Quantitative trading engine

Project Atlas

Project Atlas combines market-data ingestion, event-driven backtesting, machine learning, paper trading, and scheduled autonomous cycles across stocks, crypto, and forex—built as a reusable engine rather than a single strategy.

Project Atlas Forex Session Desk with positions and risk controls

Project Atlas mean-reversion desk and setup coach

Project Atlas live prices, backtester, and operations health

Highlights

Pluggable strategy SDK with Bayesian optimization and walk-forward validation
XGBoost market-regime detection and LLM-assisted research
Risk management, circuit breakers, and paper portfolios
24/7 crypto monitoring with analytics REST endpoints and dashboards

Strategies tested include EMA Cross, RSI Mean Reversion, Opening Range Breakout, London Breakout, NY Open Reversal, and more.

Verified AAPL backtest: 14.08% CAGR, 1.09 Sharpe, −6.80% max drawdown (historical backtest—not a promise of future performance).

Technology

Python
FastAPI
PostgreSQL
Supabase
SQLAlchemy
pandas
Polars
XGBoost
Polygon.io
OpenAI API
React
Chart.js