In this strategy, we will have a look at two correlated markets. For instance, we buy a stock when it becomes “cheap” with respect to an index and sell when it becomes “expensive”.
In my backtesting, I compared stocks with the Dow Jones index. I selected only stocks that have been included in this index since the beginning of year 2000 (that is, 20 stocks) and backtested the strategy in the period 1 January 2000 – 31 December 2017. Instead of using the index as such, I looked at the price of DIA (which is an ETF of this index).
At first, let us calculate the ratio of the daily closing price of a stock, Ps, to the price of DIA, Pd. In other words we compute Ps/Pd. Afterwards we construct Bollinger Bands (50,2) of this ratio. We buy a stock as soon as its Ps/Pd closes below the lower band. We sell the stock as soon as Ps/Pd will reach the upper band.
Of course, there are other possibilities, e.g. we may use RSI of the Ps/Pd ratio. The results will be different but this will not change the general idea.
Here are some selected results (using NinjaTrader software using data from Kinetick)
|Number of trades||508|
|Average time in market||126 days|
The results are quite positive, even though the maximum drawdown is perhaps too large.
This is what happens if we add a 10% stop-loss to the strategy:
|Number of trades||717|
|Average time in market||81 days|
The performance is worse but at least this strategy is at least safer in case if a stock decides to drop heavily (it has not happened here, but it may in your real trading).
This strategy can be optimized by changing the parameters of the Bollinger Bands, modify the stop-loss etc. You can consider adding some other criteria like bullish price action in the stock (e.g. bullish candles, testing a resistance or so). This I did not do in my backtesting.