Predictor
A directional lean for the next few trading days, blended from three signals you can inspect — plus the public record of how often that lean has been right. Educational, not financial advice.
Predict a stock
A lean for the next few trading days, blended from three signals you can inspect: the ML transformer's price forecast, post-earnings drift with EPS growth, and news sentiment.
Forecasting price, weighing the last earnings report, and sampling news tone.
A "60% up" call is only worth something if the market rises about 60% of the time it's made. This replays the predictor out of sample, on data the model never trained on. Points on the diagonal are perfectly calibrated.
Recovering the model's probability and the realised outcome for every historical window.
Every prediction made here is logged the moment it's made (ticker, probability, price) and graded once its horizon passes. Nothing is edited afterwards, so the hit rate is a genuine forward record that can't be fitted in hindsight.
Grading any predictions whose horizon has now elapsed.