Discussing the article: "Trading strategy based on the improved Doji candlestick pattern recognition indicator"

 

Check out the new article: Trading strategy based on the improved Doji candlestick pattern recognition indicator.

The metabar-based indicator detected more candles than the conventional one. Let's check if this provides real benefit in the automated trading.

After that, only about 600 remained for a one-bar EA and about 900 for a meta-bar one (since it makes more trades on average) out of about 5000 setting options. Let's build a single graph based on them to compare the training stages of a one-bar EA and a meta-bar one:

Minimum reliable optimization results


The graph shows that the one-bar EA on average (as can be seen from the regression lines) had a slightly better result over the entire training period. But the meta-bar EA had a wider spread of results, both in the direction of greater losses and in the direction of greater profitability. This can be explained precisely by the fact that more trades were made on the metabars, and this worked as a kind of amplifier: unprofitable EA multiplied the loss (because they made more unprofitable trades), profitable ones multiplied the profit.

Therefore, it was the metabar EA that achieved the most profitable results. The graph shows that only that EA was able to achieve a recovery factor above 6 more than once.

Author: ds2