The Empty File, the Fast Verdict: The Report That Carried Not a Single Data Point

The file opened that morning and forty-seven fields appeared on the screen. Next to each one, the same sentence — "insufficient information, cannot assess." The junior analyst had been hunched over that screen for two hours. He said, "Sir, the boxes have to be filled. The coaching staff meets tomorrow morning." I looked at his name on the file, added mine below it, and submitted it exactly as it was. At the bottom I wrote one line: what is not in this file is its most valuable part. Three days later the file came back. This time it was populated, and what it contained broke one foundational assumption of that season's death-overs plan.

This is not the story of a match. It is the story of a moment when a cricket data desk stood in front of its own system — and the system refused to say anything. Tournament cycles produce few such moments, and when they do, nobody covers them, because there is no six in them.

What was inside the file

The forty-seven fields were the template on which we build every opposition brief. Format (Test, ODI, T20, The Hundred), the nature of the match (bilateral, ICC event, league fixture, warm-up), venue and pitch type, temperature and dew, toss influence, the DLS component — that is the first layer. The second carries scoring rates inside the powerplay, middle-overs run rates against left-arm spin, the bowling plan from the seventeenth to the twentieth over, the distribution of field placements across twelve zones, and matchup rates. The third layer is commercial — auction purse, retention, NOC, broadcast rights, WTC points, and any signal of ACU monitoring.

That day all three layers were empty. The table was structurally correct — every column, every unit, every benchmark — yet every cell returned the same string. No source named, no date, no team, no player, no format. This was not a low-information report; it was an absent report. And the most dangerous thing about an absent report is that it looks entirely normal.

I have an older precedent for this. In 2026, after coding 1,146 deliveries from Edgbaston and Lord's on India's England tour, I wrote a 2,800-word essay on the off-stump corridor. The coding threw up an inconsistency and I delayed publication by one day. What that day caught was this: 71 per cent of top-order dismissals to seam were coming from balls pitched four to six metres from the stumps, and 38 per cent of false shots were coming from inside the line. The corridor was never about the route; it was about the choice. Since that day I have held one rule — delay is acceptable, inference is not.

Three layers, one mistake

That empty file contained three separate failures, and treating them as one is the real error.

The first failure belongs to input. The source material either never reached the system or was never read. It is like a scorecard going missing — the match happened, the balls were bowled, but the record was never entered anywhere. The second failure belongs to extraction. The material arrived, an extractor ran, and instead of raising an error it returned an empty envelope. In cricket terms, the scorecard arrived without an over-by-over breakdown, and nobody took the trouble to say the breakdown was missing. The third failure is the costliest, and it sits downstream: the framework demands at least three conclusions and two hidden-information items in every dimension. Eight dimensions, five slots each — forty slots. A junior analyst with a deadline will fill forty slots. This is precisely where an empty file quietly turns into a plausible lie.

The danger was never that the report was empty; the danger is that it could have arrived with all forty boxes filled, and nobody would have caught it.

If the format is unknown, every number is an orphan

One technical point needs stating plainly, because cricket journalism skips it repeatedly. Format is the unit without which every other metric is meaningless. A T20 economy rate is not comparable to an ODI economy rate; a first-innings run rate in a Test is not comparable to a fourth-innings run rate. If the file does not record the format, every number in it has lost its address.

I have a concrete example. In 2026, after the pandemic break pushed the IPL to the UAE, with empty stadiums, I worked from Kolkata on a sixty-match death-overs audit. I coded 1,440 balls from overs seventeen to twenty. The data showed slower-ball use at the death rising from 22 per cent in 2026 to 31 per cent in 2026, and audible wicketkeeper calls changing how captains relayed field changes. That number is inert on its own. It needs three things to mean anything — venue, the nature of the format, and the bowler's type. A rising slower-ball share on a slow UAE surface and a rising slower-ball share on a bouncy Eden Gardens surface do not produce the same conclusion. I drew the twelve zones before anyone thought to ask why — but those zones were drawn for Eden Gardens in 2026. The Impact Player rule has rewritten death-over economics; the old grid will not do that job today. Every rule change forces a re-baseline, and the re-baseline date must be published alongside the analysis, so the reader knows which map they are looking at.

Now consider the empty file whose format field is blank. Any number entered into it can be attached to any of the three examples above — and attached the wrong way. This is where a data desk deceives itself: it hunts for the answer to the question without ever checking what the question was.

Absence of risk signals is not proof of no risk

The sixth section of the file was a risk matrix — injury, workload, calendar density, financial exposure, rules and integrity, public opinion, systemic risk. All six categories read: cannot assess.

The distinction here is one cricket rarely grasps, and it is the whole point of an audit. Not finding risk signals is not evidence that risk is absent. When a fast bowler's workload data is unavailable, you cannot say his workload is fine. The two statements sit worlds apart, and journalism collapses them into a single sentence. My rule is that where data is missing, the words "no risk" will never appear. What will appear is "could not be assessed." The difference is not semantic; it decides which sentence a coach acts on in the next meeting.

The same logic governs the seventh section. A captain's captaincy, a bowler's yorker rate, a batter's strike rate — reaching a verdict on any of them requires three things: over or ball context, phase context, and the batter's or bowler's type. Lose one and you are not producing a conclusion, you are producing an estimate. And in a tournament cycle, estimates are in fastest demand, because the flag is flying, the story is running, and everyone needs a name to hang either the credit or the blame on.

The counterintuitive side

Here is what makes this episode interesting, and what will sound at first like a man defending his own error.

The most useful report of that season was the one that carried no conclusion at all. That was true not in spite of its emptiness but because of it. When the file came back empty, it was re-sent three days later — this time with source, date, and whatever over-level data existed. What the second file showed would never have surfaced on the first pass. The problem was not that the data was unavailable; the problem was that on the first pass we never asked which data was being demanded of us. The empty file asked the question itself.

This is where I begin to distrust my own trade's deepest habit. Data desks are built to answer. Nobody taught them to refuse. Refusal is a skill with no market price, because a report's value is set by its length and its confidence, not its honesty. A hundred-page document with every box full carries more weight in a meeting, even when half of it is suspended in air. And this is where the old line returns — nobody audits the death overs until the death overs audit you.

My own audit threshold is simple: take down a claim in public only when it changes a selection, a plan, or a broadcast consensus. Everything else goes in the private notebook. That threshold saved me that morning, because the file contained no claim worth taking down.

What to watch next time

After that meeting we added a small rule at the desk, known internally as the empty-input protocol. The short version: an empty input is a hard stop, not a warning. Empty and unreadable get separate exit codes, so that "unclassified" can never again serve as a mask for "empty." The re-run window is fixed at seventy-two hours, and every report carries the date on which its model was last re-baselined.

The next test will happen on the field, not in a file. After the Impact Player rule, death-over slower-ball share will be measured again — this time separated by format, venue, and bowler type. If the number lands near 31 per cent again, we will know we are looking at the effect of a rule. If it climbs, the first question will be: did the rule change, or did the model go stale?

Momentum is just a shape nobody has bothered to measure yet. And every collapse has a floor plan. The file that saved us most that season had no floor plan at all — only an empty space that somebody refused to fill.