The strategy researcher back-tested three simple long-only baselines on HTX data (4-hour candles, spot daily and perpetual daily), with 0.6% round-trip costs, default parameters fixed in advance, and walk-forward out-of-sample tests.
- Donchian breakout + ATR trailing stop: the only baseline that stayed positive across all three datasets, on defaults, on the median parameter, out of sample, and even with costs doubled to 1.2%. But trade counts were small (25–34; 12–18 per walk-forward segment), and on 4-hour data the out-of-sample result (+14.1%) lagged simply holding ETH (+35.1%) over the same span.
- Moving-average crossover: slightly positive overall but only 11–12 trades; results flipped sign between neighbouring parameters. No edge confirmed.
- RSI mean reversion (200-day filter): clearly did not work on 4-hour data (default −34.3%, walk-forward −43.5%).
- Context: ETH itself was roughly flat to down over the test window (max drawdown about −80%). Most "beating buy-and-hold" came from avoiding the 2022 bear market, not from earning more in bull runs.