Format First, Numbers Later: The Invisible Foundation of Cricket Analysis

Format First, Numbers Later: The Invisible Foundation of Cricket Analysis

Hook — The Honesty of an Empty Cell

It is half past midnight. In a small Jakarta flat a laptop screen glows, showing a spreadsheet — 1,140 shots, six columns, and exactly one empty cell in the middle. The formula I built points at that cell and says: insufficient information, cannot assess.

I was nineteen then. I studied economics on campus and quietly believed every empty cell meant a personal failure. An empty cell meant my model was weak, my eye was weak, my effort was insufficient. That night I understood for the first time that an empty cell can be the most honest answer of all. Data that does not exist can always be invented — but inventing it means lying to yourself.

Shot maps are memory with coordinates, and part of memory is always blurred. The greatest discipline in cricket analysis is the courage to face that blur and say 'I don't know.' Today, a few years later, sitting in Jakarta as a transfer market administrator, I understand that empty cell was my real teacher.

Again and again I have seen analysts make their biggest mistakes exactly where an empty cell gets filled with a number. And in cricket the most common form of that mistake is a single one: reading numbers without understanding the format.

Context — Three Games, Three Worlds

Cricket's most confusing trait is that the same game speaks in three entirely different languages. A Test runs five days, a draw is possible, every session is its own chapter. An ODI is fifty overs long, where five-day patience is a luxury. A T20 is twenty overs, where one over can write the whole story.

Early in my analytical life I often weighed these three on the same scale. In 2026, aged nineteen, I hand-tagged 1,140 shots from a Liga 1 season and built an xG model in Google Sheets. That model showed the champion side scored 9.7 goals more than their xG — meaning a large part of success lay outside skill, in pure luck.

That experience gave me a rule: the first task of any analysis is to fix the format and the context, then the numbers. What league-versus-cup means in football, Test versus ODI versus T20 means in cricket — and the difference is not only tempo but the entire tactical logic.

In 2026 I calculated PPDA and field tilt across all 64 matches of the Russia World Cup and found France conceded only 0.82 xG per knockout match. That was a low-block structure nobody saw openly. I wrote: I found the low block hiding in the negative space of a shot map. What a shot map omits can matter more than what it shows.

In cricket this negative space matters even more, because cricket data is scarcer, and people make big decisions — selection, bowling changes, transfers — on that scarce data. Decide with the wrong format's benchmark and a wrong decision is guaranteed.

In 2026, when world sport stopped and stadiums emptied, I scraped records on 1,800 cricketers and built a valuation model joining minutes, age, xG and salary. It flagged seven clubs at risk of insolvency; within eighteen months three were relegated or went dormant. The silence of empty stadiums became my loudest dataset.

This piece is a distillation of that experience. I do not claim to know everything. The opposite: I want to show that cricket analysis's greatest discipline is deciding in advance which questions can be answered and which cannot.

Core Analysis — Why Format Comes First

Suppose someone says, 'this batter's strike rate is 140, that's very good.' The first question should be: in which format? In T20, 140 is middling; in ODI, 140 is outstanding; in Test, 140 is nearly impossible. The same number carries three different meanings in three places. The number does not change, but the language of the number does.

This is my deepest irritation. Many analysts take an indicator built in one format and drop it into another, then are surprised the prediction fails. Just as in football 'league form does not transfer internationally,' in cricket 'T20 run-scoring skill does not transfer to Tests.' Not only skill — the whole benchmark fails to transfer.

I call this format-blindness. It has three layers, each causing distinct harm.

Layer one — benchmark confusion. Every format has its own natural ceiling. In T20, eight to nine runs an over is now normal; in ODI, five to six; in Test, three. Judge a Test batter by T20 standards and everyone looks a failure. The reverse is also true. A benchmark is format-specific, or it is not a benchmark but a bias.

Layer two — tactical logic confusion. In Test cricket time is your friend; you can leave the ball, waste overs, tire the opposition. In T20 time is your enemy; every dot ball is a small loss. These two logics are opposite, so the same decision is clever in one place and foolish in another. Not scoring for five overs is strategy in a Test and suicide in a T20.

Layer three — filling data gaps wrongly. Tests give a large but slow sample; T20s a small but fast one. Data density differs by format. Whoever decides on a huge Test sample may err on a small T20 sample, and vice versa. Sample size and format speed must be read together.

Held together, these three layers explain why force-filling an empty cell is so dangerous. No data means I know nothing; but no data, plus filling the cell with the wrong format's data — the gap between these two is vast. The first is honesty, the second is deception.

A Framework of Data Discipline

In my work I use a simple framework I try to follow in every analysis. It is no magic formula, just a discipline to avoid format-blindness and the empty-data trap.

One — fix the question. Question before numbers. If the question is vague, no matter how large the data, the answer stays vague. I often see people start with data, not a question. So what they find differs from what they meant to seek.

Two — fix the format. Which game, which version, which conditions — this should be written before the analysis. Cricket analysis without format is impossible, because every decision's logic is tied to the format.

Three — supply a benchmark. Before quoting a number, say what range makes it good or bad. A benchmarkless number is mere decoration.

Four — admit the empty cell. Where data is absent, state plainly: insufficient information, cannot assess. This is not weakness; it is professionalism.

Five — publish the limitations. Every model has limits. Hide them and the reader trusts the conclusion while you know not how far it stands.

These five steps are like vows to me. Every transfer window is a monastery where numbers take vows. There is no room for ego here, only for discipline.

Negative Space in Emerging Markets

My real interest is markets like Bangladesh, the UAE and associate cricket. Here data is thin, cameras few, stadiums often empty. So mainstream models undervalue these players — the model sees only what is easily measured and skips what is hard.

In this gap I hunt for opportunity. A bowler's economy rate may suggest he is cheap. But if his overs are usually in the powerplay or at the death — the most risky phases — the same economy's value transforms. Contextless numbers mislead; contextual numbers illuminate.

Here is my favourite line: the database did not replace the game; it translated it. A player invisible to the mainstream often sits outside our narrow yardstick — we do not see him because our instruments never learned to measure him.

From 2026 to 2026 I built a live PPDA and pressure dashboard counting pass accuracy and progressive passes under pressure. A live dashboard is a heartbeat with a refresh rate. Whose decisions hold under pressure, and whose break — that is the real data, not just runs or wickets.

This led me toward transfer arbitrage. In 2026 I modelled a young Benfica midfielder at eighteen million euros before the Qatar World Cup; after the tournament his price leapt. I do not predict transfers; I reconcile the lag between rumor and contract. The gap between market price and true value is my field.

Auditing Transfer Failure

In 2026 I built an xG-based shortlist for a Liga 1 club. My top recommendation was a 24-year-old striker with 0.58 xG per 90 and 4.1 pressures per 90. The club instead signed a 34-year-old veteran on higher wages. The veteran scored two goals in sixteen matches, and the club collapsed from fourth to eleventh.

This changed my analysis. Now I write not only success stories but failure post-mortems, quantifying opportunity cost and recovery windows. In a decision memo I separate process quality from outcome luck, because a good process sometimes yields a bad result, and a bad process sometimes wins. Confuse the two and learning stops.

Contrarian Angle — Correlation, Not Causation

Now the part that is, to me, cricket data's biggest trap. We often see a relationship between two things and jump to a conclusion, when relationship and cause are entirely different. A team wins more matches and scores more runs per match — there is a relationship, but which created which is absent from the data.

Here format-blindness meets another error: over-modeling. I am an INTJ by nature; deep down I crave sealing everything into a closed, clean system. But cricket is not a closed system. The toss, weather, dew, pitch behaviour, DLS and human beings together form a chaos no model fully captures.

So I now add a section to every deep analysis called 'unmodelled variance.' There I state plainly what my model cannot capture and how much it might affect the outcome. It is a method for lowering my ego. An analyst who will not admit his model's limits is not running the model — the model is running him.

Another contrarian truth: empty data sometimes tells more truth than abundant data. If a batter's shot map in associate cricket shows the cover region nearly blank, that may not be his weakness — it may be that he rarely received the ball there, because opponents avoided that region. A blank region is itself a message. But reading it needs format, context and the opponent's tactics.

Here I try to break my circle of solitude. I like working alone, but solitude has a danger: checking my own errors, I may manufacture more of my own. So I cross-check models with a video scout friend who matches what his eye sees to my numbers. My own reckoning says my greatest enemy as a process auditor is my own confidence.

Another trap is sitting in the judge's seat while auditing process. When a side picks a veteran and drops a youngster, it is easy to call it folly. But behind the decision may lie a constraint — financial obligation, dressing-room balance, sponsor pressure, or a coach's tactical philosophy my model does not see. My job is not to judge; my job is to separate process quality from outcome luck so learning can happen.

And most importantly, I try not to see a player only as a mispriced asset. Behind the number is a human with an age, a language, a family, a political-economic context. If skill is cheaply available, that is my market opportunity; but that player's career, livelihood and dignity should also be part of my reckoning, or the analysis turns cruel.

Takeaway — Next Match's Signal

Next time you read a cricket analysis — or write one — ask first: which format, which context, and where is the benchmark? If the answers do not come, the number may be pretty, but it is not analysis.

My own experience says that in the regular season the most valuable information hides in the most boring places — why a bowler tires after four overs, why a team slows after the powerplay, why a youngster gets a chance and cannot take it. These are not headlines; they are signals. And signals always arrive before headlines.

I have learned that the real work begins once you stop fearing the empty cell. Admitting what is unknown, reading what is known with format-awareness, and labelling estimates as estimates — these three are perhaps cricket data's greatest discipline. In the next over, the next match, the truth may hide in exactly this place — if we learn to see the format before the number.