Algorithmic paper-trading bot for US equities. TypeScript monorepo: a strategy library with backtesting engine, and a real-time trading dashboard.
Monthly-rebalanced cross-sectional momentum (blended 6-1/9-1/12-1) with an RSI(2) mean-reversion overlay gated by the SPY 200-day regime filter, and optional LLM-based news sentiment. Risk layer: volatility-targeted position sizing, gross-profitability quality filter, sector concentration caps, correlation-adjusted sizing, and portfolio-level volatility scaling (Barroso & Santa-Clara 2015).
packages/lib: signals, risk, strategy, backtest engine, SQLite bar store, Alpaca market data + paper trading, CLIpackages/web: Next.js dashboard with a candlestick chart, equity curve, signal heatmap, positions, and order blotter
pnpm install
pnpm test # run the test suite (vitest)
pnpm dev # start the dashboard
./scripts/run-backtest.sh
./scripts/paper-trade.shLive data requires Alpaca paper-trading keys in ~/.tap/secrets.env (ALPACA_KEY_ID, ALPACA_SECRET_KEY); the backtest runs against cached bars.
Personal learning project. Paper trading only. It does not place real-money orders, and nothing here is financial advice. Backtests are historical simulations and not indicative of future results. Provided as-is, no warranty.