title says it all, when you backtested your edge on different pairs, how much did your data differ from each pair, was the change huge? were some pairs unprofitable compared to others?
Yes of course. Different pair has different character. It’s common a strategy works on a pair, doesn’t work on another.
Just learn the character.
It is a mystery why the same strategy fails on one of two very similar and correlated pairs. There will be nothing in their charts which shows significant differences, nothing to suggest any strategy would fail. But this happens, and it can persist for months.
I put this down to probability, not inherent differences in either the two pairs, nor in how the market views them or trades them.
The outcome is that your best strategy will eventually fail on its best pair, but it will start to make money on its worst pair.
Keep your eyes open.