Stock AI
Adaptive next-session analysis across the curated master universe. The scored call is frozen in the final premarket run, while hourly market intelligence tracks new information that can change the outlook overnight.
Important: These are model probabilities and paper-model results, not guaranteed predictions or investment advice. The investor score uses an equal-weight $100K simulation where UP is modeled as 1x long, DOWN as 1x short, and NEUTRAL as cash; it excludes fees, borrow costs, slippage and taxes.
Hourly intelligence
Critical Market Movers
The five highest-impact signals are reranked from fresh global news plus cross-asset price confirmation. A high impact score means the signal is broad, fresh and confirmed; it does not prove causation.
Current driver bias: loading…Updated: loading…
What changed since our prior forecast?
Comparing the latest hourly evidence with the prior forecast…
Current forecast
Updated: loading…Predicted Direction
The evening outlook is diagnostic; the final premarket prediction is the call later evaluated against the completed regular session.
| # | Company | Ticker | Stock Price | Direction | UP Prob. | DOWN Prob. | Confidence | Key Triggers |
|---|---|---|---|---|---|---|---|---|
| Loading Stock AI predictions… | ||||||||
Continuous calibration
How the model learns
1. Predict
Before the next session, store the probability, base price and feature values used for every ticker.
2. Verify economic outcome
After the regular session closes, score the frozen final-premarket call using each asset’s open-to-close move, raw direction, Brier score and its contribution to the equal-weight $100K paper portfolio.
3. Learn without chasing one day
Magnitude-weighted errors feed one bounded session update. Rolling 5-, 20- and 60-session results provide context so one unusual day cannot aggressively rewrite the model.
Model-learning status loading…
Independent evidence framework
What Stock AI investigates
Each ticker is evaluated independently. Company news, price state, history, peers and only materially relevant macro or geopolitical events are combined into the displayed Key Triggers. A future LLM layer can deepen semantic event interpretation without changing this evidence contract.
War escalation / ceasefireStrait of Hormuz / oil shockRussia / Ukraine sanctionsChip shortage / foundry constraintsU.S. election / tariff policyCPI surprisePayroll surpriseFOMC surpriseEarnings / guidance20Y similar-state history20%+ drawdownsDRAM / HBM / memory cycleSupply-chain disruptionsSector / peer behaviorYear-end tax / seasonal effects