Testimony of an Empty Notebook: The Unwritten Rules of Data Verification in Cricket Analysis

Testimony of an Empty Notebook: The Unwritten Rules of Data Verification in Cricket Analysis

Last night I opened my laptop by the window of my home in Barishal. I opened a coding file for a Bangladesh Premier League match — a sheet that was supposed to hold fourteen overs of entries, two hundred ball-level tags, and a passing grid divided into five vertical lanes. The file came back empty. No error message, no crash, just blank.

I did not close the laptop. I made coffee and sat down again. Because an empty file had left me two questions — did that match genuinely contain nothing worth coding, or had my pipeline failed somewhere? Those two questions do not have the same answer. And that distinction is the most neglected chapter in cricket analysis today.

For years I have followed one rule — no data means no decision, only a guess. You can build analysis out of guesses, but a guess is never reproducible. And without reproducibility it is not analysis, it is a story.

Since 2026 I have done this work from Barishal. I rewind BPL matches, code pass by pass, splitting them across five lanes. The start was monotonous — fourteen matches of Sheikh Russel KC and Abahani Limited Dhaka, alone on a laptop, 1,842 passes. I wrote down every entry zone, the height of the defensive line, the gap between midfield and defence by hand. That first thread earned 4,300 shares. But the shares were not the lesson; the lesson was the habit of replacing vague phrases like "good movement" with "reception in the right half-space."

That is where my analytical framework was built. Every article began with a five-lane grid, zones 1 to 5, and ended with one reproducible drill for the Barishal U-18 side. Editors noticed that readers had started copying my diagrams. I understood then that if a reader cannot rerun the analysis, it is not knowledge, it is just a performance.

But this whole framework rests on a single foundation — data. And there is one thing about data I rarely hear said, though I see it every day: data sometimes comes back empty, and the urge to fill an empty dataset is an analyst's greatest enemy.

Modern cricket analysis now works like a pipeline. The first stage extracts information from a source — title, information points, relevant entities, core viewpoint, time sensitivity. The second stage builds deep analysis on that extracted information. Format, player technique and data, team standing, league commercial structure, governance, risk, public narrative, and industry transmission — analysis proceeds along these eight dimensions.

The problem is that if the first stage holds nothing, the second has nothing to stand on. Then the analyst faces two paths. One path — fill the gap with guesses, which looks like analysis but is really fiction. The other — admit honestly: there is no information, so there is no analysis. The second path is unpopular in cricket, because filling gaps is easy and readers want a story.

But cricket is a strange game. A decision built on false information is not merely wrong — it spreads. A conclusion drawn from one bad data point becomes another analyst's foundation, settles on social media as a "confirmed truth," and three months later nobody dares challenge it. That is why I verify data before I trust the eye test. My rule is simple: if I cannot rerun it, I do not write it.

So what does an empty file actually say? Here is the real work. An empty file is not itself a conclusion — it is a signal. The question is: a signal of what? Two possibilities.

First possibility — the source itself is thin. Suppose the original text was only a fixture announcement, a photo or video caption, or a brief news item carrying no claim at all. In that case the absence of information is natural. This is not a data gap, it is the nature of the content.

Second possibility — the source was fine, but information was lost somewhere in the pipeline. Perhaps the extraction stage failed, perhaps the file format broke, perhaps a filter dropped a crucial entry. In that case the empty file is lying — "there is nothing" — when in fact there was plenty.

Distinguishing between these two possibilities is the analyst's real duty. And this is exactly where most analysis stops in the wrong place.

Let me explain. Suppose a match coding sheet has no entry for the first six overs of the powerplay. Two explanations exist. Either nothing notable happened in those six overs — consecutive dot balls, no strokeplay, no pressure. Or the camera or data feed had a problem, and those six overs were the most important of all. In the first case the conclusion is "the powerplay was flat." In the second the conclusion is "the data is incomplete, no conclusion is possible." The two conclusions are entirely different, yet the raw file looks identical.

This is where I use a simple test I also teach to the Barishal U-18 coaches — the null-case check. The method runs like this.

First, ask whether the gap is filled elsewhere. If the rest of the match carries data at normal density and only one section is empty, that is probably an extraction fault, not a feature of play. Because a game rarely organises itself that cleanly.

Second, return to the original source. An empty analysis does not mean the source is empty too. Often the source has a title, a date, a context — it simply has no claims or information points. Then the type of source itself decides: is this an analysable article, or just an announcement?

Third, set a deadline. Suppose I conclude this is a pipeline fault. Then within exactly what time will I verify it? This matters, because the longer a guess goes unverified, the more it is accepted as truth. In my own practice I have made this rule strict: I write down guesses with a timestamp, then update them. I never blend observation and guess on the same page.

These three steps are no complex technology. They are a habit — the habit of respecting doubt.

Now let me address a separate trap, one deeply tied to verification. Suppose you have a bowler's excellent death-over economy. The number is true. But over how many overs? At which ground? Against which opponent? If the sample is only six overs, and those on spin-friendly pitches, then "excellent" is a feeling, not proof. Here I follow one rule: data from one format cannot be stretched into another. T20 economy cannot be measured in Tests; a home average does not hold away. If you cannot draw that line, even a huge dataset only multiplies confusion.

Venue and environment work the same way. Rain, dew, grass on the wicket — without those three, a powerplay decision is only half-made. When I analyse a match I first check whether the data density is sound and whether context is written down. If both hold, I analyse half-spaces, third-man runs, pressing triggers. If not, I stop. Stopping is not weakness; stopping is professionalism.

Here I add another point many analysts skip — the shelf life of analysis. Even if a fact is true, it has an expiry. You cannot decide today on form from six months ago. So I attach a date to every fact. A number without a date is half-complete.

Now a specific cricket example at the centre of my work: the left half-space. I have written many times — the left half-space is not a trend; it is a door. But that sentence only means something when I know how many receptions happened there, which bowler is opening that door, and which batter is walking through it.

Suppose a left-arm spinner bowls on a spin-friendly wicket with a left-handed batter at the crease. The angle is created naturally — how far outside off the ball lands at release, how much cover the batter offers, which angle the fielder stands at. Every one of those components can be coded. But if the coding sheet has no entry for those overs, I can write the words "left half-space," yet I cannot prove them.

Another caution here. In my experience analysts share a few specific traps. One, coordinate worship — a pitch map drawn so finely that the map feels like the match. Two, treating one league's data as universal truth. Three, dependence on a favourite phrase — the word "half-space" becomes so dear that even when the door is shut, the analyst still tries to walk through it. Four, silence-led late takes — a hypothesis written late, therefore verified late. There is only one escape from these four traps: check pitch, weather, team news and sample size, and state the uncertainty plainly.

I know this whole discussion sounds dry. But I am certain it is now the real frontier of cricket analysis. We write endlessly about tactics inside the game — powerplay plans, death-over matchups, spin angles. But we almost never write about the tactics inside the data. Yet one bad data point can flip the explanation of an entire match.

I remember one episode. In 2026, when the whole world stopped and stadiums stood empty, I was watching archive matches and mapping coaching shouts and pressing triggers. One thing became clear: empty stadiums, full notebooks. When sound drops, you can hear — centre-backs calling the line, midfielders triggering the press. That period taught me that an empty space is not always a void; sometimes it is a place where we simply did not listen carefully.

In the same way, an empty data sheet is not always a void. Sometimes it is a hint — either the source is thin, or the pipeline is broken. Telling the difference is the work.

Now I come to the part where I go against the common view.

The conventional belief is that an analyst's job is to gather information and reach conclusions. I say the real job is the reverse — deciding which information is useless, and which cannot be used. Gathering is easy; discarding is hard.

More precisely. In cricket we see enormous confidence built on thin information. Six sixes in one match — "this batter is a finisher." Three wickets in one series — "this spinner is a match-winner." These conclusions are not wrong, they are incomplete. And when an incomplete conclusion is stated with confidence, it is more damaging than a lie, because readers believe it.

But there is a danger here I rarely hear discussed. It is this — saying "there is no information" cleanly can itself become a trap. Because when a result looks clean, it feels inherently credible. "No information, so no analysis" — the sentence is honest, but it is not always correct. If the real problem is data loss, that clean result offers false reassurance.

Think about it — if a pipeline genuinely has a data-loss problem, then every "clean" result it produces is suspect. Today one match's data is lost, tomorrow another. Each time the analyst thinks, "there was nothing special in that match." Yet each time something is actually being lost. That false reassurance is the most dangerous, because it is not an obvious error — it is a silent error. And in cricket analysis, silent errors cost the most.

So I say: evidence over reputation. I avoid the idea that a big name guarantees a big performance. I start with numbers. If the numbers are incomplete, I say so plainly. Analysis that cannot stand without a source will collapse one day. Today's social-media speed throws new claims at us daily, and under that pressure many move ahead without verifying. But cricket is a game of patience. So should its analysis be.

So what will you do in the next match? I offer one simple habit. Keep three columns in your notebook — observation, hypothesis, verification. In the first, write only what you saw. In the second, what you are guessing, with a date. In the third, when and how you will verify it.

Three weeks later you will see that most guesses were either confirmed or discarded. The discarded ones are your real learning. Do not panic when data comes back empty. Ask — is the source thin, or is the pipeline broken? The answer may be hiding a large tactical truth, or a large gap. In either case your job is the same — look for yourself, do not guess.