Backtested my crypto strategy properly, got zero edge. What am I missing?
Built a crypto analysis tool over the last few months (planning and coding alongside Claude, which caught a few of my wrong assumptions along the way). It's not a bot, it reads the market and outputs a plan: entry, stop, take-profits, direction, and reasoning. Manual execution, 10 coins.
The strategy is the standard YouTube-lecture stuff: multi-timeframe support/resistance, confluence zones where levels line up, RSI/ADX/Bollinger, enter at the zones.
I backtested it across 3 years, ~6,000 trades. Then got paranoid the backtest was lying to me, so I went through the code line by line and re-derived individual trades from raw candles, checking for lookahead, making sure fills and exits were honest. It's solid.
Result: zero gross edge. Coin flip before fees, negative after. I also built a base-rate engine (bucket the indicators into market "configurations," trade the ones with a historical lean), same thing. The patterns separate returns in hindsight but carry no info you'd actually have at decision time.
I keep landing on "public indicators don't carry a tradeable edge, everyone sees them at once so it's already priced in." But I don't want to accept that without asking people who've done this.
- Is that the expected result, or did I mess up conceptually?
- If you found something that worked, was it different data (funding, order flow, on-chain), different timeframes, or execution over signal?
- Wanted to test funding/OI/liquidations but Binance only keeps ~30 days. Free source for longer history, or do people just collect forward?
Not selling anything, just trying to work out if I'm chasing something that isn't there. Happy to share methodology.