Likely goalscorers: how AI estimates the chance that a player scores

Joueur face au but dans un stade éclairé

How does an AI estimate the likely goalscorers in a match? It does not name the player who will score: it estimates, for each player, the probability of scoring at least one goal. That probability rests on a few measurable factors, from expected minutes to expected goals (xG), and is usually below one in two, even for the best strikers.

Even the best strikers do not score in every match

A real-world benchmark helps set expectations. In the 2025-26 Premier League season, Erling Haaland (Manchester City) finished as the league's top scorer with 27 goals in 35 appearances. Yet those 27 goals came in just 20 matches: he scored in around 57% of his league games and failed to score in the other 15. For a less prolific player, the chance of scoring in a given match is much lower.

The factors that matter

1. Expected playing time

A player can only score while on the pitch. A starter who plays 90 minutes does not have the same probability as a substitute who comes on in the 70th minute, even if both are equally effective per minute. Statistics are therefore expressed per 90 minutes, then scaled to the minutes the player is expected to play in this match.

2. Shots and expected goals (xG) per 90 minutes

xG measures the quality of chances. According to Opta, every shot is given a value from 0 to 1, estimated from almost one million shots based on distance, angle, shot type and defensive pressure. A player's xG per 90 sums up the volume of his shots and the quality of his chances. It is more stable than his goal tally, which is very sensitive to short spells of good or bad finishing. Our guide to expected goals (xG) explains this metric in detail.

3. Penalties

A team's designated penalty taker has an extra source of goals. Opta gives a penalty a fixed value of 0.79 xG, reflecting its historical conversion rate. This bonus is modest on average, as a team does not win a penalty in every match.

4. Role and position

A centre-forward playing as the main finisher does not have the same profile as a midfielder tasked with building play. The position he occupies in this match, and any change of system, directly affect the number of chances he can expect.

5. Opponent and context

A defence that concedes many chances raises the expected goals of the whole opposing team, and so those of its forwards.

6. Form and absences

Recent form says something about a player's confidence and role, but a handful of matches is a fragile sample. Absences work both ways: an injured or suspended player cannot score, and a team-mate's absence can redistribute chances and penalties.

Worked example with the Poisson distribution (fictional figures)

A player's goals are rare events. If λ is the number of goals he is expected to score in the match, the Poisson distribution gives the probability that he scores none: e^(−λ), with e ≈ 2.718. The probability that he scores at least once is therefore:

P(at least 1 goal) = 1 − e^(−λ)

Take an imaginary striker:

non-penalty xG: 0.45 per 90 minutes;
expected playing time: 80 minutes, so 0.45 × 80 / 90 = 0.40 xG;
he takes the penalties, and we assume his team wins one in 10% of matches: 0.10 × 0.79 ≈ 0.08 extra xG;
total: λ ≈ 0.40 + 0.08 = 0.48.

Calculation: e^(−0.48) ≈ 0.619, so P(at least 1 goal) ≈ 1 − 0.619 = 0.381, or about 38%. The same calculation applied to other scenarios:

Scenario	λ	P(at least 1 goal)
80 minutes, no penalties	0.40	1 − 0.670 ≈ 33%
80 minutes, with penalties	0.48	1 − 0.619 ≈ 38%
Comes off the bench for 30 minutes, no penalties	0.15	1 − 0.861 ≈ 14%
Very open opponent (0.60 xG per 90), 80 minutes, with penalties	0.61	1 − 0.543 ≈ 46%

Three takeaways. Even in the most favourable scenario, the player has less than a one-in-two chance of scoring. Playing time weighs as much as talent: 30 minutes instead of 80 cuts the probability by more than half. Finally, the order of magnitude matches reality: with Haaland's 2025-26 average (27 goals in 35 matches, about 0.77 per match), the formula gives 1 − e^(−0.77) ≈ 54%, close to the 57% of matches in which he actually scored.

The limits of the calculation

Playing time is an assumption. Official line-ups are only known shortly before kick-off.
The Poisson distribution simplifies the match. It assumes a steady rate of chances; a red card, an early goal or a tactical change alters everything.
A single match neither proves nor disproves a probability. At 38%, the player would fail to score in roughly 6 matches out of 10.

How Prediction AI presents likely goalscorers

Prediction AI is a predictive sports analysis platform powered by AI. For each team in a football match, the tool searches the web for its top scorers and assist providers of the season, across all competitions: goals, assists, matches played and availability (available, doubtful or absent).

The AI model then lists 2 to 4 likely goalscorers, able to score at any point in the match, ranked from most to least likely. Each comes with a probability and a short justification backed by a figure. The estimate takes into account the player's recent goals, whether he takes penalties and the opposing defence. An absent player never appears; a doubtful player may appear with a reduced probability, flagged as such. Probabilities are framed to stay realistic: even a star striker in top form rarely goes above 55%.

These probabilities are the AI model's estimates based on those facts, not a shot-by-shot xG calculation like the example above. The analysis also shows the sources consulted and a data quality indicator. See also our guide on the correct score. Explore the Prediction AI plans.

Frequently asked questions

Will a likely goalscorer definitely score?

No. Even a high probability leaves plenty of room for failure: at 40%, the player would not score in roughly 6 matches out of 10.

Why does playing time matter so much?

Because chances build up with minutes. In the example, the same player drops from about 33% over 80 minutes to about 14% if he only plays 30.

Is the most likely scorer always the centre-forward?

Often, because he gets the most chances and sometimes takes the penalties. But an attacking midfielder who takes penalties, facing a fragile defence, can overtake him.

Probabilities describe uncertainty: they do not ensure any result.