Yesterday, I described how the “guess and check” approach that most traders try when backtesting seldom works.
What should you do for your first backtest?
Instead of trying to find “the final answer” with your first attempt, create a backtest that captures the entire range of possible trades for your system.
Here’s an example. I trade a gap system with a well-defined entry signal that may or may not occur for any given gapping stock on a particular day.
The entry signal is simple – no fancy rules to complicate it.
I know there’s some threshold for a minimum gap that I’ll end up using in the final version, but I don’t know what that is yet.
I also know that I’ll end up applying some additional rules, but I’m not sure what they’ll be.
The first backtest I run will look for the entry signal with essentially no filtering (or as little filtering as possible).
But won’t you end up with more trades in your backtest than your account can handle?
Of course! Way more. Sometimes ten times more than what I end up trading.
My first backtest answers the question: how many times does this simple pattern occur?
I know I’m not going to end up trading every occurrence of the signal, but I start with a comprehensive list.
Once I have that list, I’m ready to answer the next question: under what circumstances is the signal profitable to trade?
In other words, what subset of the entire list of occurrences are worth trading, and which ones should my system ignore?
That’s the crux of creating a trading strategy – and will be the topic over the next several days.
But if your first backtest uses a “guess and check” approach, you’re not even close to being able to answer the most important question when creating a strategy.
-Dave
