We often hear this proverb in the market. When the price of a security is driven by fear, the majority of the investors/traders tends to make irrational investment decisions. As a trader we can gain from this opportunity but how to measure fear in the market? One way is to look at the volatility Index(India VIX) but it can be a very subjective interpretation. So searching for an answer, I ended up reading “Buy the fear and Sell the greed” by Larry Connors. What Larry is suggesting is to look at the shorter period Relative Strength Index(RSI) of security or index to find fear and hence irrational behavior of market participants. So, let’s take this approach with slight variation and test it on BANKNIFTY.

Here are some assumptions and the way of running test:

- Only closing price has been considered in this research
- Test period is approximately 15 year i.e., from Jan, 2005 to Aug, 2020
- All Historical data was taken from NSE’s website

Let’s figure out what happens when there is extreme fear in the markets. Below charts show total percentage returns for the next 2, 3 and 4 days for different ranges of 4-period RSI. X-axis shows different buckets of RSI(4) and Y-axis shows the total percentage return of BANKNIFTY for the next 2, 3 and 4 days.

As above charts suggest historically BANKNIFTY often bounced back from the oversold zones. So, the proverb **Buy the fear and sell the greed** is valid as data is also suggesting the same. Now, what about taking a 6-period RSI instead of a 4-period RSI?

Below chart is the same as above but RSI period is set to 6 instead of 4.

Results are not encouraging compared to the initial one. One explanation can be given that, “**Over the days a widely traded index tends to return to the mean**”. So, now we will only use 4-period RSI moving ahead.

Let’s say going long on the closing as soon as BANKNIFTY hits particular RSI(4) readings and closing the trade at the close after 2, 3 or 4 days. Below is Profit-factor (Net Percentage Profit/ Net Percentage Loss) of this simple vanilla idea.

According to the historical data, the best combination we can have is RSI(4) < 20 and carrying trade to exactly 4 trading days. This combination has an average return around 0.7% with a 1.55 profit factor over the last 15-years.

Above strategy is just a plain vanilla implantation of our idea of Buying fear. Now as we technical analysts are always looking for trends in the market let’s add some trend filters in it.

Instead of going long at every oversold zone, what if we go long if and only if BANKNIFTY is above its 200-day simple moving average on the closing basis? Here is the profit-factor for the same.

As we can see, the performance of the strategy significantly increased by adding a 200-day moving average as a trend filter. Performance can be increased further by adding additional constraint in the plane vanilla strategy like fix rupee/percentage stop-loss after taking trade.

## Conclusion

We can conclude that, historically large edge remained by buying the fear. Furthermore Shorter period RSI have more predictive ability than longer period RSI. As this research is based on the behavior of market participants rather than data-fitting on the large enough dataset, I think a positive edge will also remain intact in the future also. Strategies can be created around this historical edge to profit from this irrational behavior from Market participants.

Until then, Happy Trading!

Aren’t the RSI(6) result better than RSI(4) in the above graphs <10 and <20 range?

Pls explain as its mentioned the opposite by you?

NO, it is calculated based on total cum. return generated by particular criteria. and now i have also updated graphs of cum. return instead of avg. return that were there before.

Hope this clarify.

The profit factor of RSI 4 < 10 is 1.61 for 4days, which is better than 1.55 of RSI 4 < 20 for 4 days, but you have mentioned the later as the best combination, why so?

also above 200MA results also shows RSI 4 < 10 gives a better 4 days result.

Yes, it is as you mentioned. But here for the simplicity I have only shared profit factor graphs. if you dive deeper in the other factor like total return generated and total numbers of trades criteria I have mentioned will outperform.