No Tracking, No Ledger: How Asian Domestic Cricket Manufactures the Wrong Selections

Sylhet International Cricket Stadium, press box, third row, a 2026 BPL evening. Every ball I log four things — line and length, contact quality on a zero-to-five scale, shot direction, and whether the field placement needs to move. That night an opener made 72 off 44. The stands stood up; the next morning's reports used one word, immaculate. My notebook said something else. Eleven of his twenty-four boundaries came off the edge or the splice. Four more went square of a short boundary that Mirpur would have turned into a catch. The scoreboard was selling one story; the paper was writing another. That gap is what Asian domestic cricket never measures, because it never installs the instrument that measures it.

I built the first xG ledger in Sylhet — for football, off 132 matches and 14,800 shots. I have since tried to translate that coordinate language into cricket. The translation fails often, because football remembers who scored while cricket remembers only where the ball finished, not where it pitched.

Asian franchise cricket does not lack money or matches. The IPL began in 2026, the Bangladesh Premier League in 2026, the Pakistan Super League in 2026, the Lanka Premier League in 2026, the UAE's ILT20 in 2026. Broadcast quality often matches England or Australia; every six is captured by eight cameras. The problem is not the broadcast. The problem is the log.

Selection effectively runs on three layers. Layer one is the scorecard: runs, wickets, strike rate, economy. Layer two is event data: ball-by-ball, pitch maps, shot direction. Layer three is spatial tracking: release point, release height, ball speed, seam, swing, shot coordinates, fielder movement. In the County Championship, the Sheffield Shield or the IPL, layer three is now routine. Large parts of Asia's domestic circuit sit at layer one, with layer two partial.

The number comes from my own logging, not an official audit, so carry the error bar: of the major Asian domestic T20 fixtures I watched live or on recording between 2026 and 2026, fewer than five percent had layer-three data publicly available. Everywhere else, the scorecard is the only truth.

The national teams pay for that. What reaches a selector is layer one plus the human eye. Without layer three, nobody can answer why a strike rate looks the way it does; they can only see how much.

Working a live xG desk at the 2026 World Cup taught me that a final hands you two truths: the scoreboard and the process. France beat Croatia 4-2, while my model read 2.1 to 1.8. The result was that night's truth; the process was the tournament's. Cricket prints the first truth daily and never prints the second.

A strike rate is an incomplete number because it cannot separate venue dimensions. Mirpur's square boundary is roughly 66 metres, parts of Sylhet 67 to 68, a smaller ground 54. Same shot, same contact, same batter — different runs. That difference enters the strike rate under the label of skill. I add a stadium-effect variable to my ledger, holding boundary distance and outfield pace apart for each ground. Without it, comparing one BPL batter's 148 strike rate to another's 136 is meaningless.

Wickets are the most volatile statistic in T20, and they carry the most trust. Bowled or lbw means the ball beat the batter. Caught at deep midwicket means the batter middled it into the only fielder nearby. Both produce one wicket on the card, yet the first repeats far more reliably than the second. When I started logging domestic ball-by-ball in 2026, roughly a third of the spinners' wickets were deep catches — mistakes in shot selection, not evidence of bowling skill.

Spin economy in Asia is written in release height, which almost nobody logs. Subcontinental pitches are slow and skid, so the best spinners vary trajectory by changing where the ball leaves the hand. The gap between Mehidy Hasan Miraz's home record and his away record is best explained this way, not by the scorecard. Wanindu Hasaranga's 16 wickets at the 2026 T20 World Cup made him the tournament's leading wicket-taker on the ICC record, but the bigger fact is that on the same slow surfaces he sometimes bounced the ball and sometimes skidded it. Without release-point data that is folklore; with it, it is a training protocol.

The selection ledger breaks at exactly this point. Suppose a batter runs at a 150 strike rate domestically. Without layer two or three, the selector sees 150. If the ledger showed 28 percent of those runs came off the edge, and his boundary percentage drops to 11 on slow, low surfaces, the arithmetic changes. In the other direction sits a 128-strike-rate batter with cleaner contact and a rising death-overs strike rate against pace. Picking the first costs roughly one international slot per series — near thirty matches across eighteen months. Five such calls in a cycle leaves 150 matches of evidence dark.

Tracking does not replace the strike rate, it explains it. The selector who uses that explanation makes fewer errors over time, because he learns to draw a line between luck and skill.

No training protocol means the hardware is wasted. A ball-tracking installation costs a large share of a mid-sized domestic league's budget. Where hiring scorers is already hard, demanding six cameras is not serious. The rule at my Sylhet desk: treat what we can measure as the standard, not what should be measured. Two trained loggers, one replay screen, one standard table — that is step one.

Training that desk meant building two junior loggers, taught four columns only: line and length, contact, shot coordinates, field placement. Within six months their inter-logger reliability passed seventy-five percent. That does not replace spatial tracking. It does close much of the gap between layers one and two with pen and paper.

A question nobody asks: does a bowler's economy measure his work or the ground's dimensions? An orthodox spinner going at 6.8 in Mirpur and 8.4 in Sylhet is not described as worse in the second match. A death bowler who nails yorkers still looks expensive when the boundary is short. In both cases there is no ledger, only a ratio.

The pandemic season of empty stadiums made this visible. Empty stadiums taught me that silence has its own expected goals. Home advantage nearly vanished, and bowlers' economies dropped by roughly two runs an over, because part of the fear was manufactured by crowd noise. Without a ledger, that silence never enters a column — and if it does, it becomes a variable with no measured baseline.

The franchise market and the process ledger read the same numbers and ask different questions. An auction price rises on recent runs and strike rate, not on a venue-adjusted contact profile. Franchises with layer-two desks make fewer expensive mistakes on the same night. The transfer market is not a bazaar; it is a probability engine with agents.

Esports taught me that reaction time is a currency and drafts are ledgers. Players there are assessed round by round, map by map, while in cricket a three-match tournament sometimes decides a seven-year career.

One trained logger at an Under-19 camp produces more structural value than a former star's one-week visit. The budget line exists for the second and not the first. That asymmetry is not one coach's fault or one board's fault; it is the output of an incomplete accounting that books coach education as an expense and celebrity presence as investment.

Now the case against my own argument. More data is not the fix for Asian cricket, because the deficit is vocabulary and coaching, not hardware. A tracking system bought and shelved is furniture. A selector who cannot define release height will reach the same conclusion with ten crore data points.

Second caution: correlation is not causation. Successful international players usually have high strike rates, but raising a strike rate does not produce success. The causal path runs through skill against better bowling. Domestic strike rates are partly manufactured by weak attacks and short boundaries; that does not translate.

Third: ledger worship is its own hazard. In 2026 my model undervalued wrist spinners because my coordinate log had no release height. The model was not wrong; it was incomplete — and treating an incomplete model as complete was my real error.

Fourth: market signals and process models do not belong in the same box. An auction price is a probability, not proof of skill; a model output is an explanation, not a prediction. Conflating them lets anyone blame the process for a defeat, which is as wrong as crediting only the result for a win. The World Cup final gave us two truths: the scoreboard and the process.

Over the next six months, watch which Asian domestic league publishes open spatial data first, which board puts logger training inside its coach-education budget, and which franchise walks into an auction with venue-adjusted contact profiles. Until those three answers arrive, Asian cricket will keep reading scorecards and guessing at process. A spreadsheet is a monastery, and I take vows in columns and rows. I do not chase results; I audit the process until it confesses.