Assessing Timeframe Quality with Trend, Memory, and Randomness Measures
Summary
The analyzer proposes judging whether a market timeframe is suitable for trading by combining several rolling measures: linear-regression signal-to-noise ratio, lag-one return autocorrelation, the Hurst exponent, a volatility-clustering oscillator, and normalized Shannon entropy of discretized returns. It interprets stronger trend fit and persistence as evidence of structure, while higher entropy suggests more randomness. Hurst readings near one-half are described as random-walk-like, with higher and lower values associated with trending and mean-reverting behavior.
The document supplies example windows, entropy-bin counts, and weights for a balanced setup, trend or breakout use, intraday use, and avoiding choppy conditions. These settings are recommendations without accompanying tests, thresholds for a combined score, or performance results. The measures can be noisy and regime dependent, and the explanation does not define the volatility-clustering calculation or demonstrate that the proposed weights select profitable timeframes. Treat the settings as hypotheses to validate, not established defaults.
Key ideas
- The analyzer combines trend fit, return autocorrelation, Hurst exponent, volatility clustering, and return entropy.
- Higher entropy is interpreted as greater randomness, while Hurst values above or below one-half indicate trending or mean-reverting behavior.
- The document gives separate example weights for general use, trend following, intraday trading, and chop avoidance.
- It provides no backtest results or validation for the recommended parameter settings.
- The volatility-clustering calculation and combined scoring method are not fully specified.
Tags
This summary was written by Stratmill's research agent from the original; it is not a copy of the source.