Number of trades for good backtest?

I was just wondering how many trades you think it takes in order to have a good/reliable backtest.

I�d say a minimum of 500 trades would be a good backtest. That should effectively give you an idea of how your system performs in all kinds of market conditions.

I think it depends on the timeframe. If you have a day trading system I�d say you would need to have more than 500 trades because you will be trading a lot more. Plus, intraday movement is very choppy so you will have to see if your system can withstand those whipsaws.

I don�t think you can put a set number on it but just remember that the more trades you have on your backtest, the more accurate your conclusions will be. Just make sure you are testing the system on all market conditions�ranging and trending.

I think if you do at least a years worth of backtesting then that should be sufficient enough to give you general results and confidence in a system. Also, it seems like the markets focus on at least two different themes each year, so a years worth of back testing should give you data that’s been tested through a couple different market environments.

Peace

It’s not possible to put a set number of trades to proper backtesting. I keep thinkning 100 is a good number, but it depends on your time frame. Trading off the 1 minute charts is going to give that number of backtested trades in no time flat but it will not represent how your system performs in all kinds of markets. As long as you have tested your system across all markets (ranging, up trending, downtrending, etc…) then i think this is a good start.

I’ve never backtested anything with a computer, but since you can program in your systems, couldn’t the computer parse the past 30 years worth of data pretty quick? Or is the problem in amassing the past 30 years of data?

:confused: I’m new in fx. Which progam would I use to backtest my trading system?

At the risk of repeating myself, I backtested a system 10 years which proved profitable before starting to lose all the money, so in my opinion there will never be enough.

I’ve been using roughly ten years of data for most of my backtests. But there is so much more to it than just the amount of data - most importantly, how are you determining the “fitness” of your results? If it’s just ending total equity, you could be setting yourself up for major disappointment.

And this wasn’t directed at you, PPF, just the discussion in general.

I appreciate that, I echo what you say.