The Price of an Auction and the Price of a Field: The Seven Who Made No List

The retention list landed around seven in the evening. Fourteen names, four more flagged as available on loan. I was at my work table in Khulna, turning the pages of an old notebook — ball-by-ball figures from the 2026-26 domestic 50-over season, counted by my own hand. Reading the list, I stopped at one place.

The bowler who conceded 6.4 an over after the 35th over across the season is nowhere among those fourteen names. The one who is, conceded 7.9 in the same phase. The gap is a run and a half per over, but it accumulates across 43 death overs — roughly 65 runs. Nobody asked why.

The number was not new to me. What was new was seeing it plainly: in this market, price and performance are not written in the same currency.

Neither the Dhaka Premier League nor the Bangladesh Premier League has ball-by-ball data that anyone sells here. In the IPL, several companies build a feed for every delivery; here a domestic scorecard stops at a table: runs, wickets, overs. Where the ball pitched, which foot the batter was on, where the keeper stood — all of it vanishes into the margin of a scoresheet.

That absence is not neutral. A player with no record cannot produce evidence at a negotiating table. Yet almost every franchise decision is now taken at a table — who stays, who is released, how much of the wage bill sits in which slot, what the release clause permits. In this transfer window, Bangladesh's franchises hold on to familiar names while spending heavily on overseas slots. There is effectively no public method for valuing local bowlers. The overseas names in circulation are proven brands; a left-arm quick like Mustafizur Rahman built his place in international franchise markets over years, and that has become the standard of expectation here.

The wage bill is the least discussed part. Inside a squad limit, a few slots belong to national players and the rest to domestic ones. Spend one of those slots on a player with no public death-over evidence, and the decision rests on a seven-minute clip and memory. The release clause then decides where he goes mid-season. That is the whole of it.

I have worked this way since 2026. Building a model by hand for a football league taught me that a model can stand without a provider's feed — shot angle, distance, defensive pressure. That season the model placed a young midfielder above the league's leading scorer. The piece did not travel far, but I kept the notebook. Coming back to cricket, the habit came with me.

What I have seen from the stands year after year: a domestic bowler's best days happen outside the camera's line, and his most visible days may be his worst.

Before the indices, the method — because a story without numbers is only a story.

Sample: domestic 50-over competition, 2026-26 season, 58 matches. Cut-off date: 12 February 2026. Legal deliveries: a little under 32,000.

I built two indices.

The first, the Death-Over Pressure Index (DPI) — runs conceded per over after the 35th, adjusted for the batter's domestic strike rate, wickets falling, and the required rate. A bowler who took the last overs after five wickets had gone down carries that reality inside his figure.

The second, Impact Per Phase (IPP) — separate scores for powerplay, middle and death, weighted 25, 25 and 50 per cent. The last ten overs are the hardest work in cricket, so that is where the weight sits.

Then I attached a market price to roughly two dozen names — retention value, auction sum, or simply how often a name is called in popular cricket conversation.

What came out: price against powerplay strike rate correlates at 0.67 — strong. Price against DPI correlates at 0.21 — weak. The market treats as real the metric that is most legible on a scorecard: powerplay runs and total wickets.

Seven bowlers entered my top ten by DPI. None of them is on that retention list.

Something else caught my eye. Five of those seven had fewer legal deliveries than their team's top four bowlers — because they were not picked for the big matches, or never given a new-ball over. Where the price figure comes from the most visible place, nobody opens the ledger of opportunity.

Now the confession that sits in every piece I file. My model cannot see where three fielders are standing near the boundary. It does not know what the ball will do if the pitch is damp. It does not measure wind, does not know about a batter's injury, and understands nothing at all about dressing-room politics. These numbers are questions, not verdicts.

Every number is a person who never got to explain themselves.

One example. A left-arm pacer bowled 43 overs in the death phase last season. His economy was 6.4. In several matches he came on in situations where the batting side needed twelve an over, and he never once went for nine. The scorecard says he conceded at 6.4. My notebook says the second line is a record of context, not of talent.

Look from the other side. A spinner took 18 wickets in 14 matches at an economy of 5.9 — handsome figures. But eleven of those 18 wickets came in innings where the opposition had already passed 220 and was taking risks. His real cost in the death phase was 9.1. In conversation, he was cheap and reliable.

Take an opener with an overall strike rate of 140 — 155 in the powerplay, 112 at the death. Two numbers, two different stories, and which one gets written depends on which scout saw which clip first.

Go to the women's domestic circuit and the picture is bleaker. There, the question of ball-by-ball data does not even arise; many scorecards are not updated online either. There is no route to valuing a player because the route was never built. Bowlers like Nigar Sultana Joty, Nahida Akter or Marufa Akter have no public death-over model; the work they do has no document filed anywhere.

One thing about method, plainly. Every match, I drew ball-by-ball lines onto a paper grid — over, bowler, batter, runs, wicket probability. Then at night I moved it into a spreadsheet. 58 matches means 58 nights. Nobody pays for this work, so nobody does it. I built the model by hand, because the league deserved to be counted. No provider's feed existed there, so counting ball after ball became a kind of prayer.

Now look at the franchise side. Much of what scouts bring to the table is a seven-minute clip sent by an agent. Retention decisions revolve around three questions — will the name pull a crowd, is there room in the wage bill, will next year's release clause set a trap. Performance enters inside those questions; it does not knock on the door from outside.

Transfers are stories wearing spreadsheets like coats. And the last page of the story is usually written in numbers, because a number ends the argument.

There is a list in my notes I call the ledger of noise. Statistics that sound meaningful and explain nothing. Total wickets sits at the top. What does 24 wickets mean for a bowler? That he kept his side in the contest 24 times, or that he bowled four of them in the last two overs when batters were forced to gamble? Overall economy falls into the same trap. Dot-ball percentage without context is a bigger trap still.

In one place I fought my own model. The hypothesis was that left-arm quicks gain an advantage at the death from the angle. The arithmetic said otherwise: correlation 0.08, essentially nothing. I discarded it. I tested a second idea — age. Bowlers over thirty averaged a DPI of 7.3, those under twenty-five 7.1. The difference fell inside my margin of error. Twice my model disappointed me, and that disappointment is what taught me to trust the rest of the numbers.

There is a trap here that is easiest of all in work like mine. Suppose I said the retention list is foolish and my index is the truth. Then I would be doing exactly what I complain about.

The truth is more indifferent. A weak correlation between price and DPI does not mean the market is wrong. It means the market is buying something else.

A franchise is not buying performance. It is buying scarcity and optionality. If a left-arm pacer does not fit inside the squad limit, saving ten runs buys nothing. Holding a retention slot means the side can play another card at the next auction. That is not cricket arithmetic; it is balance-sheet arithmetic. My table does not measure it, and trying to measure it would be wrong.

Another familiar line: our domestic bowling is weak. The line is not false, but it is incomplete. In my sample, the weakness is less about skill than about measurement. If nobody keeps a picture of the ball a bowler bowled, there is no way to demonstrate improvement either, and without proof of improvement opportunity does not grow. That is a loop.

What I distrust most is the part of the data that is most profitable. Live ball-by-ball feeds are now built largely for betting companies; each delivery's small statistics travel there within seconds. The player who is the raw material of that feed does not get a penny of the profit — and the decision on his price is taken from the most easily available slice of that feed, which is the smallest slice of his work. That is the darkest side of datafication, because there the data grows while the understanding does not.

So no single cricket number can judge a person — not an auction price, not my index. Both are partial light. The difference is only this: I can open my partial light and show it to you myself.

Two things to watch in the next retention window. First, which franchise publishes its own index first — not just the table, but the method and the margin of error. Second, who takes the new ball in the first two overs. A bowler given the first two overs of the powerplay is inside the system; one who arrives in the 16th over never reaches any table. Those two places will show whether this market's eyes are opening at all.