Our method, in plain words

How Prodict works

What goes into a pick, what the model does with it, and how every result is graded in public.

The short version

A statistical model estimates how many goals each team is likely to score. Those estimates become probabilities for each market, which we compare with bookmaker odds. An AI model then reads the numbers, chooses among the markets and explains its reasoning. After full time every pick is graded and added to the public record, wins and losses alike.

  1. 1

    Match data

    For each fixture we gather both teams' recent goals scored and conceded, the averages of the league they play in, and the bookmaker odds on offer.

    • Goals scored and conceded by each team in recent matches.
    • League baselines: how many goals a typical team scores in that competition.
    • Pre-match bookmaker odds for the markets we cover.
  2. 2

    The goal model

    A Poisson model turns each team's expected goals into a probability for every scoreline. Scoring rates are adjusted for the strength of the opponents a team has faced and anchored to the league baseline.

    • Scoring rates adjusted for the quality of the opposition.
    • League baselines keep each estimate grounded.
    • Scoreline probabilities add up to probabilities for 1X2, over/under goals and both teams to score.
  3. 3

    Odds and edge

    The model's probabilities are compared with the probabilities implied by bookmaker odds. The gap between the two is the edge: a sign that a price may be generous, not a guarantee.

    • Example: the model gives a home win 60%, a fair price of about 1.67.
    • A bookmaker offering 1.90 implies about 53%.
    • That gap is the edge the model sees. Picks with an edge still lose, often.
  4. 4

    The pick and the record

    An AI model reads the numbers for each match, chooses among the markets and writes the reasoning you see on the analysis page. After full time the pick is graded and published.

    • Each pick comes with a confidence level and the reasoning behind it.
    • Every pick is logged before kick-off and graded after full time.
    • Graded picks, won or lost, stay on the public record.

Judge the record, not our word

No model wins every bet, and ours doesn't either. Instead of promising a hit rate, we publish every graded pick so you can check the results yourself. Only stake what you can afford to lose.

View today's predictions

All picks, every day

Cancel in two taps. No prediction is guaranteed — judge us on the record.

See plans