Corner Kick Calculator
Predict corner kick totals using the Poisson distribution. Estimate expected corners for each team, over/under probabilities, and corner race markets.
Over / Under Probabilities
| Threshold | Over % | Under % | Fair Over Odds | Fair Under Odds |
|---|---|---|---|---|
| Over 7.5 | 80.6% | 19.4% | 1.24 | 5.14 |
| Over 8.5 | 70.0% | 30.0% | 1.43 | 3.33 |
| Over 9.5 | 57.9% | 42.1% | 1.73 | 2.38 |
| Over 10.5 | 45.4% | 54.6% | 2.20 | 1.83 |
| Over 11.5 | 33.8% | 66.2% | 2.96 | 1.51 |
| Over 12.5 | 23.8% | 76.2% | 4.21 | 1.31 |
Corner Race (Home vs Away)
Poisson Corner Model
Corner kicks follow a Poisson distribution, similar to goals. Each team's expected corner count is the average of their own corners for and the opponent's corners conceded:
Expected Home = (Home_Corners_For + Away_Corners_Conceded) / 2 Expected Away = (Away_Corners_For + Home_Corners_Conceded) / 2 P(k corners | λ) = e^(-λ) x λ^k / k! Corner Race: P(Home > Away) = sum over all h > a of Poisson(h, λH) x Poisson(a, λA)
The model assumes corners are independent events — a reasonable approximation for pre-match analysis. Actual corner counts can be influenced by match state (teams trailing push for more corners), possession styles, and opposition tactics.
League averages: Bundesliga 11.2, Premier League 10.5, La Liga 9.8, Serie A 9.5, Ligue 1 9.9. Use these as baselines and adjust for team-specific data.
Corners as a Poisson Process
A corner kick is a textbook counting-process event: a discrete occurrence that arrives at a roughly steady rate over a fixed interval. A Premier League match averages around 10.5 corners across 90-plus minutes — roughly one every nine minutes of play. Whenever you count independent events at a known average rate, the Poisson distribution is the natural model. It is the same machinery used to price goals, just running at a much higher rate — and because the rate is high (λ ≈ 10 instead of λ ≈ 1.4), the distribution of totals is wider and more symmetric than a goal distribution.
The rate estimate is the whole game. A team's corner count in a given match is driven half by what its own attack generates and half by what the opponent's defence gives up, so the calculator blends both: expected home corners = (home corners for + away corners conceded) / 2. Style matters more for corners than for goals — crossing teams with overlapping full-backs win far more corners than possession sides that work the ball centrally, and deep defensive blocks concede corners constantly because every clearance and blocked shot restarts the pressure cycle.
For totals, sum the Poisson tail above the line: P(Over 10.5) is the probability of 11 or more corners. Half-ball lines cannot push, so the over and under probabilities always sum to 1 and fair odds are simply their reciprocals. Note how wide the distribution is: with λ = 10.3, the standard deviation is √10.3 ≈ 3.2 corners, which is why even the "obvious" Over 8.5 in an average match is only about a 70% shot, not a certainty.
Race markets are where the model earns its keep. "Most corners" treats each team's count as an independent Poisson and sums the probability over every scoreline where one side out-corners the other — with the tie as a genuine third outcome, which is why the home and away prices alone never add to 100%. "Next corner" and first-to-N races follow from a neat property of competing Poisson processes: each successive corner belongs to the home team with probability λ_home / (λ_home + λ_away), independent of the score so far. A first-to-5 race is then just a best-of sequence of those weighted coin flips, and corner handicaps come off the same home-vs-away grid. Convert any probability gap you find into a stake decision with the EV calculator.
One honest caveat: real corner data is slightly over-dispersed — the variance runs a little above the mean because corners cluster. A trailing favourite in the last twenty minutes can win four corners in five minutes as pressure compounds. The plain Poisson model therefore under-prices extreme totals a touch, so treat far-out lines (Over 13.5, Under 6.5) with more caution than the middle of the distribution.
Corner Model Formulas
λ_home = (HomeCornersFor + AwayCornersConceded) / 2 λ_away = (AwayCornersFor + HomeCornersConceded) / 2 λ_total = λ_home + λ_away P(k corners) = e^(−λ) × λᵏ / k! P(Over t.5) = Σ P(k) for k ≥ t + 1 Most corners (race): P(Home most) = Σ P(h; λ_home) × P(a; λ_away) over h > a Next corner / first-to-N: P(next corner is Home) = λ_home / (λ_home + λ_away)
Worked Examples
Default inputs give λ_home = (5.3 + 5.0) / 2 = 5.15 and λ_away = (5.2 + 5.1) / 2 = 5.15, so λ_total = 10.3. The model prices Over 9.5 at 57.9% (fair 1.73) and Over 10.5 at 45.4% (fair 2.20); the most likely exact total is 10 (12.5%). A hypothetical book price of 1.90 on Over 9.5 would be a +10% EV bet: 0.579 × 1.90 − 1 = +0.10.
Suppose the blend gives the home side 6.2 expected corners against 4.1. The race market prices out at Home most 69.1% (fair 1.45), Tie 10.3%, Away most 20.7% (fair 4.83). Each individual corner is home's with probability 6.2 / 10.3 ≈ 60% — a useful anchor for next-corner and first-to-N prices in the same match.
Frequently Asked Questions
Why do corner kicks follow a Poisson distribution?
Corners are discrete events that arrive at a roughly steady rate through the match — a Premier League game averages about 10.5 corners, roughly one every nine minutes of play. Counting independent events occurring at a known average rate is exactly what the Poisson distribution models, which is why the same math used for goals prices corner totals well.
How do I estimate expected corners for a match?
Blend each team's corners won with the opponent's corners conceded: expected home corners = (home corners for + away corners conceded) / 2, and likewise for the away side. Use at least 8-10 recent matches, keep home and away samples separate, and nudge the number up for crossing-heavy teams and for matches where a favourite is expected to camp in the opposition half.
How are corner race markets priced?
For 'most corners', model each team's count as an independent Poisson and sum the probability over every grid cell where one side out-corners the other — the tie is a real third outcome, which is why home and away prices don't add to 100%. For 'next corner' or first-to-N races, each successive corner belongs to the home side with probability λ_home / (λ_home + λ_away).
What is a good corner over/under threshold?
Anchor on the league baseline: the Bundesliga averages about 11.2 corners per match, the Premier League 10.5, Ligue 1 9.9, La Liga 9.8 and Serie A 9.5. Books typically offer lines from 8.5 to 12.5; the value sits wherever your match-specific expectation diverges from the league average the market defaults to.
What are the limitations of the Poisson corner model?
Real corner counts are slightly over-dispersed: corners cluster when a trailing team piles on late pressure, so the plain model marginally under-prices extreme totals. It also ignores in-match dynamics — red cards or an early goal can change a team's whole approach — and it assumes the two teams' corner counts are independent, which weakens when one side dominates territory.