214 Names, Last Place: The Audit Story of a Cricket Analytics Desk

Starting with a metric anomaly: 214 names, and half of them ignored

In February 2026 the Lahore Qalandars analytics desk had three people, and I was the third — hired into a franchise that had finished bottom of PSL 2 with three wins from eight. I had one shot and one condition: turn every player into a number before the draft. I built a 214-player valuation sheet from scratch — every domestic cricketer with a T20 strike rate above 120 across the previous two National T20 Cups, graded into Platinum, Diamond, Gold, Silver and Emerging bands. The sheet was ready before the 2026 draft. This was not a cricket revolution; it was an Excel file with a date next to every row — the date on which that number could be proven wrong. The franchise ignored half the sheet. The half it used contained three Emerging picks, and by the end of the season those three had outscored the single Platinum signing on which the largest sum had been spent. That was the first moment I understood that price is a hypothesis and the points table is its test — putting both in the same column is professional suicide.

Context: the noise of the auction and the silence of the data

Franchise cricket's commercial apparatus runs on auction purses, retention calls, draft pools and valuation sheets — and this is its noisiest part. Every season hundreds of names are placed on a table, a sum is written next to each, and that sum is immediately treated as a quality score. This mistake is not only journalists'; it belongs to the desks that call themselves data-driven. In 2026, when I built that sheet, I had one simple question in front of me: does a domestic T20 strike rate survive against international or franchise-level bowling? In answer I recorded, for every player, the number of balls they had actually faced. Of the 214, sixty-one did not even have a 300-ball sample. That is, roughly a third of the sheet rested on a number any statistically serious audit would reject. From that day every claim I made carried a question with it: how many balls is this number built on? If there was no answer, that sentence left the paragraph.

Core: where xG stops and cricket begins

In 2026 I logged all 64 World Cup matches by hand — shot maps, xG, PPDA — in a Lahore flat with two monitors and a dead air conditioner. Then I committed the blasphemy: translating expected goals into cricket. I built "Expected Runs" off 41,000 ball-by-ball PSL deliveries, weighted by batter, bowler, phase and field restriction. I published it as a 12-part newsletter. Eleven parts were read by roughly forty people. Part seven — on death overs — was forwarded by an analyst in Multan and travelled from there.

But the real lesson was not the one readers saw. The real lesson was inside the model, and it is this: porting xG to cricket is a translation problem with a documented failure rate. In football a goal is a continuous event — chances are created, reduced, accumulated. In cricket a wicket is a discrete event: it either happens or it does not, and one wicket changes the weight of the entire model. Football produces dozens of shot events in 90 minutes; a T20 innings has 120 balls, and roughly half of them have an outcome of zero or one on the scoreboard. Even the best-built model generates as much noise as signal on a single-innings sample. I learned this after coding 64 matches, and it is why I now write boundary conditions of imported metrics rather than their marketing.

At 33, when I first built a franchise valuation sheet, I did not think that 21 years later I would still be asking the same question — only more strictly.

Contrarian angle: never put price and output in the same column

Here is the thing that is most dangerous for an analytics desk, and it is not the data itself. The danger is that the auction price is the most legible number in the ecosystem, so it silently becomes a quality score. A big purse buys a narrative that the table has not yet settled. In the 2026 sheet the mistake I made was merging three columns — price, role and output — into a single rating. Later I separated them and began tracking surplus value against contract, not salary against reputation. When you do that, you see something nobody saw on auction night: an Emerging pick delivering four times its contract value, and a Platinum signing delivering 0.6 times its price. Both sit in the same table, and putting both in the same column is the mistake every franchise repeats every season.

The second paradox: the most-quoted numbers in cricket are often the least useful. Hundreds, fifties, career totals — these are accumulation statistics, not situation-adjusted output. A batter averaging 40 at a strike rate of 140 can be more valuable than one averaging 55 at 120 — and that difference only shows when you look at phase-wise data, not the aggregate.

Takeaway: next time you watch an auction

When a player goes for a record sum at the next auction, do not treat that sum as proof of his performance. Treat it as a hypothesis that the next 18 months of the table will test. And when someone tells you they have built an advanced model, ask one question: how many balls is this number built on? If the answer is "enough," then it is not a model — it is an opinion placed in a spreadsheet. In cricket, where one wicket changes the weight of an entire match, sample size is the only honesty left.