Home-Advantage Coefficient Is Decaying in Franchise T20 — The Driver Isn't the Crowd, It's the Schedule

The row in my logbook read like this: the home side 142/3 after 18.3 overs, needing 41 from 24. Seven wickets in hand, a set batter at the crease, and the opposition's lead spinner had already bowled three of his four overs. The next twenty-six deliveries produced four dot balls, two boundaries and one needless run-out. Match over.

That evening thirty-two thousand people in the ground went quiet at the same moment. I marked the fixture in my logbook in exactly the colour I have used for 237 matches over the past decade — a decision taken in front of a scoreboard is not really a decision, because it was made long before the scoreboard asked the question. One match is not evidence. Eighteen hundred matches are a question.

For eight years I have worked on a coefficient I call the Home-Advantage Residual. My database holds 1,842 franchise T20 matches across six leagues from 2026 to 2026, each with ball-by-ball timestamps, field-set stills and scorecards. I split every innings into four layers: powerplay (1–6), middle (7–15), death (16–20), and above them an environment layer — toss, dew, pitch reuse, travel distance, and the number of days between matches.

The method was born in London in August 2026. I was writing for a betting syndicate and had published a report predicting Burnley's relegation. The model had their xG differential at -12.4 and a forty-point finish. Burnley finished seventh and qualified for the Europa League. I went back through all thirty-eight matches and found that set-piece xG (+6.8) and post-shot goalkeeper xG (+4.2) were not in my equation at all. The Burnley model broke, and I rebuilt it one clean row at a time.

In May 2026 the Bundesliga returned to empty stadiums. Across the first three matchdays the home win rate fell from 43% to 21%. I cut home advantage by 0.35 goals and tracked it for six weeks, returning 12.4% ROI. The Bundesliga came back, and the silence rewrote every home-advantage coefficient I owned.

That translation has conditions in cricket, and I state them plainly. A football match contains roughly a thousand passing events; a cricket innings contains 120 balls, and a single delivery can reverse the direction of the whole match. Per-event variance is far higher here, which means home advantage has less time to compound. France taught me that a low block is just a different kind of data; cricket's powerplay pressure is not a football defensive block, it is a separate grammar. I never paste football coefficients onto cricket — I translate the question.

Now the numbers. Home win rate in my sample was 57.4% in 2026–17, 52.1% in 2026–23, and 51.3% in 2026–25. In runs the decay is sharper. Net home advantage — home run rate minus away run rate — was +0.42 per over in 2026–17 and is +0.11 in 2026–25. Roughly a fourfold compression in seven seasons.

Phase splits tell an uneven story, and the unevenness is the real finding. Home batting advantage in the powerplay has fallen from +0.38 runs per over to +0.09. Yet at the death, home teams are 0.61 runs per over better with the ball than visitors, and 0.33 runs per over worse with the bat. The ball still works at home; the bat does not. My reading: chasing adds scoreboard pressure, and that pressure is what breaks equations like 41 from 26. In the powerplay, a young home batter on a small ground wants to take the big option early, because the cost of risk feels lower in front of a familiar crowd.

Pitch reuse is the second contributor. In the second match of a double-header, the average first-innings score drops 6.4 runs and the chasing side's win rate rises 4.8 percentage points. Across seven venue curation reports I found that a strip used a second time gains roughly 0.9 degrees of spin turn while the bounce flattens — the slog and the cut lose value. The home team certainly knows how the surface will behave, but that knowledge cuts both ways: the side batting second holds both the information and a defined target.

Third: teams playing with three days or less between matches show a home win rate of 48.9%, 2.4 points below the baseline. Stripped of data jargon this is physiology. An overseas quick is flown in for a three-match window — routine in Tim Southee's franchise calendars — and the bill is paid in the last six overs of the second innings. A franchise spinner like Rashid Khan works three leagues, three continents and three different surfaces in a single year, and his spin-tracking numbers shift with every move. I have never stood inside a physio room, but the link between schedule density and bowling load reads clearly without extra proof. Home advantage exists when everything else is equal; on a compressed calendar nothing else is equal.

Geometry matters too. Where boundary dimensions are under 65 metres, the home powerplay strike-rate advantage is 0.29; on the bigger grounds it is 0.06. On small grounds there is less room for a spinner-catcher, and catching positions differ — that geometric memory forms after five or six games a year.

Then the fourth variable, where the biggest mistake lives: the crowd. In my sample, LBW decisions go the home side's way 0.31 times per innings against 0.27 for the visiting side — a gap of 0.04. Crowds returned after 2026, but only 0.02 of that gap returned relative to the pre-pandemic level. If the crowd were the primary engine, nearly all of it would have come back. It did not.

Dew is misread as well. In late-start matches the second innings carries a 7.2-point higher win rate, but this happens regardless of home or away. Dew is not a home advantage; it is a chasing advantage.

Squad construction adds another layer. Where the overseas quota is fully used, home advantage is effectively zero (+0.04); where one or two overseas slots go unused, it rises to +0.24. The more of a home core a side fields, the deeper its feel for local conditions — but modern franchise calendars rarely let that familiarity persist.

Test cricket tells a completely different story. In the five-day format home advantage still sits around 53%, because there is time to learn conditions and because swing-and-spin environments give local bowlers an unequal edge. Home advantage has not died in cricket; the T20 calendar is slowly erasing it.

This is where I have to stop, because my own model has set a trap. When a residual falls, people assume the cause has been found. It has not. My home-advantage coefficient is a residual — what the model could not explain, banked rather than understood. The Burnley lesson of 2026 was exactly this: the variable outside the model decides the outcome. League parity has risen, squad rotation has risen, travel logistics have improved, and video and data sharing have levelled preparation. None of that happens in front of a camera, so none of it becomes a headline.

There is a further branch: local interest in pitch curation. Home management talks to the curator regularly, but a curator's career is built on the character of the surface, not on any single result. That tension makes the coefficient noisier, and it may explain why two matches at the same venue can produce completely different coefficients.

One last thing, which took me eight years to write down. Evidence and explanation are not the same object. Falling home win rates are evidence. The crowd is not the cause is my explanation, and explanations change. I no longer treat the model as prophecy; I treat it as a confessional, and beside every coefficient I write what is missing. Beside this piece I am writing the same list: injury reports, internal bowling workloads, a curator's mood, airport delays. None of them are in my equation.

What I will watch next round is clear. Is there a double-header, is there a turnaround under three days, and how wide is the night dew window? If those three boxes tick, I will stop giving the home side a separate edge and move that edge into the toss column. The last row of my database still reads a locked gate at 140. The question is not whether home advantage is dead. The question is which variable I have not measured yet.