Win rate
The share of closed signals that were confirmed. On its own this number says almost nothing: a strategy with an 80 % win rate can lose money if the rare losses are larger than the frequent gains.
Note: until a single signal closes, a win rate does not exist. The interface shows a dash there, not a zero — a zero would mean "all lost".
Profit factor
The ratio of favourable moves to unfavourable ones. One means break even. Below one you are losing. A value from 1.3 is usually considered sound; anything above 3 over a long period deserves an especially careful check for overfitting.
Maximum drawdown
The worst fall from a reached peak. This number answers the question "how much will I have to endure", and it is what most often makes people abandon a working strategy. Look at it before the final profit.
Number of trades
Metrics computed over fifteen trades mean nothing. A hundred is the rough minimum; fewer describes an episode, not a property of the strategy.
Marks about filters that did not apply
Next to the result the service shows which of the enabled filters did not apply in this run and why. The reasons are of two different natures:
- no data for the period — temporary: once data appears, the filter
works;
- unavailable in a backtest — fundamental: the adaptive threshold
and dynamic weights adjust from already closed trades, and inside a historical run that would be looking into the future. In live mode they do work.
Two runs with the same configuration but a different set of applied filters are different computations. They cannot be compared directly.