Does rain change an AFL game?

Why this was worth a look

“It’ll be a low-scoring one in this weather” is the most confident sentence anyone says in a Thursday night group chat, and I had eight seasons of rain readings sitting beside eight seasons of scores, so I went and counted them.

What the rain actually does

Across 1,329 open-air games from 2019 to 2026, the pattern is real and small. Games with no rain recorded average 164 points. Under a millimetre, 156. From 1 to 2mm, 152. From 2 to 5mm, 144. At 5mm or more, 143.

That is a drop of 21 points from the driest band to the wettest, and it arrives in order, which is the part that looks convincing.

The pool includes 2020, whose shortened quarters averaged 121 points across 151 open-air games. Dropping that season lifts every band by five to nine points and leaves the gap between the driest and the wettest at 21. The level moves; the finding does not.

The boxes in the first figure say it better than the averages do. The middle half of dry games runs from 143 to 184 points. The middle half of the wettest games runs from 126 to 159. Those two ranges overlap between 143 and 159, which is another way of saying that a typical wet game and a typical dry game are the same game. One standard deviation across all 1,329 games is 32 points, half again as wide as the 21-point gap the bands are there to show.

The wettest game we hold is Adelaide against Collingwood at Adelaide Oval in Round 18, 2022, with 12.8mm recorded. It finished 86 to 91: 177 points, well above the dry average. The second wettest, on the same rainfall at the same ground, finished 126. Two games, the same reading, 51 points apart.

Docklands is the control group nobody planned. 317 games under a roof since 2019, averaging 169 points against 161 for every open-air game with a reading. That is not proof that a roof raises scoring — Docklands has its own dimensions, its own tenants and its own fixture — but it is the cleanest comparison available, and it points the same way the rain bands do.

The weather that moves the needle is warm, not wet

Sorting the same games by temperature instead of rain spreads them further apart than rain ever manages.

Games under 10 degrees average 150 points. Games at 25 degrees or more average 179. That is a 29-point swing against rain’s 21, from bands holding 97 and 64 games — thin, but no thinner than the wet end of the rain scale.

This is not a claim that heat makes goals. Hot games are played in different months, in different cities, and often by different clubs; the temperature is a label on those games, not a cause of the scoring. It earns a mention because it is a bigger signal than the one everybody talks about, and because nobody has ever said “warm one tonight, expect a shootout”.

Wind does less than either. Dry games in under 10km/h winds average 166 points, and dry games in 20 to 30km/h average 164. Only 9 dry games were played in more than 30km/h. Those 9 averaged 152, with a winning margin of 44 against the usual 31, which is an interesting number drawn from nine games and stays an interesting number.

Where the rain lands

Rain is not spread evenly across the grounds, and neither is what it appears to do to them.

Perth Stadium is the sharpest split: 53 wet games averaging 142 points against 164 in the dry, a 22-point gap. The MCG, with 115 wet games, drops from 165 to 155. Two grounds go the other way — the SCG averages 171 in the wet against 168 dry, and Sydney Showground 181 against 170 — on 16 and 14 wet games each, a sample that can point anywhere it likes.

The NRL answers the same question the same way. No rain, 47 points a game; under 2mm, 44; 2 to 5mm, 42; 5mm or more, 37, from 18 games.

Punt carries the same split for one player at a time, on every player page. What a wet-weather split can tell you about a single player, and how seldom it can tell you anything at all, is its own article.

What this cannot tell you

The reading is the day’s rainfall at the ground, added afterwards from a weather service, not a measure taken at the first bounce. A downpour at 6am and a downpour at the first bounce are the same number here, and they are not the same game. If the feed ever carries rain by quarter I would run this again, and my guess is that the gap widens a little and the samples stay just as thin.