-
Notifications
You must be signed in to change notification settings - Fork 0
Home
Automated, risk-defined algorithmic chart pattern screener for the S&P 500.
Scans 500+ equities daily on GitHub Actions with zero paid APIs and delivers verified morning setups.
- π‘οΈ Strictly Defined Risk: Every confirmed signal includes an entry price, an ATR-buffered structural stop, and a reference target. Setups requiring >15% risk are automatically dropped.
-
π No Forced Patterns: Strict geometric curvature algorithms (parabolic regression vs. V-shapes) coupled with a composite quality score (
$\ge 60$ ). - π Trend-Gated: Built-in SMA50 / SMA200 trend gates prevent fighting dominant macro momentum.
- π¬ Rigorously Backtested: Includes a walk-forward replay engine and historical git signal tracker to measure real hit rates and R multiples.
- β‘ Serverless & Free: Runs completely within GitHub Actions free tier using Yahoo Finance data. No servers, no subscriptions.
| Pattern | Setup Type | Trigger Condition | Structural Stop | Reference Target |
|---|---|---|---|---|
| β Cup & Handle | Continuation | Daily close above handle high | Handle low β 0.25 ATR | Entry + Cup depth |
| π€ Inverse Head & Shoulders | Reversal | Daily close above neckline | Right shoulder low β 0.25 ATR | Entry + (Neckline β Head) |
| πΊ Bullish Wolfe Wave | Reversal | Daily close back above line 1β3 | Point 5 low β 0.25 ATR | Line 1β4 at the ETA |
Tip
CONFIRMED vs WATCHLIST: Setups breaking out on the latest bar are flagged CONFIRMED. Setups with completed geometry within 5% of their trigger wait on the WATCHLIST.
Explore the full technical design, algorithms, and configuration:
-
ποΈ Architecture and Data Pipeline
Constituent loading, symbol normalisation (e.g.BRK.BβBRK-B), Yahoo Finance throttling, last-bar alignment, and the nightly scan schedule. -
π Pattern Catalog
Exact geometric criteria, parabolic RΒ² roundness formulas, neckline tilt calculations, Wolfe EPA/ETA projections, and score composition. -
ποΈ Configuration and Tuning
Reference guide for every threshold, quality score cutoffs, 5% watchlist proximity, age limits, and false-positive controls. -
π§ͺ Testing and Contributing
Offline test harness, walk-forward backtester (tools/backtest.py), signal outcome tracking (tools/evaluate_signals.py), and how to add a pattern.
- π¬ GitHub Discussions: Join the community, suggest new patterns, and discuss setups.
- π Latest Scan Report: View the most recent confirmed signals and watchlist.
- π» GitHub Repository: View source code, open issues, or contribute.
π Documentation
π¬ Community
π Links