The Third Fielder Called Dew: An Audit of IPL's Home Advantage and the Variable No Spreadsheet Can Hold

An April night at the Wankhede. The fourteenth over of a chase. The captain takes his spinner off and hands the ball to the umpire to be towelled. Nothing on the scoreboard looks unusual. What is unusual is outside the ground, on the weather app: dew point 21.4°C. I had every match of that season. When I separated the games where the dew point stayed above 20 during the second innings, the gap was clean — middle-over scoring in the chase ran roughly 0.9 runs per over higher in those matches, and in dry games at the same venue the gap disappeared entirely. That was less a discovery than a question, and I had spent years calling the question by the wrong name: home advantage.

I have watched from the same Wankhede gallery for fourteen years, the one where the spinner's release and the dew-sheened ball stay in a single frame. What I see there never reaches the scorecard: the bowler's thumb turning wet over by over, the keeper's glove ruined after a third spell, the captain's face at the twentieth over when he understands he answered the wrong question at the toss. This piece is an audit of one thing — how many separate variables we have bundled under the phrase "home advantage" in the IPL, and how few of them exist on any dashboard we own.

METHOD BOX: how this number was built

Dataset: 1,100-plus IPL matches, 2026 to 2026, ball-by-ball, with venue, toss, innings order, local weather (dew point where available) and travel distance from the previous match city.

Three eras I never merge: Era-1 (2026–2026, eight teams, no Impact Player), Era-2 (2026–2026, ten-team expansion and auction inflation), Era-3 (2026 onward, Impact Player rule).

Published weights, v3.2: scheduling 0.34, pitch allocation 0.28, dew/humidity 0.21, travel 0.09, crowd 0.08. These weights move season to season, and every change is timestamped in the file.

What the model cannot see: the curator's intent, ticket pricing, a player's private circumstances, dressing-room politics, and the effect of a rain forecast on mental preparation. The open model worked because it admitted what it could not see.

CONTEXT: we are naming the wrong variable

Every analyst knows home teams win slightly more. In the IPL they do. The question is why — and that is where the conversation stops. The commentary box supplies "confidence on their own ground" or "the crowd behind them". Neither claim can be measured, dated or retracted.

My position is old and I repeat it: Indian cricket's data desks won the tactical argument years ago while the TV box still sells unfalsifiable vocabulary. The fix is not more data. The fix is transparency — what went into the variable, what was left out, and which rule era applied. A clean dashboard is not evidence; it is a question wearing a suit.

CORE: three layers, and the cracks between them

Layer one — scheduling. Nobody writes it down, which is the problem. The IPL play-off format hands Qualifier 1 and the Eliminator to the top two teams at their home venues, meaning the strongest sides get the most "home" matches at the peak of their form. Aggregate the season and you credit that window to the ground rather than to the cricket that earned it. Scheduling alone carries 0.34 of the weight in my model — a third of home advantage sits in the calendar, not inside the walls.

Layer two — pitch allocation. This is where the data desk goes quiet and the curator starts talking. Chepauk's slow, scuffed surface threads a rope for the home spinners and not for the visiting quick. Dharamsala's altitude, around 1,457 metres, rewrites both swing and the batsman's lungs. Chinnaswamy's bounce and short boundary make 200 unsafe. Mohali's dimensions and wind combine into something a visiting side cannot read in two overs. None of this is tactics. It is geography and administration, and it carries 0.28 of the weight — the second-largest share belongs to the man running the roller four days out.

Layer three — dew. Everyone sees it; nobody measures it. Dew does not merely wet the ball; it shifts the centre of gravity of an entire innings. The first-innings spinner becomes unusable after the break, the captain leans on the seamer for yorkers, and every throw from the fielding side arrives slippery. Dew is a fielder with no place on the scorecard, and it plays every evening match.

CORE: the 2026 natural experiment, and what it cost

In 2026 the whole IPL moved to the UAE — sixty matches, three venues, Dubai, Abu Dhabi, Sharjah. It was the cleanest natural experiment the league has produced, because every match was neutral. Isolate that season and home advantage on points-per-game collapses into statistical noise. That was the moment I understood I had been calling a bundle of three things by one name, and two of them — pitch allocation and crowd — had simply been removed.

The second lesson was between venues. Sharjah's boundaries were visibly shorter than Dubai's and the effect showed directly in scoring rates. My original note recorded the gap per ball; a later check put the unit per over. That is the mistake I scold others for, so I log it against myself: the unit changed and the conclusion did not, but trust does, unless you write the correction back.

One clarification matters here, because this is where most of the cheating happens: the UAE season was not a zero-crowd experiment. There were no spectators, but travel patterns changed, the heat schedule was different, and above all nobody owned a pitch. Remove all of that at once and you cannot say which variable did the work. My 2026 conclusion — that home advantage is mostly scheduling and pitch allocation — pointed the right way, but I wrote it with more certainty than the data held. Two editors sent it back. A digital outlet ran it. The file sits in my private log of pieces that turned out right and were refused first. It gives me no comfort. It reminds me that being right and being able to prove it are different jobs.

CORE: the Impact Player changed every old chart's denominator

From 2026 the Impact Player rule arrived, and anyone stacking 2026 and 2026 run rates on one graph to claim "improvement" is doing bad work. The rule gave teams a twelfth player, but its use is not symmetrical. Batting first, it usually arrives as an extra batsman and adds five or six runs at the death. Chasing, it arrives as a sixth bowler and pays off in the middle overs, where the spinner used to be held back. Era-3 home advantage is not comparable to Era-2 — the denominator changed, not the numerator.

I keep one rule for every comparative table: when the rule era shifts, refuse the chart. Cross-era comparison is arithmetic dressed as insight. A model without a story is just arithmetic; the 2026 U-17 World Cup taught me that in a Guwahati hotel room, where a possession-value model built from scratch had England's Rhian Brewster as the tournament's most efficient shooter two rounds before the final. The model was right. Published without the story, nobody read it.

CORE: selection is the binding constraint, and Ranji is an information market failure

If home advantage lives largely in pitch allocation and dew, the real question is not how a visiting team wins. It is who owns the local information. In India, the largest reservoir of that information is the Ranji Trophy, where hundreds of players perform in conditions nobody publishes ball-by-ball.

I read this as a market failure, not a romance problem. When a tournament's data appears only at the knockout stage, selectors have no comparative basis and decide on the one innings they happened to watch. India's second XI is the hardest team to make in world sport — easy to assemble, hard to select, precisely because we cannot compare players performing on three different pitches in three states in the same week. The same logic prices the auction: when a franchise spends a large purse on a name, the right question is not whether the player is good but how far his price sits above the base rate of actual output — and that base rate cannot exist while domestic data stays private.

CONTRARIAN: treating the crowd as a variable is lazy, and I have been lazy

Here is the part I like least and need most. In 2026 I used the neutral venue as a proxy for an empty stadium. It was not one. A neutral venue changes travel, accommodation, pitch identity and weather at the same time. Change four variables together and you cannot pick one out. I blamed the crowd when the crowd was probably the smallest part of the bundle.

When the crowd vanished, the model kept its shape and lost its meaning. If spectators really were the dominant variable, home advantage in the UAE would have gone negative, because pitch allocation was gone too. What appeared was zero — and zero does not mean the crowd does not matter. It means our measurement still does not record it separately.

Being the market sceptic is a reputation, and reputations want feeding. The lazy version is dismissing the curator's testimony, the ticketing manager's experience and the empty-seat count as noise. I counted the empty seats and found a variable no spreadsheet could hold. At one Pune match the upper tier was bare while the pavilion side was full; the board published one attendance figure. The number was not false. It was incomplete. Noise is the cheap version of my own method, so I keep those variables in an explicit log instead.

None of this article's inputs comes from a franchise insider. The data is public scorecards, public schedules, public weather. Where I went beyond the public record I measured it myself; where I could not measure it, I have said so. A franchise that gives me data wants something back — usually a frame in which its auction error stays invisible. When the relationship is too close to test, hand the piece to a younger desk.

TAKEAWAY: what to watch next cycle

I have no forecast, and if I had one I would not publish it. I have a range, a sample, and the conditions under which the frame breaks. Three signals matter next: public, match-level dew-point data, which would move the toss debate out of captaincy instinct and into the weather column; the Impact Player usage curve, split between teams treating it as a twelfth batsman and teams treating it as a sixth bowler; and ball-by-ball Ranji publication, which is a pricing issue before it is a journalism issue. Whoever still asks why a team won should first answer a smaller question: what was the dew point that night?