I have an interesting dilemma I've ran into while developing a strategy I've adapted to nonjascript where the backtest results are so skewed from live performance(which in and of itself isn't strange) but not in the usual way. No matter whether i try using tick replay or high resolution order fills, i cannot figure out how to optimize this thing bc if i do find backtest results that are decent, it will perform terribly live, but with bad results, the live run will wildly outperform it's backtest data. So far, i can't find any template which improves on the default settings and any deviation from it greatly reduces profitability. Now, I'm not exactly deadset on changing things that don't need to change, I've given up on that and I'm trading with it (automated with manual intervention) but I'm really curious if anyone else has experienced it. I've made quite a few strategies and yes, i get it, backtest results aren't always reliable but this is something different altogether.
Keep digging. Eventually you'll find the Holy Grail somewhere on this planet and you'll make loads of money and you'll be living your best divine life