Premier League: Are the Odds Accurate?

3,800 matches. 11 seasons. Max calibration gap: 2.7pp across five odds clusters, 3.5pp across ten deciles. The EPL is the best-calibrated betting market in European football.

Matches

3,800

Goals/match

2.83

Home win

44.6%

Draw

23.2%

Away win

32.1%

Max Δ (decile)

3.5pp


Season-by-season overview

SeasonnGoalsHomeDrawAwayO2.5BTTS0:0
2016/173802.8049.2%22.1%28.7%54.2%50.8%7.1%
2017/183802.6845.5%26.1%28.4%51.1%48.9%8.4%
2018/193802.8247.6%18.7%33.7%53.7%51.3%5.8%
2019/203802.7245.3%24.2%30.5%52.1%51.1%5.5%
2020/213802.6937.9%21.8%40.3%50.0%48.9%7.9%
2021/223802.8242.9%23.2%33.9%53.9%50.0%5.8%
2022/233802.8548.4%22.9%28.7%52.6%51.6%6.1%
2023/243803.2846.1%21.6%32.4%64.7%61.6%2.9%
2024/253802.9340.8%24.5%34.7%56.6%57.4%4.2%
2025/263802.7542.6%27.4%30.0%55.0%56.1%7.1%

2020/21 — COVID, behind closed doors. Home advantage collapsed to 37.9%, away teams won 40.3%. 2023/24 — goal explosion: 3.28/match, only 2.9% goalless draws. 2016/17 — last season with home advantage above 49%.


Home advantage is declining

From ~49% to ~41% over 11 seasons. The trend predates COVID. The 11-season average is 44.6%, but the last three full seasons average 43.2%.

Possible causes: VAR reducing referee home bias, improved away preparation via video analysis, standardized pitch conditions. The data confirms the trend without proving the cause.


Market calibration: implied vs actual

3,800 matches split into 10 deciles by implied home win probability.

Implied H%nImpliedActualΔ
4–17%38012.5%10.8%−1.7pp
17–26%38021.5%25.0%+3.5pp
26–32%38029.1%27.9%−1.2pp
32–38%38035.1%37.4%+2.2pp
38–42%38040.0%39.5%−0.5pp
42–48%38045.1%44.2%−0.9pp
48–55%38051.1%54.2%+3.1pp
55–63%38058.3%58.4%+0.1pp
63–73%38067.7%69.5%+1.8pp
73–91%38079.8%79.5%−0.3pp

Maximum decile gap: 3.5pp. Grouped into five broader odds clusters, the maximum gap is 2.7pp — the figure used in cross-league comparisons. Neither is systematically wrong enough to overcome the ~4.1% bookmaker margin. The EPL 1X2 market is efficient.


Most common scores

#ScoreCountFreq
11:141410.9%
21:03338.8%
32:13158.3%
42:02907.6%
51:22657.0%
60:12596.8%
70:02316.1%
80:22065.4%
92:22025.3%
103:11915.0%

Top 3 (1:1, 1:0, 2:1) = 27.9% of all results. Top 10 = 71.2%. Correct score bet on 1:1 wins 10.9% of the time.


Goals markets (Over/Under + BTTS)

ThresholdFrequency
Over 0.593.9%
Over 1.578.3%
Over 2.554.4%
Over 3.531.9%
Over 4.515.4%

BTTS: 11-season average 52.8%. Stable at 48–52% until 2023/24 (61.6%) pushed it higher. 2025/26 finished at 56.1%.


Halftime to fulltime

HT result→ Home→ Draw→ Away
Home winning78.3%15.0%6.7%
Drawing36.8%34.7%28.5%
Away winning11.3%17.2%71.5%

Home comeback rate: 11.3% (1.7× higher than away at 6.7%). Level at HT → home wins 36.8% vs away 28.5% — the clearest residual home advantage.


Corners, cards & referees

Corners ↔ Goals: r = −0.036 — near-zero linear correlation in full-match totals. This does not test set-piece quality or team-level profiles. Avg 10.4 corners/match. Similar near-zero correlations across all 5 leagues.

Cards: avg 3.50 yellows/match. Red card in 10.7% of matches — the least card-heavy league in Europe.

RefereeMatchesGoals/matchHome%
A Taylor2953.0043.1%
M Oliver2873.0442.5%
C Pawson2292.7541.9%
S Attwell1952.8845.1%
C Kavanagh1932.6344.6%
P Tierney1902.7941.1%

Oliver: 3.04 goals/match. Kavanagh: 2.63. Gap of 0.41 — likely fixture assignment, not referee influence.


Can you profit from betting the EPL?

Bet Home

−3.4%

Bet Draw

−7.2%

Bet Away

−5.6%


ROI breakdown by cluster

ClusternHomeDrawAway
Heavy Fav664−4.0%−3.6%−10.7%
Clear Fav722+0.0%−11.6%−6.8%
Slight Fav706−6.1%−3.5%−2.1%
Even531−3.3%−0.8%−8.0%
Underdog1,177−3.5%−11.7%−3.0%

Best case: Home in Clear Fav at break-even (+0.0% over 722 matches) — pure noise. No reliable strategy beats the margin.

⚠ Retrospective ROI across 11 seasons — not a betting signal. These clusters reflect historical mispricing that the bookmaker margin largely absorbs; the pattern may not persist forward.


Dixon-Coles ρ for the Premier League

Empirical fit: ρ = −0.010 — the smallest correction of any league.

At league-average λ (1.56, 1.28), standard Poisson underestimates 0:0 by 0.26pp. With ρ = −0.010, the gap shrinks but doesn't fully close. Bundesliga needs ρ = −0.096 — roughly ten times stronger.

For most EPL matches, Poisson and Adjusted DC produce nearly identical score matrices. The difference only matters for extreme λ values. Full cross-league comparison →

Data: 3,800 EPL matches, 2016/17–2025/26 (complete). Source: football-data.co.uk. Analysis: ImpliedScore. Last updated: May 2026.

Scroll to Top