Rolling Coin and Moving Average Selection with Double-Crossover Signals
Summary
This article describes a crypto futures framework that periodically selects liquid contracts and tunes a double moving average crossover strategy for each one. It first ranks USDT contracts by dollar trading volume, then backtests multiple fast and slow average pairs for every candidate. A composite score incorporates win rate, profit factor, drawdown control, and a bonus for coins with elevated volatility; each coin’s highest-scoring parameter set competes against the others to form a trading whitelist. The selected pairs generate long and short crossover entries, with trailing profit protection and position scaling based on Bitcoin volatility conditions.
The author presents this as an application of an experimental “harness” approach, while acknowledging that the framework uses full historical samples without holdout validation. Selecting the best parameters and then the best coins compounds selection bias, and the scoring weights and volatility bonus are empirical. The document reports no specific performance results and says the assumption that recent winners persist has not been statistically validated. Its central lesson is to separate candidate generation from evaluation while treating rolling historical selection as a hypothesis requiring out-of-sample and live assessment.
Key ideas
- The framework selects liquid USDT futures contracts using recent dollar trading volume.
- It tests multiple double moving average pairs per coin and ranks coins using their best historical score.
- The composite score combines win rate, profit factor, drawdown control, and a volatility percentile bonus.
- Selected crossover strategies use trailing profit protection and Bitcoin volatility states to adjust exposure.
- Using the same historical data to choose both parameters and coins creates compounded overfitting risk.
- The article identifies the lack of holdout validation and empirical scoring weights as important limitations.
Tags
This summary was written by Stratmill's research agent from the original; it is not a copy of the source.