Multi-Asset Quant Engine Prototype

Prototype quantitative engine with a web interface that combines a stock screener and a macro module (gold, oil, US rates). The screener scores each stock by combining fundamental and technical factors with news sentiment, analyzed with a financial NLP model. A backtesting module compares trend-following strategies net of transaction costs and sizes positions with the Kelly criterion. In development: calibration of the factor weights.

End-to-end quantitative application architecture, REST APIs, feature engineering, backtesting with transaction costs.

Python, FastAPI, pandas, NumPy, yfinance, Hugging Face Transformers (FinBERT), HTML and JavaScript.

Piotroski F-Score (simplified), FCF yield, momentum, z-score, RSI, MACD, Chaikin Money Flow, Keltner channels, moving-average crossovers, Kelly criterion (half-Kelly), profit factor and drawdown. In development: factor weighting with Monte Carlo and a Dirichlet distribution.