Every figure carries its n and standard error. A verdict appears only on the favourite / longshot buckets at 2 SE with at least 1,000 sides, and only where it matches all competitions. Nothing here is about a team.
Where the price is wrong · Pinnacle 1X2, each fixture counted once per side · since 2017
Favourites are underpriced: priced 69.8%, won 73.0% (+3.2 pt, 1,062 sides, 2.3 SE)
Longshots are overpriced: priced 16.0%, won 13.9% (-2.1 pt, 2,159 sides, 2.8 SE)
+3.2 ptFavourites (≥ 60%)priced 69.8% · won 73.0%± 1.4 pt · 2.3 SE · 1,062 sides · underpriced
-0.2 ptMid-range (25–60%)priced 40.5% · won 40.3%± 0.8 pt · 0.3 SE · 3,659 sides
-2.1 ptLongshots (≤ 25%)priced 16.0% · won 13.9%± 0.7 pt · 2.8 SE · 2,159 sides · overpriced
the band curve is built for all / domestic / cups over the full and 5-year windows only — thinner cells would be noise
Calibration · price vs what happened, per outcome
Match result (1X2) · Pinnacle · since 2017 · 3,440 fixtures · 70% closing, 23% T-8h
Outcome
Priced
Actual
Residual
SE
n
Home win
43.0%
41.3%
-1.7 pt
0.8 pt
3,440
Draw
25.3%
25.9%
+0.6 pt
0.7 pt
3,440
Away win
31.7%
32.8%
+1.2 pt
0.7 pt
3,440
Over / under 2.5 · Pinnacle · since 2017 · 3,008 fixtures · 66% closing, 27% T-8h
Outcome
Priced
Actual
Residual
SE
n
Over 2.5
50.7%
51.6%
+0.8 pt
0.9 pt
3,008
Under 2.5
49.3%
48.4%
-0.8 pt
0.9 pt
3,008
Both teams to score · bet365 · since 2017 · 3,243 fixtures · 92% closing, 3% T-8h · bet365 fair price (Pinnacle stopped pricing BTTS on 14 Apr 2026); bet365 runs ~1 pt under Pinnacle on 'yes'
Outcome
Priced
Actual
Residual
SE
n
Both score
51.3%
53.8%
+2.5 pt
0.9 pt
3,243
Not both
48.7%
46.2%
-2.5 pt
0.9 pt
3,243
What each book charges · 1X2 overround (sum of implied probabilities minus 1), median and 90th pct per book · since 2017
Pinnacle 3,4402.7% · p90 3.4%
Unibet 3,2835.4% · p90 6.6%
bet365 3,1715.6% · p90 6.4%
Ladbrokes 2,7086.4% · p90 7.2%
WilliamHill 3,1316.6% · p90 9.7%
Betfair 3,2387.1% · p90 10.8%
Market vs the provider's model · log-loss on the 90-minute result, fixtures both priced · since 2017 · lower is better
0.9608Pinnacle log-loss2,685 fixtures
0.9847Provider log-losssame fixtures
+0.0239Provider minus market6.8 SE
How the market has priced home advantage · Pinnacle 1X2, price the market put on each outcome vs what happened, Jul–Jun seasons; the current season is partial
Season
Home priced · actual
Draw
Away
n
2025/26
41% · 39%
27% · 26%
32% · 35%
380
2024/25
42% · 40%
27% · 28%
31% · 32%
380
2023/24
43% · 42%
26% · 29%
31% · 29%
380
2022/23
43% · 42%
26% · 26%
31% · 32%
371
2021/22
44% · 39%
24% · 26%
32% · 35%
380
2020/21
42% · 42%
24% · 25%
34% · 34%
477
2019/20
44% · 40%
25% · 22%
31% · 37%
282
2018/19
44% · 43%
25% · 30%
31% · 27%
346
2017/18
45% · 43%
24% · 22%
31% · 35%
376
Coverage · Pinnacle 1X2 fixtures per line regime; movement is coverage only — not evaluated until the series holds a season
Closing line (within 2 h of kick-off)2,425 · 2017-05-14 → 2026-09-14
T-8h line (2025 onward prematch)803 · 2017-05-13 → 2026-05-24
Other (within the 24 h guard)212 · 2017-05-13 → 2025-02-06
Movement series (hourly + final hour)40 · from 2026-08-22 · not evaluated