Aegis Trading Lab
Evidence before execution.
Quantitative research systemOverview
Aegis is not presented as a profitable trading bot. It is a research laboratory built around reproducibility, conservative simulation, fixed promotion rules, and an intentionally absent live-order path.
The problem
Strategy ideas often look convincing before fees, slippage, data leakage, regime changes, and untouched out-of-sample evidence are considered.
What I built
I separated data validation, forecasting, strategy, risk, paper execution, journaling, and reporting, then required every candidate to pass fixed evidence gates before promotion.
Engineering decisions
How the system works
Hard safety boundary
Forecasting cannot size positions, bypass risk, alter configuration, or submit orders. Live mode stays locked independently of research output.
Honest research
Momentum, mean-reversion, carry, trend, rotation, pooled-flow, and copy-audit branches retain rejected evidence instead of moving thresholds after results.
Reproducible pipeline
Checksum-verified public data, chronological folds, next-bar execution, fees, slippage, fixed risk limits, and versioned evidence keep experiments auditable.
My role
Product capabilities
Selected stack