Paper Trader

Returns over time — deposit-adjusted portfolio % change over standard windows —

Daily briefing —

Since you last looked — material events in the window, ranked (no snapshot scanning) —

Trade suggestions — co-pilot, not auto-executed —

Risk panel

🧨 Position blow-up ladder — each held name shocked alone, idiosyncratic (no beta) —

Earnings radar — scheduled gap risk on holdings & watchlist —

• loading…

Portfolio Greeks — Black-Scholes, live IV from yfinance —

ML scorer · per-position outlook — predicted 5-day forward return from DecisionScorer MLP —

Position thesis — per-holding integrated view: news, scorer, technicals, last decision, verdict —

Drawdown anatomy — current DD from peak with per-position contribution —

Confidence calibration & signal attribution — do high-confidence trades actually win? which signal types pay? —

Capital deployment & liquidity — is the book pinned with no dry powder? —

Decision pipeline health — is the live Opus trader actually deciding? NO_DECISION = parse failure —

Decision failure forensics — WHY a cycle produced no decision, with the raw Opus excerpt —

Decision drought drift — what the bot's inaction cost: portfolio vs S&P while it wasn't trading —

News edge — does a high ai_score headline actually predict the move? (SPY-abnormal) —

Scorer reliability — how far DecisionScorer predictions actually land from reality —

Scorer ⇄ Opus disagreement — where the ML safety net and Opus are fighting on held positions —

Portfolio analytics

Behavioural edge — payoff ratio, breakeven win-rate, the disposition effect —

Loser autopsy — why each closed trade lost: verbatim thesis, hold, failure mode —

Winner autopsy — why each closed trade won: verbatim thesis, hold, success mode —

Track record — per-name closed-trade memory: the verbatim history the trader now sees in its own prompt —

Concentration honesty — pairwise return ρ & effective independent bets —

Capital paralysis — the trap, its cost, and the single sale that unlocks it —

Open-book alpha — how much of the open P&L is selection vs just SPY —

Overtrading & churn — how often it re-buys a name it just closed, and how fast —

Thesis drift — is the reason each position was opened for still true? —

Signal-feed health — is the live trader receiving any news, or flying blind? —

Runner heartbeat — is the decision loop still cycling, or has it wedged/died? —

Decision reliability — headline NO_DECISION % vs the true post-restart rate —

ML gate edge — does the DecisionScorer earn its complexity OUT OF SAMPLE, or would a one-line rule do? —

Funded suggestions — which BUY/ADD idea is fundable, and the sale that unlocks it —

Signal follow-through — did it act on the news it saw, and did acting pay (vs SPY)? —

News source edge — which of ~17 collectors' scored headlines actually precede the move (vs SPY)? —

Behavioural scorecard — do the independent behavioural checks agree on a problem? (no grade, just concordance) —

Sector pulse — semis & optical —

DRAM / semis heatmap — 5d momentum & news pulse —

Deduped signals — syndication collapsed, urgency decayed (halflife 4h) —

• loading…

MACD Signals — per-ticker momentum scan · ⚡ = high-conviction cross-up below zero ▼

Trade outcomes — closed round-trip statistics · win rate · expectancy · hard-exit attribution —

Decision log

Model leaderboard — ranked by avg yearly vs-SPY % · 🐒 random baseline anchored at the bottom —