Is footy momentum real?

Basketball's "hot hand" was famously dismissed as a fallacy — then quietly un-dismissed. We ran the same test on AFL goals. Momentum is real. But the warning sign isn't the one most coaches watch for.

New to r-values, p-values and confidence intervals? Stats made simple for coaches explains every number in this piece in plain English — no propellor hat required.

Every footy fan believes in momentum. The three quick goals, the crowd lifts, the other team can't get it out of the centre — a run-on feels like the most real thing in sport. Yet the most famous study in sports statistics said the equivalent belief in basketball — the "hot hand," the player who's "in the zone" — was an illusion. People saw streaks in what was really just randomness.

What's less famous is the sequel. In 2018, two economists found a subtle flaw in that original method: it accidentally understated streakiness, and once corrected, a real hot hand reappeared. So the honest state of the science isn't "momentum is a myth." It's "momentum is real, and we nearly fooled ourselves into denying it."

We put AFL goals to the test.

Goals cluster more than chance allows

If momentum weren't real, goals would fall in a random order — your goal, their goal, your goal — with streaks no longer than coin-flips produce. We counted the number of separate scoring "runs" across our captured games and compared it to what pure randomness would produce.

Randomness would give about 515 separate runs. We saw 469. Goals clump together more than chance allows — a gap this large happens by fluke less than once in a thousand times (a Wald–Wolfowitz runs test, z = −3.37). Footy goals are genuinely streaky. Momentum isn't only in the crowd's head.

And the deeper into a run you go, the more likely it continues:

After one goal, the same team kicks the next about 53% of the time — barely above a coin flip. But after three in a row it's 68%, and a team that's kicked six straight kicks the seventh 81% of the time. The run-on feeds itself. (And remember the 2018 correction: this kind of count slightly understates streakiness, so the true effect is, if anything, stronger.)

The fingerprint shows up in results, too: the winning team's longest goal run averages 6.4 straight; the losing team's, just 2.8 — a large, clean separation (Cohen's d = 1.34). Games are often decided by one team's burst that the other never matches.

But the warning sign isn't speed

Here's where we expected to find the actionable trigger — and didn't. The intuition is that quick goals are the danger sign: three in five minutes means the dam has burst, throw numbers behind the ball and clog it up. We tested it directly.

It's not true. Runs that start fast — two goals inside five minutes — go on to average 3.4 goals. Runs that start slowly — more than five minutes between the first two — average 4.2. The chance a run reaches four-plus goals is essentially identical whether it started fast (36%) or slow (37%). The speed of the goals tells you nothing about how long the run will last.

So momentum is real, but it doesn't announce itself with tempo. What predicts a long run isn't how quickly it began — it's simply that it has begun and kept going. The trigger to act on is the count (they've kicked three on the trot), not the clock (they kicked them quickly).

A fair caveat

Some of this is team strength: good sides score in bunches partly because they're good, not purely because they're "hot." But the runs test controls for that — it compares each game's goal order against that same game's totals, so a lopsided final score can't manufacture the result. The clustering is real beyond what the scoreline alone would produce.

What we can't prove is that intervening works. We don't capture coaching moves, so we can't show that flooding the back half for five minutes actually breaks a run. We can only say the runs are real, they extend, and they decide games — so a deliberate circuit-breaker is a reasonable bet, even if unproven here.

What this means for your team

• Treat the run-on as real, not superstition. Three on the trot genuinely raises the chance of a fourth. Momentum survives a proper statistical test.
• Watch the count, not the clock. A burst of quick goals is no more dangerous than the same goals spread out. The signal to act is "they've kicked three in a row," full stop.
• Have a rehearsed circuit-breaker. A pre-planned way to clog the game up — extra numbers behind the ball, a deliberate slow-down, a positional switch — is worth having ready for when a run starts. The runs are real and they decide games; breaking one early is plausibly high-value (though we can't yet prove the intervention works).
• Build your own runs. The same coin works both ways: winners' runs reach 6+, losers' stall at 2–3. Manufacturing your own burst — and pressing it once it starts — is as much the game as stopping theirs.

In short

• Footy goals are genuinely streaky — they cluster more than randomness allows (a result this extreme happens by chance less than 1 in 1,000 times). Momentum is real.
• Runs feed themselves — the chance the next goal continues a run climbs from 53% after one to 81% after six — and winners' longest runs (6.4) dwarf losers' (2.8).
• Speed is a red herring. Quick goals don't extend runs any more than slow ones. The warning sign is the count, not the tempo — and a rehearsed circuit-breaker is worth having ready.

A note on the data

This pools every goal, in order, across captured local-league AFL games, after filtering out demo data and unfinished matches. The runs test compares the observed number of scoring "runs" to the number expected if goals fell in a random order given each game's totals (so it can't be faked by lopsided scores). Continuation rates and longest-run figures are counted directly from the goal sequences.