How to read VEKTORIUM's picks
A model pick is more than 'team A wins'. Every VEKTORIUM pick shows three numbers, and they answer three different questions. Learning to read them together is how you decide whether a pick is worth betting.
Win probability: the chance
This is the model's core answer: the chance each team wins, as a percentage. If it says 58%, the model estimates that, over many replays of this matchup, this team wins 58 times out of 100. It's an honest probability, not a guarantee — a 58% bet will still lose 42% of the time.
Confidence: how much the evidence deserves trust
Confidence is a score out of 100 for the evidence behind the pick, not for the pick itself. It is built from what was known before the first pitch — whether both starting pitchers were known, whether the lineups were posted, how much of the season each team has played, whether the weather came from the feed — and from how closely the three models that make up the ensemble agree with each other. Two picks can both sit at 58% with very different confidence: one with confirmed lineups and models that agree, another missing a starter with the models split. Confidence says how much to trust the number; it never says which side is more likely, it cannot be read as a win rate, and a confident pick still loses. Whether it should change how much you stake is a question the published record has to answer — until it does, read it as a warning about how solid the inputs were, not as a licence to bet bigger.
Fair price: the odds the model thinks are correct
The fair price converts the model's probability into an odds line. If the model says 60%, the fair price is about -150. Now compare: if the best book price available is -140, the market is offering better odds than the model thinks are fair — that gap is your edge. If the market is at -170, the market is more pessimistic than the model, and there's no value.