RSI is mostly used as a reversal or contrarian indicator. In most stock markets, this has worked really well over the last 30 years. But what if we flip it and use it as a momentum indicator instead to make a RSI momentum strategy? Can we make money that way?
We can use RSI as a momentum indicator by buying on rising RSI values and selling on falling RSI values. This article explains how to do it and how to backtest RSI as a momentum indicator. Unfortunately, RSI momentum doesn’t seem to work on stock indices.
What is the RSI?
We need an understanding of the RSI indicator to understand how to use it. We have written about it previously in a separate article:
Also, there are at least two other ways of measuring relative strength in addition to the original RSI linked above (but we prefer the original one):
What is momentum in stocks?
Before we go on we need to define momentum. Momentum is based on inertia – the tendency for an object to do nothing or to remain unchanged (in physics). This includes mostly the direction of the stock or asset, but also the velocity of the movement.
Put short, if a stock has risen over a certain period, it’s more likely to continue in the same direction than to reverse the momentum. It’s kind of a trend-following strategy and the opposite of a mean reversion strategy.
How does this work in practice? A lot of research has shown that there is such a momentum premium in the stock market, but it only exists for relatively short periods of time. In general, momentum over the last 3-12 months predicts the best results in the coming 3-12 months. (We also go into depth about this an other strategies in our quantified investment course.)
For example, if the S&P 500 rises in one month, you go long at the close and hold it for one month. At the end of the next month, you look at the performance again and stay long if the performance is positive or you sell if the trend was negative. Check out these two trading strategies as examples:
How does RSI work as a momentum indicator?
RSI is mainly used as a mean oscillating indicator: we go long at low readings and sell (or go short) at high readings. These are classically oversold (what happens when markets are oversold?) and overbought (what happens when markets are overbought?) readings. For stocks, this has worked very well over the last three decades, like these strategies (for example):
If RSI works well as a mean reversion indicator, could it possibly work as a momentum indicator, which is basically the opposite strategy? It sounds contradictory, but it might be worth a try. Also, many readers might know that there are other markets/asset classes than stocks. If a strategy works well on stocks, it’s not logical that it should work on all markets. No markets are the same!
RSI momentum strategy (backtest)
According to many articles and books, the RSI indicator works pretty well as a momentum indicator. But as usual, this is mostly anecdotal evidence with very little quantified backtesting backing their claims. We strongly recommend to backtest all your trading ideas and strategies before you commit money (if you are new to backtesting you might be interested in our backtesting course).
Let’s go on to backtest a RSI momentum strategy:
We backtest the S&P 500 (SPY) from inception in 1993 until today. When testing for momentum we do the opposite of mean reversion: we buy when the RSI is rising and sell when the RSI is falling.
We backtest the following RSI momentum strategy by using optimization (read here for how to optimize a trading strategy):
- Parameter 1: the # of days to use in the RSI-formula (the number of days from 5 to 50 with intervals of 5)
- Parameter 2: the threshold of entry (from 50 to 75 with intervals of 5)
- Parameter 3: the threshold of exit (from 75 to 50 with intervals of 5)
This optimization requires 360 tests (10 x 6x 6), quickly done in Amibroker.
We can’t display all the results because of lack of space but here is an excerpt from the backtest results:
Column 3 to 5 shows the three parameters we used in the optimization of our RSI momentum strategy. To balance a decent amount of trades and good results, it seems like the best results is around 15-to 25 days for the RSI. However, the profit factor (what is a good profit factor in trading?) is often all over the place by small changes in the variables, meaning that the element of randomness is huge.
Conclusion: Does RSI momentum strategy work?
It doesn’t seem that RSI momentum strategies work very well on daily bars in the S&P 500. We also tried our RSI momentum strategy with weekly and monthly bars but to no avail. We even tried on other stock/indices ETFs but we find little evidence that is a viable trading strategy.