The Hum of Empty Cells: Null-Handling and the Ethical Kill Switch in Cricket Data

The spreadsheet began to hum, and I knew the broadcast was over. Half past midnight, and the only light on the Hackney flat's table was the laptop screen. The coffee had gone cold; the deadline sat three hours away. In my hands was a data package in which every cell was empty. No headline, no source, no information points, no team or player name. Beside each field, one sentence: insufficient information — assessment not possible.

In thirty-one years of cricket journalism I had never seen such an empty handoff. Since I walked out of a London sports radio station in 2026 after an on-air argument about Burnley's expected goals, I had learned one thing — a single number can be more honest than a story. But when the numbers themselves never arrive, the journalist is left with one decision: fabricate, or stop?

(Context)

Modern cricket coverage now runs on two layers. The first is raw material: scorecards, ball-by-ball logs, pitch reports, DRS records, fielding maps. The second is analysis: extracting meaning from that raw material. In the pipeline I work in, there should be no visible gap between the two. Analysis without information points is invention, and invention is fake news.

For the past few weeks, mid regular season, I have been noticing a pattern. Teams' pressing intensity — a football PPDA-style proxy — has been dropping across a three-match window. To make that claim I need specific information points: which match, which over, which bowling change, which field setting. If the package lacks them, the claim is my imagination, not journalism.

Right now in England's County Championship and South Africa's domestic T20, the same kind of story is forming. One side's powerplay strike rate has fallen from 145 to 121 across three matches, yet the scorecard hides it because wickets have not fallen. That sort of signal is the real news to me — long before it becomes a headline. But to prove it I have to go match by match through ball-by-ball data.

This is where my role doubles. I am storyteller and method-builder at once. The story wants fast extraction; the method wants slow verification. Last night's empty package reminded me that the second is the real work. Some imagine a data journalist is a number factory. We are judges of numbers, and a judge's first duty is not to rule without evidence.

(Core)

The question is simple: when a data journalist receives zero data, what does he actually do? The answer is less simple, because the system encourages you to build.

Think of 2026. I was sitting in a London sports radio studio, drawn into an on-air argument about Burnley's lucky 16th-place finish. I pulled up their 2026-17 expected goals: 42.1 for, 44.8 against — a minus 2.7 differential. The number said Burnley were a mid-table side, not relegation fodder. My producer called it spreadsheet sorcery. I quit that week.

The lesson was plain: one correct number weighs more than a thousand comments. But in 2026 I had data; last night I did not. That difference is decisive.

At Russia 2026 I tracked every side's PPDA. Hosts Russia's group-stage PPDA was 8.7 — the most aggressive pressing by a host nation in tournament history. In a pre-tournament piece I predicted their quarterfinal run, citing pressing intensity over talent. In the knockout, Spain completed 1,005 passes against Russia and still lost on penalties — Igor Akinfeev saved two shots. Six pieces in four days. My editor gave me a raise. I bought a flat in Hackney.

I ran the PPDA numbers again, and the flat in Moscow started to feel real.

The curious thing is that both cases shared one discipline. I had decided in advance which number was my thesis and which number would catch me out. Call it a pre-registered counter-metric. Watching Russia's pressing, I was also watching how high their defensive line sat and how it broke against pace.

My verification steps are almost mechanical. First I extract the number from the raw log. Then I check it against a league average or historical benchmark. Then I ask how large the sample is — three matches cannot carry a season's verdict. Finally, a qualitative check: do the coach's words, the footage, and the player's own admission match the number? If any of these four steps fails, I drop the claim — that is my ethical kill switch.

This discipline has a human side I often forget. Behind every number is a person — with a career, an injury, a family. When I build a model of Burnley's defensive record, Tom Heaton's shoulder pain does not enter it. The model never tires; the player does.

By 2026 the method had matured. When COVID-19 emptied the stadiums, I saw not a tragedy but a natural experiment. I scraped 1,200 matches from Europe's top five leagues, March to December 2026. Home advantage fell from 0.42 to 0.28 goals. Referee bias toward home teams dropped 23 percent. The Ghost Games series ran, and for the first time my data entered a policy debate about fan return.

In the ghost games, the crowd disappeared, but the pressing lines left fingerprints.

Why do these cases matter now? Because an empty stadium and an empty dataset ask the same question: are we actually measuring what we see, or imagining what we want to measure? In a crowdless ground home advantage shrank because the pressure came from the crowd, not the play. Likewise, filling empty cells with analysis pushes the truth further away.

There is a monastery in every dataset, and its silence is not empty. Silence here is not ignorance but restraint. The analyst who knows how to stay silent becomes more credible over time — because readers know that when he speaks, it has been checked.

(Contrarian)

The common assumption is that a full dataset is always better than an empty one. I doubt it.

An empty cell forces you to stay honest. A full cell — with numbers but no source, no context, no format tag — gives you room to lie. Mixing Test, ODI and T20 metrics means the number can be right while the meaning is wrong. A fast run rate is gold in T20, but suicide on day one of a Test.

Here the difference between correlation and causation steps forward. A side's pressing may rise with its wins — but is that because of the pressing, or because a team that is ahead can press more? In small samples the two are almost impossible to separate. An analyst who does not add this caveat is making stories with data, not science. I do not trust the eye test until it can survive a scatter plot.

There is a less flattering truth too. Demand for content in this industry is so strong that it creates social pressure to fill the gaps. Small clubs develop half-finished products for giants through loan-with-obligation deals — just as data journalists sometimes pass half-verified numbers off as full truth. The reader ultimately pays the method's cost, and the player pays for the error.

(Takeaway)

So last night's empty package was in fact a gift. It reminded me that my job is not to make numbers but to recover the truth behind them. In the next round my aim is singular: keep an information point beside every claim, a counter-metric beside every metric, and a human cost beside every model.

The model did not predict the goal; it predicted the regret of ignoring it.