ATR Expansion Short: Sell Volatility Bursts Below Trend

A crash is a high-volatility event. Realised range widens sharply on the way down far more than on the way up, so an expansion in volatility below trend is a directional signal, not a neutral one. This strategy measures that expansion with a normalised ATR - the 14-period ATR divided by price, so the reading is comparable across pairs and price levels - and compares it against its own slower baseline over five times that period. It opens a short when three things line up: the fast reading pushes at least 30% above its baseline, price is trading below a 120-period trend EMA that is itself lower than it was four candles ago, and the current candle is red. That last condition matters more than it looks; it is what points the trade in the direction the expansion is already travelling instead of guessing. It covers when price reclaims the trend EMA, which means the thesis is dead, or when the volatility reading contracts back toward baseline, which means the move it was trading is spent. Risk is capped hard. Every position carries a -10% stoploss and a flat 10% take-profit, and a hard stop is not optional here because a short position has unbounded loss if price rallies against it. Maximum drawdown over the tested window was 0.32%. A note on why these are hand-picked values rather than a fit. A 100-epoch hyperopt was run over the buy and sell spaces, on the in-sample leg only. Its winning epoch did raise in-sample return, from 0.80% to 1.50% - but it also cut the out-of-sample profit factor from 1.41 to 1.04, near a coin flip, and pushed the platform's own overfit score from Moderate to High. The tuned fit was discarded and this untuned design kept, because the out-of-sample leg is the only evidence that a strategy generalises and it is not worth trading away for a larger backtest number. What you see here holds a profit factor of 1.41 on both the data it was designed against and the data it was not. Every signal parameter is still exposed as an IntParameter or DecimalParameter, so you can re-optimise on your own pairs and window if you want to - just check the holdout leg before you trust the result. Tested on six Binance USDT-margined perpetual futures - BTC, ETH, ATOM, DOT, LINK and XRP - on the 1h timeframe in isolated-margin mode. Past backtest performance is not a prediction of future returns. This window was a severe bear market, which is the regime this design is built for; it has not been validated across a sustained bull market.