What it is
Donchian breakouts (buy a new N-bar high), moving-average crossovers, ATR-based trailing stops. Positions are sized by volatility so each coin risks the same. Most of the return comes from a few long trends; most trades are small losses.
trend following
Trend following: lose small often, win big rarely.
Buy new highs, sell new lows, hold until the trend bends. Low win rate, long flat periods, and the one strategy family that kept an out-of-sample edge in our own multi-coin research. Also the easiest to overfit if you tune the lookback on the same data.
Official Binance broker · Non-custodial · Live only when you confirm · Updated
What the backtest must get right
Enough history to include several regimes (2021 to now at minimum), fees and funding over long holds, and an honest treatment of the lookback: if you tried 10, 20, 50 and 100-bar breakouts on the same data, the overfit check counts four variants and deflates the Sharpe accordingly.
A prompt that builds one
Fixed rules, few parameters, and a walk-forward in the same breath.
“Trend-follow a basket of 10 large caps on 1d: buy a 50-day high, exit on a 20-day low, 1% risk per coin sized by ATR, then run a walk-forward.”What the research says
In OptAlgo's own coin-agnostic study only fixed-stop trend breakouts kept an edge out of sample, with a drawdown you must be able to sit through. Mean-reversion and cross-sectional momentum variants did not. Your result will differ; run the walk-forward.
faq
Trend following FAQ
What win rate should a trend strategy have?
Often 30 to 45 percent. The profit factor comes from the size of winners, so judge the strategy on walk-forward return and drawdown, not win rate.
Daily or 4-hour bars?
Both work; daily has fewer trades and lower fees, 4-hour catches turns earlier at the cost of more whipsaws. Backtest both as one study so the overfit check sees them together.
Do I need to code this?
No. Describe the rules to Claude or ChatGPT through the OptAlgo connector and the assistant writes the strategy against the OptAlgo SDK, validates it for free and backtests it. You can read and download the code.
How do I know the backtest is not overfit?
Every run returns an overfit verdict based on how many variants you tried. Then ask for a walk-forward or a holdout: the strategy must keep its edge on data it never saw before it is worth paper trading.