Quantify predictive signal value using mutual information and lag analysis.
npx skills add https://github.com/trudumb/hyper_make --skill signal-audit
Please help me install this Agent Skill. Skill: Signal Audit Source: https://github.com/trudumb/hyper_make/tree/main/.claude/skills/foundation/signal-audit Command: npx skills add https://github.com/trudumb/hyper_make --skill signal-audit
Systematically quantify the predictive information content of candidate signals to guide feature selection before model building, reducing wasted effort on low-value features.
Run a signal audit on your dataset to identify top signals by mutual information and review regime-specific results before building a model
High-intent search queries and answers about installing and using this skill.
Quantify predictive value by estimating mutual information between continuous signals and targets. This approach identifies high-value features and reduces wasted effort on low-value signals before you start predictive modeling.
Mutual information estimation for signal analysis measures the statistical dependence between a signal and your target variable. It applies regression normalization and regime conditioning to quantify exactly how much predictive information a candidate feature contains.
Identify leading indicators by applying lag optimization to your dataset during a signal audit. This captures temporal effects to pinpoint signals that consistently precede target movements, helping you select features that act as true leading indicators.
Yes, you can condition signal analysis by regime for cross-source datasets. The process applies regime-specific mutual information estimation to evaluate how signal predictive value shifts across different market regimes or data segments.
Track signal decay by running audits that monitor mutual information values across temporal windows. This decay tracking identifies when a signal's predictive information content begins to degrade, allowing you to adjust features before model performance drops.
Run a signal audit before feature selection to prioritize top signals, during quarterly reviews to monitor decay, or when debugging scenarios arise. This prevents wasted effort by validating that candidate features retain predictive value.
SJTU-IPADSClean datasets and generate statistics, correlations, and visualization templates.
doanchienthangdevGuide LoRA and QLoRA finetuning with memory calculations and model merging.
20YN04Generate predictive trading signals from OHLCV data using machine learning models.
the-thinker0Quantify uncertainty with entropy and mutual information measures.
