About bsig

A research workbench for trading-signal hunting, backtesting, and live tracking. Free public data only.

Read first. Backtests use free data with no transaction costs, slippage, or borrow. Several "wins" ride on small N (BTC halvings N=3, K5 Phase-3 slip N=9). Crypto signals carry -60% to -77% drawdowns. Past performance is not predictive. Not investment advice.

What this is

An autonomous pipeline that scouts trading ideas, backtests them on yfinance / FRED / public data, applies multiple-testing correction (BH @ 5% FDR), tracks per-signal status against current market conditions, and surfaces an optimized basket. Nothing here is a recommendation to trade.

How the tabs work

Strategies — the curated high-return catalog with mechanism, rule, caveats, and a CAGR threshold filter.
Live — which signals are currently firing (active / defensive / unknown / off), live convergence, anti-signals, recent status flips, and per-signal P&L when logged.
Construction — sub-pages for Regime (current market conditions), Optimize (max-Sharpe / max-CAGR / risk-parity baskets), Portfolio (equity curve + active list), Analysis (BH/FDR survivor tables).
Pipeline — live view of the Idea Scout → Strategy Developer → Backtester loops.

Data sources

yfinance for equities, ETFs, crypto, futures. FRED for macro series. SEC EDGAR for event-driven signal scanning. Everything is free and machine-readable; no paid feeds.

Caveats worth re-reading

Most "high CAGR" regime trades (AS-B, AE-2, AG-2, AI-4, AI-5) have N=1–5 regime observations — their reported Sharpe is in-sample fit, not validated edge. Only 9 of 30 currently-active signals also pass BH multiple-testing correction. The optimization tab defaults to BH-survivors only so the math doesn't lean on those.

← Back to bsig