Can we really use REPLAY data to prove a strategy?

Hi there,

I'm new to Ninja. I've watched the CREATE, TEST, APPLICATION phase tutorial. Initially I thought - great! - someone is saying that backtesting is a very poor and inaccurate method to test strategy performance. I thought forward testing would be a much better approximation of reality...

Attached is a visualisation of data I collected after running a particular strategy for 7 hours. I collected realtime data, then for the same time period I gather the backtest data as well as the replay (forwardtest) data.  From the picture you can see that the realtime set has a very different result.

Question: I'm struggling to see how bloodhound (and yes it is a fantastic tool) can really add value if the forward testing data itself is such a poor reflection on reality (as it seems to be)?

Have more questions? Submit a request

0 Comments

Please sign in to leave a comment.
Powered by Zendesk