Umpire's Call: Ball-Tracking Arithmetic on Asian Pitches and the Uneven Calibration of DRS

On 6 November 2026, at Delhi's Arun Jaitley Stadium, Bangladesh and Sri Lanka were deep into the 24th over of a World Cup match when Angelo Mathews took a few seconds too long to reach the crease. The journalist next to me shut his laptop and said, "Nothing will come of this." Four minutes later Mathews was timed out — the first such dismissal in the history of international cricket. Nobody in the press box clapped. Nobody whispered. Everyone simply stared at the scorecard. So did I. The decision was legally watertight, and yet on the measurement side we had almost nothing: how many of those seconds went where, while a helmet strap was being replaced, was never published frame by frame. Cricket's umpiring arguments are not really about right and wrong; they are about what can be measured and what cannot. Rewind the tape; the truth usually hides between two frames. 19 July 2026. At Colombo's Sinhalese Sports Club Ground, the Player Review System took the field for the first time in an India-Sri Lanka Test. India took several reviews in that match; none worked. Over the following decade and a half DRS became a permanent fixture, yet for the whole of that stretch one country opposed it fiercely — India. The BCCI's argument was that if technology also produces errors, the natural flow of the game is damaged. The argument was not unreasonable. The problem is that the doubt was never translated into measurement; it hung at the level of decision-making — 'in or out'. In October 2026 the BCCI accepted DRS across all formats. Then came amendment after amendment — the soft signal on catches, the third umpire's power on no-balls, new definitions of the pitching and impact zones. From 1 June 2026, in the ICC's updated playing conditions, the soft signal was removed from international cricket. On paper this is a victory for transparency. In practice it has placed a single frame-based decision on the third umpire's shoulders, in a world where camera angle, frame rate and the graphical overlap of ball and glove are not supposed to leave any doubt at all. This is where umpire's call enters. Ball tracking is a forecasting instrument, not an observational one. The cameras can capture the ball at one point after pitching and at one point just before it meets the pad. The software then draws a trajectory through those two points to estimate where the ball would have been at the plane of the stumps. Estimate means prediction. And prediction carries an error margin. The rule says that if the ball covers more than 50 per cent of the stump area, the on-field call stands; below that, the technology's call wins. On the surface this looks like giving the umpire the benefit of the doubt. In statistical language it is an admission that the tracking has its own uncertainty, and that inside that uncertainty we will keep the on-field decision intact. The problem collects right here. That uncertainty is not equal at every ground. In England or Australia the ball usually bounces higher and seams less; the remaining distance from pitch to stumps is longer and the path relatively straighter. At many Asian grounds, especially on a fourth-day turning, low-bouncing track, the picture is the reverse. The ball changes direction twice inside two feet — once off the pitch, once through the air. At Mirpur or Kanpur, when a spinner releases the ball it may drift in from outside the line and strike the pad, leaving only a few centimetres to the plane of the stumps. Across that short stretch the ball travels a few more centimetres laterally. Which means the segment where the model must guess carries the weight of the entire decision. In my own ledger, across 27 Tests and ninety-nine white-ball matches played on Asian soil since 2026, I have recorded the outcome of every review — 517 in total, of which 304 concerned ball tracking or impact in line. This is not an official dataset; it is my ledger, and its pattern is what troubles me. On turning, low-bouncing wickets, the share of decisions surviving as 'umpire's call' sits noticeably higher than on bouncier surfaces. The green ball path we see in the tracking graphic carries no printed error band. A viewer cannot know that the path is one estimated path among several possible ones, and that some of those possible paths might have missed the stumps altogether. It is important to avoid the bias-hunting trap. Suppose there were partisanship — then we would expect the same frame, the same ball, the same moment to produce different outcomes for two teams. I have no direct evidence of that in my ledger. What exists is something else: a geometric inequality. The question is not about the umpire's integrity; it is about calibration. The 50 per cent threshold has been installed as a universal number, but an instrument's error margin changes with conditions. Apply one standard across two different worlds of error and the results will differ. That is the first lesson of any ledger. The second layer is logistical, and it gets even less attention. DRS is not neutral infrastructure; it is procured by the host board and the broadcaster. Which series gets Hawk-Eye, where Real-Time Snicko is used instead of UltraEdge, how many cameras are installed — all of this is negotiated before a series, and the host party has an obvious interest in the outcome. During the Covid period of 2026-21 the neutral umpire system broke down and home umpires returned; neutral umpires have since been restored, but no principled uniformity has ever been created around technology providers. So the 'same DRS' we imagine is not the same at all. From the ledger's angle there is another layer — the mechanical application of law. The timed-out rule is clear: two minutes after reaching the crease. But who measures the moment of 'reaching'? Where does the clock start — with the official at the scoreboard, or in the on-field umpire's eye? During the Mathews incident both sides were playing at full intensity, one of them on the attack, and the rule fell in such a way that a gap of a single second settled a World Cup match. Whether Mathews broke the law is not the question. The question is why the measurement component of the law is never published in recorded form. Cricket now measures pace, angle and spin to the centimetre, but measures time on the strength of an informal clock in seconds. The same applies to over rates. From 2026 the ICC began applying in-match over-rate penalties in T20 cricket and later in ODIs and Tests — an extra fielder inside the circle if the bowling is not completed within a set number of overs. The transparent part of the rule is the penalty; the opaque part is the accounting. Was time lost to the innings break, an injury, a review or DRS setup, or to the fielding side's slowness? There is no publicly released method that separates these. The penalty lands on the team total, and the player cannot tell where his own fault lies. What I think is this: in the current cycle the centre of officiating controversy will be disclosure of information, not improvement of technology. The problem right now is not a lack of information but a monopoly on it. Hawk-Eye's ball-tracking data stays with the board; the raw data of the predicted path is never published. Until that changes, the door to conspiracy theories stays open — and that is not a fault in the cricket fan's imagination; it is a fault in the open door. Now to the contrarian argument. Some will say technology reduces controversy, so where is the problem? But fifteen years of evidence suggests technology has not reduced controversy; it has relocated it. In the pencil era people argued about the umpire's eye; in the DRS era they argue about a 3D animation. The third umpire's job is not easier now but far harder, because after the removal of the soft signal there is no psychological anchor left. A conclusive decision must be made from an incomplete frame, with no allowance of time. The phrase 'clear and obvious' is written in the rule book, but it has no numerical definition. Which angle, which frame rate, how much blur — none of it is written down. That is the largest rules-technology gap of all: a written rule without a written standard. The same logic works in another argument. In the debate over short-leg catches the cricket world splits in two, when the one piece of information that matters most — whether a gap exists between hand and ball at a given frame rate — cannot be understood without watching a slow-motion replay twice. Rewind the tape and you can see the difference; the problem is that the time to rewind the tape is not given to the viewer. The broadcast chooses which moment to show, and that selection becomes the body of the story. So what should be done? I have three suggestions, all of them unpopular and probably none of them implementable in this cycle. First, condition-dependent calibration: on turning and low-bouncing tracks the confidence level for the impact zone should differ, and perhaps even the intervention threshold should differ. Second, the uncertainty band of ball tracking should be shown in broadcast — if the ball is not a single green line but a fading band, the viewer at least has less room to misunderstand. Third, the reasoning behind the third umpire's decision should exist in written form in the match report, the way Bangladeshi umpires still write match reports. I was born in Dhaka and cover the Indian cricket market from New Delhi. My vantage point lives exactly there — on the border between two countries, and in a press-box chair. My ledger is not an official document, and I want to say that plainly, because a ledger that hides its limits stops being a ledger. Still, the economics of the game has pointed in one direction for fifteen years: better measuring instruments mean less controversy — that idea is a myth. The opposite seems true. The more honestly an instrument's limits are disclosed, the less controversy there will be. On that November evening in Delhi, as Mathews climbed the pavilion steps, I thought we had all read the first page of the law book. The question that remains is this — are we dragging that same letter-bound faith all the way to the software, an instrument with no obligation to keep account of its own error?