Negative PnL even when statistics imply profit



I have a conceptual issue that maybe someone more experienced could help me out to figure it out. I am running an algo on backtest mode for a 4 month period. Some statistics:

  • The algo does 37 buys and 37 sells.
  • Out of those 37 sells 23 are exited using a software stop loss (37.83% of the exits are normal and 62.16% are stop loss, not great at all, I know…).
  • The average profit on each trade is +7.58% and the average stop loss is -3.85%
  • The average risk to return ratio is 1.96.

Now the question is, despite the accuracy is low, the risk to return ratio is about 1:2, therefore the algo should have profit however the reported PnL on Catalyst is negative. Any idea please?

Thank you very much.


I’m not sure I understand the logic. But some points to consider:

  • Are the profit and losses per trade that you are quoting taking into account transaction costs that may apply to the backtest?
  • Unless you are speaking about log returns, the order of the trades will influence the profit and loss. You cannot simply average them.
  • I’m not sure what the risk to return is. Something like the inverse of the Sharpe ratio? What does it have to do with estimating the returns?


Thank you for your answer, I had a concept understanding problem.

I think it is now more clear. Apparently one of the problems is that the stop loss was to tight.