Most investors navigate markets on noise — news cycles, sentiment, and incomplete data. Parallax Labs builds systematic market signals using machine learning, statistical research, and exploratory data evidence. The objective is disciplined, transparent signal intelligence rather than discretionary market timing.
Continuous classification of volatility, liquidity, and market regime states using probabilistic models and historical market data. Designed to track structural transitions and alert conditions that can affect portfolio risk.
GMM · HMMWalk-forward information coefficient tracking across a live strategy universe. Ensemble gradient-boosted models rank securities by probability-weighted return expectations each month.
Gradient Boosting · IC GuardDynamic exposure models with drawdown-aware controls. Volatility targeting, cash flexibility, and tail-risk overlays are reviewed through the production rebalance cycle.
Factor Models · Vol TargetingWalk-forward tested over 166 monthly decisions. Designed to reduce look-ahead bias and data leakage.
Institutional-style models — gradient boosting, hidden Markov regimes and market geometry — distilled into systematic monthly signals. Built for disciplined investors who want evidence, risk controls, and benchmark transparency.
Machine learning models trained on historical price, macro, fundamental, liquidity, and market-geometry features. Results are historical simulations using point-in-time data, walk-forward validation, and bootstrap robustness analysis. Live results may differ materially from backtested performance.