What I was looking for
Bugs that still produce plausible numbers, in an Expert Advisor audit and across trading libraries.
What I found
Oldest bar for newest, losses for wins, excess kurtosis for raw: the result is still a number of the right type in the right range. The report fills in and the number gets trusted.
barter-rs fed its win rate the loss count, so it reported 0.30 where the answer was 0.70; three independent libraries agreed on the right value. An MT5 Expert Advisor I audited read its price arrays oldest-first and sold on a bullish MACD cross. Both the issue reporter and the author had missed it.
The move
I test against a known answer, not against "it runs": type and range checks cannot see a flip.