Here are my thoughts on my original message, where I asked which strategy you think is best, and then later posted the responses.
There was some insightful feedback, so it’s worth going back and reading those threads if you haven’t already.
This is one of those rare strategies where one or two of the columns I’ve added to the backtest are highly predictive.
In fact, each version of the strategy is just using increasingly extreme values for just two columns.
For that reason, I’m not worried about curve fitting, even with version 4 of the strategy.
If I had to use several rules that the Strategy Cruncher suggested to create version 4, that would have been a different story.
When you identify a column or two that are highly predictive, it gives you lots of ways to trade the strategy. (There are no wrong answers to my original question.)
In this situation, I like starting with a version that trades frequently and doing so with tiny size.
This lets you quickly see if the trades are measuring up to the backtest in live while risking a negligible amount of money.
More on this topic and next steps soon…
-Dave
