Valuing the Empty Sheet: In Cricket's Data Desks, N/A Means Unknown, Not Safe

The Sheet at 2:47 AM

At 2:47 in the morning, a payload lands on a Delhi data desk. Six fields, one answer across all of them — unavailable. No title, no source, no publication date, an empty list of information points, an empty list of viewpoints, and the article type logged as 'Unclassified'. A single field is populated: the domain label, and it is populated wrongly — cricket_world, where the approved taxonomy requires 'Cricket'.

I have seen hundreds of these sheets in forty-one years. Every time, the same thing happens. Somebody stands up and says, 'Sir, there is nothing in this — which means there is no risk.' That sentence has cost me more than any bad model I have ever run.

A sheet with nothing written on it and a sheet that says everything is fine are two entirely different documents on a data desk. To the man who reads it at nine the next morning in a briefing, they look identical. Both look empty. That is where the error begins, and no ledger ever records it.

Three Blocks, One Final, One Run

In 2026 I joined a Delhi trading desk that priced international cricket. Fourteen traders, me the only woman, and the only one who had never played the game. I never considered that a deficiency, because my job was not to watch the cricket. It was to price the phases.

I rebuilt our T20 model around three fixed blocks — powerplay, middle, death. Each block had its own par score, each had its own economy benchmark. The par score moved with the surface; the blocks never moved. That was the discipline: the narrative could change, the blocks could not.

21 May 2026, Rajiv Gandhi International Stadium, Hyderabad. Mumbai Indians made 129/8. That is well below par on any surface, and at the innings break the market made Mumbai underdogs — an understandable error, because 129 is a low score.

My death-overs economy benchmark did not move. On that surface, that evening, 129 was defensible, and the reason was not the score. It was the gap between the two sides' death-bowling economy. Mumbai won by one run. Rising Pune Supergiant finished on 128/6. Krunal Pandya took the man of the match award for 47 off 38 balls.

I stopped writing about momentum after that night. Every T20 piece I file now opens with a three-block split, an explicit par score for the surface, and both sides' death-overs economy. Contributors get a one-page template and a hard cap on narrative sentences. If a claim cannot be expressed as a phase number, it does not survive the edit.

I priced the final in three phases before I knew how it would end. I repeat that sentence because the method inside it is what catches an empty sheet. When you write the claim first and the result arrives later, you know exactly where you were wrong. When you write the claim after the result, you never find out.

The Anatomy of a Payload

The thing that arrived at 2:47 was not a news item. It was the signature of a failure. On a data desk we call it an empty payload, and we identify it not by its emptiness but by the pattern of its emptiness.

Pattern one: title missing, source missing, type unclassified. Those three do not fall together by coincidence. When a page yields no title, the break is at the fetch or parse layer — the page was JavaScript-rendered, or behind a paywall, or blocked by an anti-bot wall. The title is the easiest thing on a page to extract; it sits at the top. When even that is gone, the problem is the pipeline, not the article.

The second pattern: an empty list of information points. This is the field everything else runs on. A normal cricket article yields five to fifteen points, however poor the article is. A match report yields score, overs, wickets. A preview yields form, injuries, head-to-head. A governance item yields board, meeting, rule. Zero points means the text yielded nothing.

The third pattern: an entity-layer mismatch. The domain label reads cricket_world where the approved list says 'Cricket'. It looks trivial. It is not. When label mapping itself is wrong, everything beneath it is suspect, because a system that cannot get that right has probably handled names, figures and dates with the same carelessness.

The fourth pattern, and the most dangerous: time-sensitivity was never assessed. Cricket is a sport where every claim is anchored to a date. A comment made before a series and the same comment made after it carry different weights. With that field empty, you do not even know which season you are describing.

One distinction must be made plainly, because I learned it expensively at this desk. An empty payload comes in two kinds. One where the article really was hollow — a caption, a scorecard stub, a fragment left after a paywall. Another where the article was solid and our machine could not read it. The only way to tell them apart is to go back to the raw source and count by hand. If the raw text has names, scores and dates and the payload is empty, the machine is at fault. If the raw text has nothing either, the source was never worth the queue.

I keep two ledgers: what the model said, and what happened. The gap is the lesson. This sheet wrote nothing in the first ledger. In the second, it recorded that the desk could say nothing at all about that article within twenty-four hours. That is also a result, and it cannot be hidden.

The Silence of the Taxonomy

Cricket's data problem is that 'cricket' is not a word, it is a roof, and four different sports live under it. Test, ODI, T20, and now formats with their own rulebooks. Their numbers do not transfer. A bowler's T20 economy can be 8.2 and his Test economy 2.8 — both true, both the same man. Put them in one column and you get nonsense arithmetic.

So when a sheet arrives with no format marker, I write no number on it. This is not a matter of principle; it is daily practice. Duckworth-Lewis-Stern came into use in 2026, and revised targets from before that date are not comparable with today's standard on a like-for-like basis. That is a technical detail that vanishes from every series review. When somebody says a team's record in rain-affected matches is poor, my first question is: under which method, from which year.

The same applies to the World Test Championship. Its points system does not weight every series equally, and adding points from one series to points from another to describe 'form' is like adding banknotes in two currencies and calling the total savings. Without a format marker and a competition marker, none of it is possible.

Here is something thirty years taught me that the data world still refuses to accept. The most useful information in cricket's numbers is not in the number. It is in what is written beside the number — which format, which surface, which weather, against which side, under which rulebook. Metadata is not decoration. Metadata is the only thing standing between a number and a claim.

And when the metadata is gone, the desk has one duty left: to say that we do not know. Do not read that as weakness. On a trading desk it is the strongest sentence available, because it stops you putting money down. A desk that has not learned to state its own ignorance will, sooner or later, take a large loss in a single night.

The Format Marker: The Tag Without Which All Arithmetic Fails

In 2026 I extended the framework to football and built a dead-ball expected-goals model for the Russia World Cup. The model said one thing: the market was using possession as a proxy for control, that it was a weak predictor, and that set-piece xG was systematically undervalued.

That tournament produced a record 29 penalties. On 15 July 2026 at Luzhniki Stadium in Moscow, France beat Croatia 4-2. Two of France's four goals came directly from dead-ball situations — Mandžukić's own goal off Griezmann's free-kick, and Griezmann's penalty. I had written down in advance that this might happen.

The first VAR World Cup taught me to price dead balls before anyone argued about them. That same week a client replied to my note in one line: 'Women don't read defensive lines.' I answered with the spreadsheet.

I bring that up because it carries the lesson that attaches to tonight's sheet. I had no option to say 'I don't know'. I had to produce a number in advance, with a sample size and a confidence interval attached.

My desk rule since then: a sample size with every number, a limit with every claim, and the limit above the claim rather than below it. A caveat buried in a footnote is not a caveat that gets read.

Apply that rule to an empty sheet. A sheet with no sample, no format, no unit — any sentence written on it breaks the rule. That is why I took no figure off that payload. What I did was name its failure and put the name at the top, not the bottom.

Entity Recall: Losing a Name Costs More Than Losing a Number

There is a count on every data desk that nobody watches — entity recall. Of the names present in the raw text, how many were extracted.

It matters because cricket journalism stands on names. Teams, players, boards, selection committees, franchises, broadcasters. When the machine cannot extract names at all, every inference about the article becomes a gamble. I have reached the conclusion that losing a name costs more than losing a number. Without a number you can still ask someone what happened in the match, and you will get an answer. Without a name you have no question left to ask.

So the rule here is this: ten articles from every batch are read by hand before and after, and the names counted. If even one name is missing from any article, the whole batch report is treated as suspect. It sounds harsh. It is. But the cost of that harshness became clear the day a whole series review went out with a player's name replaced by 'an all-rounder'. The sentence meant nothing, and no reader could have known something was missing.

That connects to a larger point about cricket's politics. Teams and players the data desk watches get audited three ways — scorecard, video, commentary. Teams it does not watch get audited none. The bias is not a conspiracy. It is the real effect of stadium aura and media pressure: a big name draws cameras, cameras draw data, and data draws the absence of 'N/A'. The result is that smaller teams look like they have no history, and an absent history gets read as 'no record', when the truth is that no record was ever made.

Two Ledgers: ₹48,390 Crore and the Empty Ranji Stand

The biggest, cleanest, easiest dataset in cricket journalism is broadcast rights. At the June 2026 e-auction, the Indian Premier League's media rights for the 2026-27 cycle sold for ₹48,390 crore. That figure is real, verifiable, and costs nobody any effort to type.

I could publish it every week and be right every week. That is precisely the danger.

Because there is a second ledger, and nobody prints it. How many reports were filed from the Ranji Trophy knockout rounds, how many spectators stood in how many grounds, how many matches never got a single television frame, how many players gave not one interview in any language. That count is hard because it is never recorded. And what is never recorded is never valued.

In March 2026, Mumbai beat Vidarbha by 169 runs at the Wankhede to win the Ranji Trophy — the club's 42nd title. That was a big story and it got the space a final deserves. The quarter-final that season, played in a smaller city, probably earned you one paragraph. I am not writing this as a complaint. I am writing it as a reconciliation.

My standing thesis, formed over forty-one years of watching: the IPL did not damage Test cricket in India. It damaged the Ranji Trophy's claim on relevance, and nobody priced that damage.

Now attach that to the empty payload. When you have one ledger reading ₹48,390 crore and another full of empty cells, the empty cells will be read in your report as 'no risk'. This is where a data desk's honesty is decided. Both ledgers belong on the same page, side by side, with nobody protected from the other. The day we print only the first, we have told the reader the second does not exist.

The crowds left, and the baseline I rebuilt without them has held ever since. That baseline was built on empty stands, not full ones. And it is that baseline which tells me that the blank space on any sheet is usually the story of somebody not being watched, not the story of something not happening.

How N/A Gets Read on a Betting Desk

I am a betting analyst, and this should be written without wrapping. My entire framework rests on the difference between risk and uncertainty. Risk is what you can measure and price. Uncertainty is what you cannot. Do not put the two in the same cell.

An empty payload is not a risk signal. It is uncertainty. When a desk writes 'no anomalies found' about a match, it should also write how much data it looked at. 'Zero anomalies, from zero data' is an incoherent sentence, and nobody stakes money on an incoherent sentence.

Let me be clear that there is no odds signal, no money-flow observation, and no betting inference in this piece. What there is, is a professional discipline, and it says: never read the unknown as low risk.

I learned that in Hyderabad, on the night the market looked at 129 and made up its mind while my benchmark stood still. The market had data that evening. The problem was not a shortage of data. The problem was that it read the score and did not read the death-overs economy. One run decided the final, and I had already written down why it might.

An empty sheet is a step further along. There, you do not even have a score to read. And where there is nothing to read, the gravest danger is that an experienced man starts filling the cells from memory. I will say without hesitation that the most common failure on cricket desks today is not a bad model. It is a cell filled from memory.

Pricing the Dead Ball First: The 2026 Notebook

Dead balls are where the market is lazy and the data is patient.

I write that when I need reminding that price is set where nobody is looking. The dot ball is cricket's cleanest example. Six or seven dot balls in a T20 innings can turn a match, and commentary mentions none of them. What is not an event is not valued, and what is not valued gets priced at zero. That is the oldest error in pricing.

The same error attaches to an empty payload, in the opposite direction. The market prices a dot ball at zero when it is worth a great deal. The market reads an empty sheet as 'nothing' when it is in fact 'unknown'. Both share one mistake: something is being judged by its absence rather than by its value.

I keep two ledgers: what the model said, and what happened. The gap is the lesson. On this sheet the model said nothing. Instead of treating that as a lesson, we treated it as a hazard, and that is the difference between one desk and another.

What Was Never Recorded: Kapil's 175* and the Standard of Evidence

18 June 2026, Tunbridge Wells. India were 17 for 5 against Zimbabwe in a World Cup match. Kapil Dev made 175 not out. India won.

No footage of that innings exists. BBC staff were on strike and the cameras were not running. What survives is testimony — teammates' accounts, the scorecard, newspaper reports.

I am one of the few witnesses to that innings, and I say this every time somebody claims that what was not recorded did not happen. That innings taught me that evidence comes in grades, and the grades must be labelled. Where there is tape, rely on the tape. Where there is no tape, state the grade of the testimony. Where there is not even testimony, write that we do not know.

The problem with an empty payload is not that it lacks information. The problem is that it does not say why it lacks information. That is the distance between the innings of 2026 and the empty sheet of 2026. In 2026 we knew why there was no tape. In 2026 we do not even know why the sheet is blank.

The Null Gate: The Rule That Stops the Empty Sheet

One concrete recommendation comes out of all this, and it is technical, not emotional.

Put a hard gate in the pipeline. If the information-points list is empty, or the title is unavailable, stop. Do not proceed to analysis. Route it back for re-extraction. The gate is necessary because our entire system assumes that what arrived is complete. When half arrives and we treat it as complete, the other half gets filled in — and the filler is not a machine. It is us, working from memory.

Second rule, about language. Never print 'not available' as the equivalent of 'no risk'. It looks like a matter of word choice; it changes the meaning of the whole document. In English, N/A and NIL are one letter apart, and investment decisions rest on that letter. My template now reads: 'Not available — meaning: unknown; meaning not: nil.' It is not an elegant sentence. It works.

Third rule, about labels. When a domain label does not match the approved list, do not begin the analysis. Mislabelled data is more dangerous than missing data, because missing data is visible and mislabelled data gets used.

Fourth rule, entity recall, as above. Ten articles read by hand, every batch. It is expensive. I do not care. I care about the night an all-rounder's name went missing and nobody noticed.

Contrarian: The Real Danger Is Not Invention, It Is Reading Blank as Clean

For several years the entire argument about cricket data has focused on one question: when will artificial intelligence lie. The question is fair and I do not dismiss it. But it pulls attention away from the real hazard.

The real hazard is not what the machine makes. It is what the machine does not make and we make ourselves. An empty sheet that somebody read as 'all clear' does more damage than any invented statistic, because an invented statistic provokes argument and gets caught, while an empty sheet provokes nothing. Nobody challenges a document with nothing in it.

One more thing belongs here, and it is the most uncomfortable thing my career has taught me. The commentary box is the last un-audited room in the sport. The claims made during a match — this pitch is slowing, this bowler is weak at the death, this side folds under pressure — carry no sample size, no confidence interval, and no obligation of verification. That habit leaks into the data desk, where it simply hides inside a spreadsheet.

I should also disclose, because hiding it would break my own rules, that many of my sources sit inside boards and franchises. Their presence means my calls get returned, and occasionally a leak reaches me that reaches nobody else. My standing rule: no story sourced from access is published without a second, non-access source. It slows me down. It keeps me honest.

And a last point, directly attached to that sheet. The Indian cricket writer must decide one day whether he covers the board or works for it, and most have quietly chosen. The data desk meets the same question in a new form: will you publish the data you were given, or also the data you were not given? Printing the blank cell — saying plainly that we do not know — is the other half of that choice.

The Signal Ahead

Back to 2:47, where we started, now knowing what came out of it. The sheet with six blank cells tells us nothing about a match, a player or a contract. It tells us one thing, and that thing matters to the whole industry: somewhere in our information supply chain a link is broken, and that break is waiting for someone's attention.

Over the next twelve months, with the next auction, the next rights cycle, the next WTC window closing, every desk will face the same question of what to trust. My answer stays what it has been for forty-one years: trust what you counted yourself, and leave blank what you did not. A blank cell is not a disgrace. A falsely filled cell is.

Every recalibration I have done began by admitting the old baseline was honest then. That applies to this sheet. It honestly said it had nothing. We were in a hurry to fill it instead of listening — and had we filled it, we would never have known what we lost.

The question is not what was in that sheet. The question is who will be the first person to fill it in, and whether he will put his name under the cell.