The Empty Column, the Clean Report: Cricket Analytics Keeps Reading 'No Data' as 'No Risk'

1:12 a.m., Kolkata. A report opened on a newsroom screen: eight sections, twenty-two table rows, and in every row a tidy cell carrying one phrase — insufficient information. No title. No source. No format. The skeleton had survived; the body had not.

The desk read it in three seconds and moved on: no red flags.

I count deliveries the way other people count regrets — quietly, past midnight. The document that landed with me last week had nothing to count. The information-points list was empty. No team, no player, no date, article type unclassified. A report containing nothing was converted, in three seconds, into an assurance that nothing was wrong. This is not a failure of one newsroom. It sits inside the chain through which cricket information now travels.

Context: a four-link chain, and its weakest joint

That chain has four links. Fetch — scorecards, ball-by-ball feeds, broadcast graphics, franchise tagging sheets. Extraction — pulling the actual prose out of a page. Classification — deciding whether this is Test or T20, a match report or an auction rumour, a bilateral series or a league. And analysis. The visible investment in cricket goes into the fourth link: dashboards, models, specialists, the twenty-minute studio segment. The weakest joint is usually the second, and its failures never make a sound.

In a tournament cycle that weakness gets expensive. When thirty or forty matches are cleared in forty-five days, a report is generated after every one of them and the human minutes available per report keep shrinking. The day the schedule is heaviest is the day the checking is lightest. An empty payload slips through most easily on exactly the day nobody has time to stop it.

One technical point gets skipped in cricket coverage far too often: without a settled format, no number can be benchmarked against anything. A T20 innings tops out at 120 legal deliveries, a full match at 240. A Test spreads roughly 2,200 to 2,500 deliveries across four innings. Strike rate itself means two unrelated things — runs per hundred balls for a batter, balls per wicket for a bowler. An economy under seven reads as strong in T20 and as noise in a Test. Where the format is undetermined, average, strike rate and economy all hang in the air without ground beneath them.

Core: an empty cell is not a zero

An empty payload is not a finding; it is an unmeasured variable, and the report's architecture makes it look like a finding. That is the root of the whole problem. A table with twenty-two filled rows looks like a working document even when every row says that nothing could be established.

The illustration is simple. Suppose a bowler's economy column is blank and your dashboard averages blanks as zeros. You have just discovered the most economical bowler in the league — nought runs per over. He does not exist. This is not a hypothetical fear; it is the standard error of spreadsheet-based analysis, because a missing field and a zero field can occupy the same cell.

The spreadsheet had 4,180 rows. Not one of them explained the noise. Across the 2026–18 Ranji Trophy I sat in the old Eden Gardens block and hand-tagged every delivery — seam position, release angle, pitch landing on a nine-zone grid. The most useful column was never wickets; it was false shots induced, the count of deliveries that forced a batter into the wrong shot. That column taught me what a blank actually means. Some rows were blank because the camera never caught the release. Treating those as zero is the same as claiming the batter was never once discomforted on that ball.

One result fell out of that tagging which bears directly on this argument. The reserve seamer Mukesh Kumar's wobble-seam release sat measurably wider than the squad's other right-arm pacers. I wrote it into a two-thousand-word newsletter that nobody at the association read. The finding was correct; there was no reader. Information is valued by its consumption, not by its accuracy.

Return to that document. It contained a technical inconsistency that narrates the entire story: title, source and article type failed together, while the schema skeleton survived intact. A summarising model handed real text would normally emit at least a title. Three fields collapsing at once points to retrieval, not summarisation — the page never opened, or it stalled behind a paywall, or what came back was a navigation shell rather than an article.

A second, smaller-looking item: the domain label arrived as cricket_world when the schema requires Cricket. That is not a spelling matter. The label decides which benchmark table the system pulls. A wrong label means every number downstream is compared against the wrong baseline, and a comparison against the wrong baseline cannot produce a right answer.

DRS has a rule called umpire's call. When ball-tracking lands inside the tolerance band, the on-field decision stands — neither overturned nor confirmed. An empty dataset holds the same status. It is not out, and it is not not out; it is undetermined. The mark of a good report is that it prints undetermined as undetermined instead of converting it into a clean statement.

The same inconsistency repeats at the selection table. When a player's red-ball data is thin, selectors fall back on reputation. India's second XI is the hardest team in the sport to make, and its central problem is not talent — it is an incentive structure that reads missing data as nothing proven. Ranji's competitive meaning keeps thinning, Impact Player distortions have been bolted on, and through all of it the column that stays emptiest is durability with the red ball. At the January 2026 mega auction KKR spent ₹7.4 crore on Dinesh Karthik and built the squad around spin. On 25 May 2026 at Eden Gardens, Qualifier 2: SRH 174/7, KKR 160/9, a thirteen-run defeat. My brief had flagged Rashid Khan's googly share in the middle overs; the plan still asked batters to attack him. Rashid took 3/19. What my sheet did not contain that evening was the delivery sequence on which the instruction to attack was supposed to apply.

Contrarian: the fault is in the design, not the model

The industry's instinctive response is that the model is crude, the algorithm incomplete, the trust in AI premature. I read it the other way. The fault lies in a design built to produce an output under every circumstance. A pipeline that never returns empty-handed never learns how to return empty-handed.

Look at the incentive structure. An analyst who says I have nothing loses standing. An analyst who files twenty-two rows of insufficient information and hands over a clean table gets praised for depth of work. Both contain identical information: zero. The reward is distributed differently. Every auction paddle goes up before the logic arrives; my job is to arrive earlier — and the same rule governs reporting. An institution obliged to announce fourteen or fifteen names will announce fourteen names whether or not it has an argument for the fifteenth. Cricket has shown us that structure repeatedly.

The broadcast bias runs the same direction. What is visible gets celebrated — the hundred, the fifty, the milestone. What is invisible gets no ceremony at all: a blank column, a delivery that was never tagged. That is why missing data is so powerful. It goes missing quietly.

A Russian summer taught me that weather is a variable nobody puts in the model. In that summer of 2026 I argued that France's pressing triggers explained why a T20 ring field should start two metres closer to the bat. The weakest link in that piece was not the boldness of the comparison. It was that I never stated where weather entered my model. A variable absent from the model will be absent from the output too — and it will never announce itself as a zero. It will simply not be there.

In governance the same omission is the most expensive of all. When a tournament's integrity report states that no suspicious activity was found, the first question should be whether surveillance ran at all. The 2026 Cronje affair, the 2026 Pakistan spot-fixing case, the 2026 IPL spot-fixing case share one lesson: a blank report and a clean report are not the same document. The effect travels downstream as well. Fantasy platforms and broadcast graphics run on the same feed, and where data is absent they read stability — when in fact nothing was read at all.

Takeaway: watch one sentence in the next report

Through this tournament cycle, test one sentence in every match report you read. No injury concerns and injury data was not collected look alike on the page, but the first is a finding and the second is a confession. The fix is cheap: a validation gate between Stage 1 and Stage 2 that refuses a payload with empty information points and populates a null_reason field. The cost is a day's work; the saving is all the trust that would otherwise be spent reading an empty table as an all-clear. Next time a report tells you everything is fine, ask the only question that matters — is everything fine, or was nothing measured?