Hey OP, honest question, do you realize the first image, a backtest of Ebay, indicates that your algo underperfoms a buy and hold strategy? If a long position was opened near the peak in early 2023, at about $50/share, it would be at ~80% profit today, more than double the indicated P&L of 36% from backtesting.
Can you backtest other stocks, notably S&P 500 members, over the same or similar periods and see if your 'mean reversion logic' outperforms or underperfoms?
You should always use the return of a buy and hold from the start of your data as a baseline when testing these sorts of algorithmic/technical analysis trades. If it can't beat a buy and hold, it's non-competitive. Either your execution is flawed, or the idea itself is.
I dunno, I wouldn’t call you a losing trader. This looks a lot like Lance’s p/l curve. It’s a lot of nothing for years, but you’re at least not losing it all because you’re trading small. You finally figure your shit out, and your curve skyrockets.
This looks like exactly what I would expect a professional traders curve to look like after they finally figured things out.