My Overfitting Conundrum

I've been struggling with overfitting on my mean reversion strategy and I'm trying to figure out how to balance performance and robustness. I've been testing different walk forwards and it's been hit or miss. Has anyone else come across a reliable method for preventing overfitting on time-based strategies? I've also been considering incorporating more robust metrics into my backtesting, but I'm not sure where to start. Any advice or experiences would be greatly appreciated.