Match analysis, model explainers and football data insights — how PredictIQ Pro’s predictions work, what the numbers mean, and how to use them.
PredictIQ Pro doesn't rely on one model. Five components each read a match — a Poisson goals model, CatBoost, LightGBM and devigged market odds — and a meta-learner blends them into the final win, draw and loss probabilities.
September 28, 2026 · 2 min read
Expected goals (xG) assigns every shot a probability of becoming a goal, based on factors like distance, angle and shot type. Adding those probabilities up shows how many goals a team's chances were worth, which is often a better guide to performance than the scoreline.
September 28, 2026 · 2 min read
A prediction model is well calibrated when its stated probabilities match reality: across many matches forecast at 60%, about 60% should happen. Calibration matters more than raw accuracy if you plan to build on the probabilities themselves.
September 28, 2026 · 2 min read
Win, draw and loss probabilities always add up to 100%, and the favourite is rarely a certainty. A 49% favourite loses more often than it wins. Read the probabilities as a spread of outcomes, not as a single predicted result.
September 28, 2026 · 2 min read
The Poisson distribution turns a team's expected goals into the probability of scoring 0, 1, 2 or more goals. Combine two teams' distributions and you get a probability for every scoreline, and from those, the win, draw, loss, over/under and both-teams-to-score markets.
September 28, 2026 · 2 min read
Decimal odds convert to an implied probability with 1 divided by the odds. Across all outcomes those probabilities add up to more than 100% because of the bookmaker's margin. Devigging removes that margin to reveal the market's underlying view of the match.
September 28, 2026 · 2 min read
To use the PredictIQ Pro football predictions API, create a free account to get an API key, then send it in an Authorization header to endpoints such as /predictions/live at api.predictiqpro.com. Responses are JSON with win, draw and loss probabilities per match.
September 28, 2026 · 2 min read
PredictIQ Pro runs a Model Context Protocol (MCP) server at mcp.predictiqpro.com/mcp. It lets AI agents such as Claude call football predictions and data directly as tools, instead of you writing custom API glue code.
September 28, 2026 · 2 min read