I build trading bots. Last year, I programmed hundreds of trading strategies, multiple combinations of strategies, and experimented with different closures like stop-loss, take-profit, stop-market once the price reaches a certain level, opposite signal, different levels etc.
However, when I backtest these strategies on TradingView, they consistently underperform in the long term, especially after accounting for fees.
Does anyone have a strategy that actually works?
You've done a lot that is commendable, but if you've not achieved a good strategy, you will only be wasting your time. It's even worse if backtesting which is not deemed to be accurate when it is even profiting is now making you lose. You need to overhaul the bot(s. Well, I have many trading strategies that work but the simplest one you can try now is the price action or the chart pattern. Just learn it and apply it rightly and also vary it on different time frames. If your bot(s) works well, there is no way it will fail with this strategy.