The Spreadsheet With Zero Rows: How Cricket Writing Holds Up When the Data Never Arrives

On Monday morning at the Bangalore data desk I opened the same file I open on day one of every series. The columns were ready — format, match, innings, over, batter, bowler, score, source, timestamp. There was not a single row underneath. The headers were gleaming; the data was gone. Some upstream script had filled every empty field with "N/A" and passed it along as analysis — format unknown, match unknown, player unknown, date unknown, source unknown. Ten check items, ten failures.

At such a moment only one habit serves a desk, and it is not a writing habit. It is the habit of stopping. In front of an empty cell two roads open up. One is to write a plausible-sounding sentence, because in cricket almost any number can be attached to some series or other and satisfy a reader. The other is to admit that the data did not arrive this week and therefore the analysis will not happen this week. The first road is fast, the second is honest. In eleven years the most valuable lesson I have learned is that choosing between these two is the actual skill of the trade.

A cricket number passes through four hands before it reaches a desk. The first is the scorer at the ground, filling the ball-by-ball log — which shot, which direction, which part of the body, catch or cut. The second is the broadcaster's feed, which turns that log into graphics and often drops exactly the tags analysis needs. The third is the wire agency, which sends the score correctly but not the context. The fourth is the desk, where I sit. At every hand something falls away. A wicket gets logged as "caught" when it was a run-out at the non-striker's end. A bowler's economy gets calculated with the no-ball over excluded. These errors look small until someone builds three years of a career on them.

In the summer of 2026 I took that chain into my own hands. In a Koramangala flat in Bangalore, under a failing fan, I hand-coded 1,214 shots from all 64 matches of the Russia World Cup — every shot tagged for body part, assist type and defensive pressure. The open event data had the shot map but not the tag that says what pressure the shot was played under. I opened the notebook at 1,214 shots and found a quieter story: France's open-play xG was 1.02 per match, mid-table, and its PPDA of 16.9 was the highest of the four semifinalists, meaning the least pressing. France won the final 4-2. The algorithm gave them a middling grade; someone else lifted the trophy. That day I understood the model was not wrong, it was incomplete — it was not measuring the part of the game where France were exceptional.

Cricket shows this incompleteness daily. 19 November 2026, Ahmedabad. Virat Kohli's World Cup: 765 runs in 11 innings at an average of 95.62 and a strike rate of 90.31 — the most runs in a single World Cup, a genuine record. That same evening India lost the final. Both numbers are true and they measure different things. The 765 measures the accumulation of eleven innings; the trophy measures one evening, on a slow pitch, against one bowling plan. A desk that treats either as the whole story hands the reader half a truth and calls it satisfaction.

Look at the other number from that final. Australia chased 241 in 43 overs for the loss of four wickets — roughly 5.6 runs an over. In today's T20 age, 5.6 an over reads as slow. The truth is the pitch was slow, the target was small and there was no need to take risk. The run rate there was not an independent quality; it was a consequence of conditions. Every headline cricket number carries this hidden condition — it is an outcome of match state, not direct proof of a player's ability. A writer who strips the condition and prints the number is selling an inference, not a technical observation.

My second notebook, from 2026, was a test of that condition. On 16 May 2026 the Bundesliga returned behind closed doors. Over six weeks I compared the 83 closed-door fixtures of that season with the 223 crowd-backed ones. Home wins fell from 43 per cent to 33 per cent; home goals from 1.74 to 1.43 per game. I then ran the same test on IPL 2026 — 60 matches in the UAE, empty stands. Chasing sides won 34 of them. The crowd coefficient was never a number; it was a question. Because in that same season the schedule had shifted, the venues had shifted, players were tired from quarantine, the dew timings had shifted. How many of those 34 turned because of empty stands and how many because of dew, my spreadsheet could not say. The numbers survived in the file; the causal story did not.

At the auction table the same lesson returns in another form. Before the transfer fee there is a spreadsheet nobody wants to read — the retention count, the right-to-match claim, the purse left over, how many teams needed a player in that role, and how quickly sentiment entered the bidding. The figure that makes the headlines is that day's demand, not a certificate of quality. The same player can sell for two very different prices in two seasons while playing exactly the same way. When a desk prints the bid as if it were the cricket, the reader is left believing the market has valued the player. The market valued the shortage.

Now back to that empty file. The first thing missing was not a statistic but the format tag. This is cricket analysis's first rule and its most broken one. Without format, every comparison is contaminated. A Test opener's strike rate of 135 is fine, because the ball is new and it moves. A T20 finisher's 135 is low, because his job is to go past it. One number, two different verdicts — and format decides, not the number. The second missing element was source tier: is this an official board figure, a wire report, mainstream cricket media, or an aggregator's arithmetic. The third was the timestamp. Without these three, what gets built is not analysis but a translation of guesswork.

Cricket coverage has multiplied its numbers several times over in the past decade. It has not multiplied its accuracy. That sounds backwards, but the experience points that way. When ten metrics are available and three give different answers, the writer picks the one that supports the story already written. That is today's most common failure — not a false number, but a partisan selection of true ones. The cure is not a better model but the nerve to publish a blank page. The broadcast booth hears noise; the notebook hears patterns — and sometimes the notebook has to record that no pattern was found this week.

One more thing, which an analyst who came in from outside should always remember. I was born in South Africa and I write on Indian cricket from Bangalore. I have the log, but the feel of the ground belongs to the scorer who has sat there for forty years. Data supplements lived knowledge; it does not replace it. When my spreadsheet and the ground disagree, the first question should be what my tags left out — not whether the ground is wrong.

Three things must be mandatory for the next ingest cycle: a format tag, a source tier and a timestamp. A file that arrives without them should be marked VOID and sent back, however close the deadline. Under every table there should be one line saying where the number came from and what date it belongs to — that transparency is the trust between desk and reader. In Bangalore the data desk is another kind of stadium, and a scoreboard can lie there too. A single match can lie; a season rarely does.

The question that remains: when the scoreboard and the spreadsheet disagree at deadline, which does the desk believe — and does it have the nerve to put that disagreement in front of the reader?