Now that you have your first backtest, you’re ready to tackle the most important part of creating a trading strategy.
Remember, this is a comprehensive list of the occurrences of your original signal, so the equity curve might look terrible at this point. But that’s ok.
You’re never going to trade EVERY occurrence of the signal, so why did we include each one in the backtest?
Because you need a methodical way to determine which signals are profitable enough to trade, and which aren’t.
By including all the signals in the first backtest, you’re guaranteeing that the final version of the strategy at the end of this process will have the maximum possible number of profitable trades in the strategy.
Even though the equity curve might look bad, you should be incredibly excited when you step back and think about it.
There is a high likelihood that some combination of the trades already in your backtest will produce a profitable strategy.
You just need to determine which trades are good enough and which aren’t.
To do that, you need to apply rules to your strategy to filter out the poorest trades from your original trade set.
The only reliable way to figure out what rules to apply to filter out the bad trades is to add data points to your backtest that could be used to create a rule.
Most traders, at this point, create the rules in their code to filter out trades from their backtest, but that’s the wrong approach.
You still want all the trades in the backtest, but you need to add columns to your backtest for each rule you might apply.
This is the essence of building a robust strategy, and yet most traders aren’t even aware they can do this.
In fact, most backtesting software packages out there DON’T HAVE THIS CAPABILITY. (I’m talking to you TradeStation!)
Next, I’ll go over what backtesters have this fundamental feature.
Until next time,
-Dave
