Design, validate, and run US-equities strategies with AI assistance.

Candles + volume + indicators (SMA / EMA / VWAP / Bollinger / RSI / MACD), full drawing toolkit (trendlines, fib retracements, rectangles, anchored VWAP) with drag-to-edit handles and undo/redo, symbol comparison overlays, per-symbol saved layouts, and price alerts that email you on trigger.

Six classical setups (SMA cross, RSI bounce, 52-week breakout, EMA pullback, gap-fill, swing high) ranked by R:R. Source-performance panel tracks realized P&L per feed so you can tell which sources actually pay.

Intraday scanner (ORB / VWAP / RVOL / failed breakout), pre-market briefing, and a manual journal in one place. ~400-name day-trade universe ∪ your watchlist.

Today's biggest gainers, losers, and volatility leaders across a curated universe + your watchlist.

Your tracked symbols, unioned into every scanner and the Daily Brief. Add / remove inline.

Sub-minute signals from a WebSocket stream — ORB / momentum / mean-reversion / pair stat-arb / news shock.

Submit a proposed order — entry, stop, target. Claude returns thesis-check, risk factors, suggested modifications.

Manual or AI-generated strategies (DSL + Python). Backtest, critique, paper-trade.

Plain-English → strategy DSL via Claude. Validator + smoke test gate the output before save.

Live Alpaca paper account: positions with attached plans, transactions, and FIFO-matched performance analytics — best/worst trades, P&L by ticker, calendar heatmap, by-origin trending (1d / 7d / 30d / 90d windows).

Every market day the bot polls signals, gates them through your rule engine, and submits bracket orders. Pause / kill any time.

Autonomous scanner-evolution engine. Audits the live pool, identifies regime gaps, prompts Opus 4.7 for new scanner code, validates + novelty-filters + walk-forward-backtests, and activates survivors into a SHADOW lifecycle stage. Graduates winners to PROVISIONAL → ACTIVE, retires losers — all hands-off (with an operator killswitch and tunable thresholds).

Type a market hypothesis; Opus 4.7 generates a scanner class, a critic LLM reviews, and a walk-forward backtest decides pass/fail. You activate accepted proposals into the live roster. ~$0.15-$0.30 per cycle.

Per-regime RLS coefficients the ranker has learned from realized R-multiples. Inspect convergence, spot regimes where the model has gone in an unexpected direction, see how confident each learned weight is.

Pre-open + post-close summaries via Resend: gappers, top setups, open-position Coach verdicts, closed-trade ledger. Manual send or twice-daily schedule.