Cricket Analytics' Most Honest Finding: 'Insufficient Information'

On Monday morning, Stage One of our two-stage analysis pipeline came back to the desk in Brisbane. Ten pages of template, all eight dimensions present, and in every row the same line — 'insufficient information, cannot assess.' My first reaction was that the system had failed, that the person filling the form had ducked responsibility. The second thought slowed the first one down. What looked like a gap in that sheet was, in fact, the most honest cricket document of the week. A data desk produces documents like this every day and refuses to publish them every day, because a blank sheet does not sell to a broadcaster.

21 May 2026, Hyderabad. Rising Pune Supergiant's first full-time analyst desk lost a final to Mumbai Indians by one run. The model had already flagged Mumbai's death-bowling edge — Jasprit Bumrah and Mitchell Johnson, both under seven an over. Nobody acted on the number. The model was right; the delivery channel was wrong. The one-run final taught me that a margin is a story with decimals, and it taught me something else: a number that never reaches a room is not analysis, it is a filing cabinet.

After that defeat came a weekly column for a Pune digital outlet, and with it a rule hammered into place: no tactical claim goes out without a figure and a stated sample size. A phrase like 'death-over weakness' does not leave my desk today unless the economy figure and the innings count travel with it. That rule has moved in one direction for nine years — the definition of an information point has become stricter. An information point is a sentence carrying four things: the claim, the number, the sample, the date. Remove one and the sentence becomes an opinion, and opinions do not price.

The pipeline runs in two stages. Stage One breaks any material into information points; Stage Two maps those points onto cricket's eight dimensions — format, player technique, team landscape, league economics, governance, risk, public narrative, industry transmission. If Stage One returns empty, Stage Two has no basis for setting a price. The model is not the truth; it is the argument we can test. And where there is no argument to test, silence is the only valid output. The problem is that the industry reads that silence as failure, when in most cases it is the cheapest and most accurate finding available.

Let me take three of my own cases, because keeping receipts on yourself is this desk's only capital. First, the 2026 empty-stadium experiment. Covid pushed the IPL to the UAE, franchises cut staff, and I sat alone pulling nine seasons of results. Nominal home win rates ran at roughly 57 percent with crowds from 2026 to 2026, and close to 50 percent across the 60 crowdless matches in Abu Dhabi, Dubai and Sharjah. The gap was visible, but it was not the real question. The real question was how much of that gap survived once toss and dew were stripped out. Very little survived. In the floodlights, dew had replaced the crowd, and I have not used the word 'fortress' since. Every analysis I file now carries a short methods footnote — sample, venue, crowd status — and readers quote those footnotes back at me in arguments. I treat that as evidence the vocabulary is doing its job.

Second case: the 2026 T20 World Cup and the intent index. Working for a broadcast data desk, I measured the first ten balls of an innings — runs scored, wickets lost. On the slow UAE surfaces, India's top order was running below the tournament's powerplay baseline. India lost to Pakistan by ten wickets in Dubai and exited in the group stage. Getting the graphic onto the India–Pakistan broadcast took three attempts at the same argument in escalating formats — first a line, then a paragraph, then a table. The producers agreed on the third. A full-time data desk turns pressure into a repeatable process, but the process still needs someone at the other end willing to listen. Between matches I filed a daily Tokyo 2026 briefing, and the lesson there was identical: a metric's strength lives in its sequence of presentation, not in its arithmetic.

Third case, the most uncomfortable. The Impact Player rule arrived in 2026 and I modelled its effect on all-rounder valuations. The direction was obvious: the rule repriced the role that does two jobs. Then the model put a brake on itself. Comparing all-rounder prices before and after the rule is invalid, because the rule changed the population itself — the role you are measuring no longer has the same definition. This is the rare case where 'insufficient information' is the right answer, not ignorance but discipline. When the population shifts, historical comparison manufactures confusion. The more plainly a franchise is told this, the safer its purse.

Fourth case is the 2026 auction room in Bengaluru. My runs-above-replacement-per-crore model pushed Kings XI Punjab to KL Rahul at ₹11 crore and Andrew Tye at ₹7.2 crore. Rahul made 659 runs; Tye took 24 wickets and won the Purple Cap. The story sells, but the story is incomplete: a sample of one auction, fourteen matches, one franchise's purse. The auction room made every bid feel like a hypothesis with a deadline, and that room demanded a decision inside eight weeks that would be judged across three seasons. What the model did was publish an error margin alongside every bid. That is the difference between two kinds of analyst — one gives you the number, the other gives you the number and its boundary.

One cross-sport borrowing is necessary here, with a hard test attached. Baseball's replacement level maps onto cricket because cricket also has a countable currency — runs — and a bounded innings. The loan works. Where it breaks is equally clear: in baseball an out is a finite resource, in cricket a wicket is simultaneously a resource and a risk. So cricket's runs-above-replacement has to add wicket preservation as a separate cost. Every metric I imported from another code was tested against four cricket constraints — pitch, ball condition, over limits, wicket preservation — and whatever failed was dropped in public. That hygiene matters, because a borrowed metric is most seductive precisely where the mapping is weakest.

Now the other side. The greatest danger of a blank sheet is not its honesty; it is its use as shelter. Any desk can make itself bulletproof by writing 'insufficient information.' I saw that trap first-hand in the 2026 modelling: refusing to price all-rounders was intellectually clean and commercially useless. The franchise had to bid in eight weeks, and my answer was that the population had changed. That was not the right answer. The right answer was to publish an estimate with an error bar, run three scenarios — rule stays, rule goes, rule is modified — and disclose the probability attached to each.

That is a small confession of the most common error in this profession: using a null finding as an excuse to avoid a decision. Every data desk should audit its own nulls. My rule now is that one contrarian brief per tournament goes to the junior analyst and the disagreement is printed unedited. A one-man desk stays honest through a season only if there is someone inside it willing to object.

The 2026 T20 World Cup will be played in India and Sri Lanka and the pre-tournament model is being built right now. The number I am tracking hardest is not a projected final score — it is the first-ten-balls powerplay intent index, because in 2026 that was the number standing between winning the tournament and leaving in the group stage. The question is not who has the better data. The question is who has the nerve to walk it into the room where the decision is made — and when the answer is nobody, will our desk say so out loud?