Forex Trading 2018- TA/FA- Setups and discussion, page-4216

Hi masonhlys,

@sharks37 could probably provide a better advice than myself but when you backtest a particular strategy, you need to backtest both in-sample and out-sample period. Also like you’ve pointed out, the volume of trades is also important to study, ideally you’d want to see at least 500 trades and above but it really depends how your strategy works, i.e. does your strategy trade on daily timeframe, which obviously you’d expect less volume of trades.

So if you’re testing from 2012-2018 (in-sample period), which is a good length to test, you’d also want to test a number of years outside of 2012-2018, say 2008-2012, to see if it still performs as expected.

I guess to summarise the backtesting process - first you pick an in-sample period (generally 3 years above is good enough) and test to see if it has potential to make money. Next is to optimize the parameters within the in-sample period and then using those optimized parameters to apply onto the out-sample period to see if those parameters would work on the data that they haven’t seen before, which would validate whether your strategy is robust - doesn’t have to be exactly the same result but you’d want to see it’s still making money while keeping the drawdown percentage to a minimum.

Hope that helps.