Temperature and AFL scoring

Dry AFL matches under 10°C score 9.9 combined points below their own season average across 198 matches (t = -4.59). At 25–29°C the same comparison runs +12.6 points on 118 matches (t = 4.88). Across the whole dry sample every extra degree is worth about 1.1 points — roughly a goal and a half for a 10-degree swing.

Unlike wind, this is a gradient rather than a threshold: every step up the table adds points, and the mild 15–19°C band sits almost exactly on the season average, which is what a clean control band should do. The 30°C+ row holds 19 matches — below our pre-set 40-match minimum, which was fixed before this study was run. It is shown for completeness as directional evidence only and is excluded from our headline, our summary numbers and any lean or pricing claim.

AFL scoring has drifted a long way since 2004, so a temperature result could easily be a disguised season trend. Two checks say it is not. Every match is measured against the average total of its own season, and re-running the same bands without that adjustment moves each qualifying band by no more than about 0.8 points — the under-10°C row goes from -9.9 to −9.8, and the 25–29°C row from +12.6 to +11.9. Cold matches are also spread across the whole season, from round 1 to the finals, rather than bunched into a handful of rounds.

The league-wide numbers are the finding. They are not evenly spread, and pretending otherwise would be the easiest way to mislead with this table.

The MCG carries the largest set of cold matches and shows a modest effect. The biggest ground-level numbers sit on the smallest samples and several are marked below the minimum, so they should not be priced. Grounds with fewer than ten cold matches are left out of the table entirely.

• Source: AFL match archive joined to hourly Open-Meteo reanalysis at each venue's coordinates, in venue-local time. 2,640 dry matches, 2004–2026, 26 grounds, from -0.1°C to 34.1°C.
• Measure: combined final score, compared with the average total of the same season so the long-term drift in AFL scoring cannot masquerade as weather.
• Excluded: any match with 0.1 mm or more of rain in the window, all roofed Docklands/Marvel matches, matches without a resolvable venue, and matches without a final score. Nothing is excluded on the basis of its result.
• Minimum sample: 40 matches, set before this study was run and unchanged for it. Rows under that threshold are labelled and never used in headlines or pricing.
• Not controlled: wind (measured separately in our AFL wind study), humidity, day versus night starts, team strength, travel and finals context. Cold matches cluster in the southern states and warm matches in the north, so some of the gradient is venue and style of play rather than temperature itself.
• No closing totals are available for these matches, so this study reports scoring, not Over/Under hit rates. We will not publish hit rates without the lines.
• Spread inside the sample is ±32.6 points, so temperature moves the average without deciding any single match.

Data generated 2026-09-21. Numbers refresh as the weekly venue-weather top-up runs; corrections are dated and published on this page.

• Open-Meteo ERA5 historical reanalysis — archive-api.open-meteo.com (hourly game-window weather at each venue’s own coordinates)
• AFL Tables — afltables.com (AFL match archive, venues and final scores)
• Internal AFL match archive — venue coordinates, local start times and final scores.
• Every dated change we have made to a published study is listed on our corrections log.