About the indicator failure swings you can read in, for instance, the book by T. Bulkowski: “Trading Basics. Evolution of a trader”.
At first, let us have a look at the graph below that shows a stock price together with the RSI-index (14-day):
What we see is that the index forms a sort of “double bottom” (denoted by the circle), where the right bottom is above the left one. The pattern is also quite small so it lasts only a few days. It is also important that it occurs in the oversold region, i.e. the value of RSI is low.
Backtesting is perhaps not very straightforward because the size of the pattern is not so well defined. Therefore, the results shown below may not be 100% correct, e.g. the algorithm used for backtesting may not find some patterns or so.
In the testing, I assumed that the oversold region begins at RSI equal to 30. We enter a trade “today” if the value of RSI “yesterday” is greater than “the day before yesterday”, as well RSI on “the day before yesterday” is in the right bottom. In addition, we assume that the low of the day that corresponds to right bottom in RSI is also higher than the low of the day that corresponds to the left bottom in RSI (so not as shown in the figure above where we can see the opposite). This ensures that the uptrend has started (in addition, my backtesting confirms that this improves the performance).
We do not use any stop-losses, i.e. this is not a real trading strategy. The objective is rather to show a general performance of the pattern.
We will assume two types of exits: (i) leave after 10 days; (ii) leave when RSI becomes 70 (note that the second exit will result in trades with a longer holding time).
At first, we will test stocks that are currently in S&P 500 index. Here are some selected results (using NinjaTrader software using data from Kinetick):
|Number of trades||656|
The results for the Russell 3000 index are as follows:
|Number of trades||1850|
The results are positive and even the maximum drawdown is acceptable. Nevertheless, the performance is not so good: a relatively low percent of profitable trades combined with a low win/loss ratio.
Now, let us change the exit rules: we will leave a trade when the RSI index becomes 70, i.e. it enters the overbought zone. The results for S&P 500 stocks are as follows:
|Number of trades||600|
|Number of days held||275|
And for Russell3000 stocks:
|Number of trades||1715|
|Number of days held||264|
The results are different: the percent of profitable trades has improved (also due to the longer holding time) but the maximum drawdown is worse. Generally, we can conclude that the strategy is profitable. Nevertheless, as mentioned above, it would be safer to combine it with another strategy to increase the odds.