First of all - thanks to all the lovely people on this forum. You have all been tremendous.

I am facing this unique challenge I have never encountered before (probably because I have never tried doing things this way).

I have built an MT5 EA that I have been testing on historical data provided by my broker (ICMarkets Global in this case). I tested this using M1 data on BTCUSD from Jan 2018 to Dec 2024. In parallel, I was learning about custom symbols and importing third-party historical data. I thought I might run this EA on BTCUSD data that I downloaded from Dukascopy for free. When I did that the EA performance changed completely. I have attached pictures of equity curve of both runs. You can see that until Dec 2020 both symbols gave similar results. It is after Jan 2021 that they started going in completely different directions.

So far I have tried the following

1. Visually checked the data and confirmed there are difference but not to the extent to get such different results
2. Ran tests on both instruments with same fixed spread to see if spread is causing issues but get similar differences in results

What else can I try to find out what is going on here?