Anomaly Detector

Execution anomaly detection monitors slippage, fill quality, price gaps, and volume irregularities in real-time trading systems. This Skill applies statistical methods (z-score, moving-average deviation) and machine-learning models (isolation forest, LSTM autoencoders, one-class SVM) to identify unusual activity across multiple symbols and produce decision-ready signals.

Yes. Slippage and spread monitoring are core features. The Skill detects abnormal volume, liquidity issues, and spread conditions to alert for suspected market manipulation, mispricings, or unusual fill behavior that impacts execution quality.

The Skill implements three detection approaches: statistical methods (z-score, moving-average deviation, quantile-based checks), machine-learning models (isolation forest, LSTM autoencoders, one-class SVM), and rule-based checks to analyze execution quality and identify price gaps and volume anomalies.

Use the anomaly detector to analyze an order execution in real-time, evaluating slippage and cost impact. The Skill produces structured anomaly analyses and performance metrics that flag unusual activity and market conditions affecting your fill.

The Skill requires alpaca-py for market data and trading integration, pandas for data manipulation, and numpy for numerical computation. These handle data ingestion, preprocessing, and the statistical and ML-based anomaly detection logic.

Yes. Price gap detection is a core feature. The Skill identifies significant gaps and flags potential catalysts, helping you distinguish normal market moves from execution issues or manipulation risk.