• Registered: 2020-12-10
  • Posts: 19

Topic: Too good to be true.....

Hey guys. I've only just got back into creating strategies after a hiatus, and developed one that is just too good to be true. I generated it using 30% OOS, but also holding back 2 years of additional data.  It easily passed all Monte Carlo and walk forward, but the kicker is the Multi-market - see stats attached.

It was generated on EURUSD, which is one of its worst performing pairs with a system quality number of 4.97 (some pairs are >8).  I've immediately put it on demo on a few pairs, but interested in other experiences where you've generated a strategy where teh backtests can't possibly be accurate?

  • footon
  • footon
  • ◄≡≡≡►
  • Offline
  • Registered: 2009-08-17
  • Posts: 2,969

Re: Too good to be true.....

Don't see the stats...

When things look too good to be true, then they probably are, but isn't it nice to look at smile
Usually these things arise with a bug, in a (custom) indicator or in the program. So if you can confirm that the behaviour is not the same between live trading and backtest, then be sure to upload that strat here for a check-up!