The same deterministic process runs every day and is replayed in the backtest - so what you see live is exactly what was validated on history.
Ingest, score, gate, select, plan, and monitor - each stage is transparent and traceable.
Each cycle pulls prices, option chains, IV and greeks (Polygon); unusual options flow (Unusual Whales); insider, congress, 13F, WSB, and off-exchange data (Quiver); earnings and analyst actions (FMP); the VIX regime (FRED); social sentiment (StockTwits); and filings (SEC EDGAR). Every feed is independent and degrades gracefully.
Each candidate is scored on 13 signals, then combined with tuned weights into a single 0-100 conviction score. Manipulation and dark-pool pressure are inverted, so crowded or manipulated names are penalized rather than rewarded.
Technical trend blended with the volatility regime confirms the setup is bullish before a name can be picked - it's a long-only shares strategy. If the top score fails to clear the confidence gate, the day is reported as 'No high-quality trade today.'
For the winning name the engine sets the entry price and a ~10-trading-day (about two-week) hold plan, and screens for liquidity - so you're only buying shares you can enter and exit cleanly.
Conservative / central / aggressive share-price targets, a protective stop, a time-based exit at ~20 trading days, and invalidation conditions are attached before publishing.
Open positions are watched against live rules - the share-price target, the protective stop, or the ~10-trading-day time exit - and an alert is raised the moment one triggers.
The conviction score is engineered to be explainable, risk-aware, and validated.
The ranking, direction, pick, and exit are all produced by deterministic math. The AI layer only writes the explanation and can suggest a refinement - it never overrides the engine.
Higher-signal inputs like options flow carry more weight than lower-signal ones. Each component's exact point contribution to the final score is shown on the dashboard.
Manipulation risk is inverted before weighting, so heavy dark-pool shorting or crowded names actively reduce conviction instead of inflating it.
The identical strategy is replayed across years of market history in the backtest engine, so the gate and weights are validated, not guessed.
Before a single weight or gate goes live, it is replayed across years of actual share prices in the backtest engine - each pick bought and held to its share-price target, protective stop, or ~10-trading-day exit. Win rate, annualized return on deployed capital, and max drawdown are measured, and a gate-threshold sweep is run to confirm the recommend cutoff. The AI overlay is deliberately excluded from backtests, so the track record reflects the deterministic engine alone.
The deterministic selection above feeds every product, so they all trace back to the same point-in-time research.
The validated bullish name as shares - entry price, conservative/central/aggressive share-price targets, a protective stop, and a ~10-trading-day time exit.
An independently selected high-conviction bullish name as a single long call on a real, tradable strike - usually the shares pick, sometimes a different name when the top pick is too volatile for premium. Shows the option's live (delayed) price next to the pick-time price, and a calculator to set your own profit target and capital at risk.
A parallel stacked-EMA "cloud" engine (8/9, 20/21, 34/50, 72/89 EMA pairs) that scores trend, momentum, and low-risk pullback entries 0-6 across the universe, with concrete exit levels and a live (delayed) price overlay - pure deterministic math.
An agentic AI assistant that answers any ticker, sector, options, or strategy question. It gathers the live 13-signal score, market data, options analytics, company profile, and risk flags in parallel, then streams a structured, honest read - and flags missing data instead of guessing.
Follow a real candidate from raw signals to entry price to exit plan.
Stella is an AI research agent that uses real-time and historical market feed data to score securities and publish daily scores and illustrative outcomes. This is not investment advice. Stella is not a registered investment adviser, broker-dealer, or certified financial planner. The top-scoring idea on any day can and often will lose money. Past performance, backtests, and simulated P/L do not guarantee future results. You are solely responsible for your investment decisions; seek advice from a properly licensed professional. By using StrikeIntel you agree that StrikeIntel and its operators have no liability for trading losses or decisions made from this information.