loss in a single position. If the answer is no, then I reduce the amount allocated to the portfolio. Go from $50,000 to $30,000. Again, I ask can I handle a 50% in $3,000 position? If yes, then I have found the sizing for the portfolio.
The time where is it OK to use a stop loss is when it is large and it is there to keep you trading the strategy. You may know that after a stock loses 20%, you are more likely to make a discretionary decision. As we know, making a discretionary decision in the heat of the moment will likely lead to bad decisions. Therefore, adding a stop loss rule at 20% may be a good decision even though it will reduce returns.
Are all stocks and instruments equally good mean reversion candidates or should there be a certain level of volatility, market cap, etc. in ideal setups?
If you are looking for high returns, then you want stocks with high volatility, low price, and low liquidity.
I normally look for 100-day historical volatility over 40. The higher the better.
I also look for stocks that are above a significant moving average or near highs. My most common method is the stock is above the 200-day moving average. The reason for this is that it keeps you out the bigger losers.
Trading under the 200-day moving average is profitable but you have to deal with a lot more volatility. When I do trade below the 200-day moving average, I will change my rules to look for even more stretch setup before entering.
What are your thoughts on waiting for a stock to begin reversing versus buying while it’s at an extreme making new lows (or highs). Do you recommend waiting for some type of reversal confirmation?
This is one of those ideas that seems like it should work. One of the first things I tested 16 years ago when I got into quantitative trading was this idea.
It failed miserably.
It took a strategy from making money to losing money. In waiting for the confirmation, you are giving up a good portion of your profits. Also, I think that the exit then needs to be changed to some sort of trailing stop. This is something I should revisit and see if things have changed or not.
What is different about trading short with mean reversion?
People often think that shorting is simply reversing the rules from the long side. This is only partially correct. The main difference is shorts are better over the 200-day moving average than below.
Another important step is to make the stretch even more. You want the stock to be really over-bought before jumping into a short. These charts look like rockets and can be mentally really hard to short. Below is the chart of a recent signal from an old short strategy that I stopped trading about three years ago. This stock had nearly doubled over two months.
For example, if you had a rule on the long side that the stock’s RSI2 was under 5, for the short side you would then want the RSI2 to be above 99. Instead of using two mean reversion indicators, you may want to use three.
You may be wondering, did the strategy stop working and that is why I do not trade it? No, it is still doing fine. It is up 19% this year as of 10/31/2017. Not bad for a strong bull market.
The problem is once every few years you get hit with a stock that opens up huge. That wipes out months of profits and is very psychologically hard to deal with. Here is a stock that I entered short in the last bar.
Then a few days later, this happens.
One needs to be prepared for this to happen.
Any final thoughts or common mistakes you see traders making when it comes to building mean reversion strategies?
One of the most common mistakes I see traders make when making any strategy is trying to make a strategy that works in all market environments. Most of my strategies have a market regime focus, like a bull market or a bear market. A simple 200-day moving average works well for this.
Another common mistake is trying to add too many rules. This is normally done to try and avoid a bad trade or two. This is overfitting and should be avoided, but this issue can also come up accidentally.
Say that you have run an optimization to determine what is the best RSI value to enter on. This is a normal part of the backtesting process. But maybe the value you like avoided a bad trade. In order to keep an eye on this, I always output the worst trade from any backtest. This easily allows me to see if a variation is simply better because it missed a bad trade.
Mean reversion trading | tips & strategy – my takeaways
The first comment Cesar made that really resonates with me is on time-based stop losses. I really like his implementation of a 10-day exit regardless of how the trade performed.
Time-based stops don’t get the attention they deserve but they’re a very clean way to maximize the opportunity cost of tying up capital in a stale trade.
Something else that hits home is that Cesar looks for stocks in uptrends that are selling off towards an upward sloping 200-day moving average.
This same methodology is something I look for in my own setups for trading pullbacks and reversals. The difference is that I look for pullbacks to form at a higher point in the trend, a 20-day moving average versus the 200-day.
Finally, I love the advice he shares at the end, warning traders that try to create a trading strategy that fits in all market environments. I’ve written about this in many different forms, and I couldn’t agree more. See this post, adapt to dynamic markets or fail.
I’m a huge fan of starting each day understanding the market context and the environment we’re in before even thinking about pushing the buy or sell button. See this post on using market internals as a regime filter.
Hopefully, this Q&A has been helpful, I know I enjoyed it.
If you have questions or comments for Cesar leave them below. If you’re looking for more articles like this, be sure to check out our trading learning center.
Enjoy what you read? Share it below and be sure to tag @thetraderisk.