I'm learning [ Quant finance ] by building. Here's what 26 years of gold data taught me.
Built a multi-factor system from scratch - 15 alpha signals, walk-forward ML validation (Random Forest + Gradient Boosting), 6,393 trading days
backtested.
What I found:
- Macro signals (DXY, Fed rates, VIX) beat every technical indicator
- Best signal: VIX fear regime → Sharpe 0.56
- Worst: Bollinger mean-reversion - gold trends, it doesn't revert
- Combined system: 54.7% win rate, 0.70 Sharpe, -21.9% max drawdown
The biggest lesson? Technical analysis alone is a coin flip on commodities. The real edge is macro regime detection - dollar weakness + rate cuts
together = Sharpe 0.47.
Stack: Python, FastAPI, scikit-learn, yfinance, FRED API, real-time WebSocket dashboard and 100+ international datapoints.
Not a dashboard. A quant system. Still learning, still building.
Design learned and inspired by the best Bilawal Sidhu, a legend in Spatial Intelligence, World Models & Visual Effects.
#QuantFinance
#BuildInPublic
#MachineLearning
#GoldMarkets
#SystematicTrading
