The 462-Match Ledger: What Actually Drives Home Advantage in the BPL

I opened the hand-coded season again, and the margins disagreed.

The ledger covers four seasons. 462 matches, one row each, seven columns per row — date, venue, match number, toss, first-innings score, over-by-over run rate, attendance. When the league stopped in March 2026, I did not write op-eds. I re-coded the old rows instead, because without an attendance baseline I had no right to interpret anything that came later. It took fourteen months. Those fourteen months taught me that an empty row is not a zero.

What the disagreement looked like: a January 2026 evening at Mirpur. Rain arrived in the ninth over. The match collapsed to fifteen overs a side and the target was revised. An opener made 41 off 33 — a strike rate of 124.2. That number then travelled to four places: two newspaper scorecards, a fantasy app database, and a club media note. Nowhere does it say the innings was fifteen overs long, on a night when par was somewhere near 140. A number running loose without its denominator is the thing I fear most in this trade.

So this piece comes from that ledger, and it opens with one question: in the BPL, where does home advantage actually come from? The crowd, the pitch, or the wrong column on a fixture sheet?

Context: seven teams, three venues, an uneven calendar

The BPL began in 2026. Its window runs December to February — Bangladesh's winter, when dew settles on the Mirpur outfield and the ball slips wet into a batter's hands. Seven teams, a double round-robin, 42 league matches plus four playoffs: 46 games is a full season.

Three grounds, three characters. Mirpur is slow, spin-friendly, and at night the dew betrays the seamers. Chattogram, with the sea breeze, bats well in the afternoon but the ball holds in the second innings. Sylhet scores highest of all.

Across the four seasons I coded, every ball carries five fields: runs, wicket, bowler type (spin or pace), phase (1–6, 7–15, 16–20), and shot direction. The ledger has three versions — v1.0 (2026), v2.3 (2026), v3.1 (2026) — each with a date and a stated reason for revision. I keep version history because someone will use these numbers after me, and they deserve the chance to correct me.

That is where the first problem sits. My ledger records venues by name; the fixture sheet records them as home and away. One is geography, the other is administration. They are not the same thing — and that gap is what I spent the rest of this piece chasing.

Core: denominator first, story second

Version 3.1 of the ledger shows: in evening innings, the first six overs at Mirpur go at 7.42 an over, Chattogram 8.31, Sylhet 8.76. In afternoon matches Chattogram climbs to 8.64, because the dew has not arrived.

Death overs — 16 to 20 — invert the picture. Evening economy at Mirpur is 9.8, Chattogram 10.9, Sylhet 11.4. Read together: at Mirpur you must start with patience and finish with care; at Chattogram the start is easy and the finish is a furnace.

Then spin. Across all death-over spin deliveries in the ledger, spinners average 7.9 an over at Mirpur, 9.4 at Chattogram, 10.2 at Sylhet. Where the ball stays dry, spin is a death weapon. Where dew sits, a wrist-spinner loses his grip entirely — and that is Sylhet's batting capital.

Pace runs the opposite way. A bowler operating above 140 kph loses his advantage twice over in dew: the batter cannot see it, but it does not hit the stumps either. In Mirpur's evening 16–20 phase, wides per over run 38% higher than in the afternoon. That is a direct measure of control, and it is the column most scorecards throw away.

One record belongs here, because it defines the boundary of my model: on 7 February 2026 at Mirpur, Fortune Barishal beat Chittagong Kings in the BPL final — their second title, after the first on 1 March 2026 against Comilla Victorians at the same ground. Two finals, one venue, one set of evening dew. Barishal won both, by different routes.

Now the home-advantage arithmetic. In v1.0 — when crowds were full — home teams won 43.7%. In the fanless or limited-crowd season that fell to 37.9%. The easy conclusion: the crowd is the engine, noise makes runs.

But in v3.1 I reconciled the columns by hand and found 23 matches labelled 'home' between 2026 and 2026 were actually played at Mirpur, not at the team's own ground — broadcast allocation, security schedules, venue readiness. Strip those out and home win rate lands at 41.2%, only 4.3 percentage points off baseline. Roughly a third of the crowd's magic turns out to be a mislabelled column on a fixture sheet.

The toss numbers demand even more care. In evening matches, teams that win the toss and field win 58.4% of the time; in afternoon matches, 49.1%. The gap is dew. But the denominator matters: 274 evening matches against 141 afternoon ones. On that sample, 49.1% is not statistically distinguishable from 50% — the confidence interval runs 45.9 to 52.3. Anyone who reads that row and writes 'the toss is irrelevant by day' has left the sample size out of the column.

And those nine matches. Abandoned to rain, or stopped before a ball was bowled. The official BPL record files them as 'no result'. In one club-linked spreadsheet that reached me, the same matches sit in that club's loss column. This is not an accusation — it is a filing convention, an ordinary misclassification in domestic book-keeping. The measurable effect: that club's win rate reads 45.5% when the truth is 50.0%. Two numbers, one reality, and the decision is being taken by whichever column holds the error.

I now split empty cells three ways. True zeros: the match happened, nothing occurred. Missing at random: the match happened, nobody logged it. Unobserved: there was no data, because nobody looked. Collapse all three into 'zero' and the baseline rots — and we build next season's decisions on the rot.

Contrarian: correlation is not causation

The fashionable explanation is that the BPL is a low-scoring league because Bangladesh's batting quality is weak. My ledger cannot agree, because the claim has no denominator. The low scores cluster in Mirpur evenings — 118 matches in that bucket, on a two-paced surface ringed by dew. At Sylhet in the afternoon, the same batters score about 30% higher. Same player, same bowler, different clock. An analyst measuring the league's quality through Mirpur evenings is measuring a venue's time of day, not a league's standard.

There is a quiet cultural layer here that I always write down: scorers in Chattogram and Dhaka do not name the same delivery the same way. Travelling from commentary into the scorecard, a dot ball sometimes lands in the 'beaten' column. When I reconciled raw logs by hand, I found overs the broadcast called tight were actually five wides and two free hits. The raw log remembers the over the scorecard forgets.

My second contrarian note concerns dew. Everyone assumes dew means easy second-innings batting. In my ledger, evening second innings do score faster — 9.1 against 8.0 — but wickets also fall faster after the 15th over, from 0.41 to 0.58 per over. Dew buys runs and buys wickets with the same hand. Since a chase needs runs rather than wickets, the two numbers together suggest dew does not help the chasing side; it forces the chasing side to take risk.

And one limitation I will state plainly, because it is the ledger's most honest lesson. I have described the drop from 43.7% to 37.9% as a crowd effect many times. But cricket never gives us the experiment: same ground, same squads, only the spectators removed. That was observation, not a controlled trial. And observation without a denominator cannot decide anything — it can only leave behind a dated question.

Takeaway: the signal for next over

Three columns I will separate next season.

One: venue allocation in the fixture list. Every 'home' match not staged at the team's own ground gets flagged, and home advantage re-measured with that flag. Whoever skips this step will build a squad on a wrong column. And BPL squads are built at auction, where time is short and valuation is thin. The ledger is patient; the transfer market is not.

Two: the powerplay wicket denominator. Not run rate — wickets lost in the first six overs. In my ledger, sides losing two or fewer powerplay wickets added 47.2 runs in the last five overs; sides losing three or more added 34.8. A 12.4-run gap, and it maps straight onto results.

Three: a dew row. Humidity, the over of the ball change, the scheduled start time — a separate field on every match. Any season without that column will produce the same lazy contextual explanation again, from scratch.

I know nobody will keep these three columns. They are hard, they earn no ratings, they show up on no sponsor deck. The question stays anyway, and it is not a team's question but the record's: the column where we write the venue next season — will we mislabel it again? Results will come, tables will form, stars will be born. And alongside, a misspelt field will sit quietly, holding dew, a clock, and a fragment of truth that never makes anyone's ledger.