The Quiet Audit of the Franchise Market: Bangladesh's Immutable T20 Data Ledger

The franchise auction papers for the 2026 domestic season landed on my desk one afternoon. One team had released two experienced right-handed middle-order batters and replaced them with two youngsters whose franchise cricket record sits close to zero. On first reading, the decision looked self-destructive. Then I placed three seasons of phase-adjusted strike rates, ball-by-ball death-over logs and opponent bowling matchups side by side. The pattern surfaced: the team was not buying batters, it was buying safe overs. The retention fee for the two youngsters was roughly 40 percent lower than the veterans they replaced, and that saving kept two specialist death bowlers on the roster. In the market's language, that is a cheap call. In the data's language, it is a hypothesis whose sample is not yet large enough. The notebook was my first model, and Mymensingh was my first laboratory. In 2026, aged twenty-one, I logged 180 shots from twelve matches by hand — which ball, from which angle, struck with which body part, from what distance. The expected score that emerged from those numbers belonged to football. After Abahani Limited Dhaka's 2-0 win, my first piece argued the scoreline flattered Abahani, whose expected score was only 1.3. Those four thousand reads taught me a habit that still holds: an analysis opens with data, not with a lede. I did not discover expected goals; I submitted to them, one page at a time. Carrying that same discipline into cricket is harder. In football a shot is one event; in cricket a ball is one event, but the number of possible outcomes is limited and the value of each ball shifts by over. A boundary in the first six overs is worth less than a boundary in the last six, yet the scorebook records both identically. That asymmetry is domestic cricket's largest data gap. In a league that stages forty-six matches a season, what does a good season mean for a batter? Perhaps three hundred balls. Three hundred balls make a batter's true phase-by-phase skill hard to isolate, though not impossible. In Bangladesh's franchise market, a player's price is still set by total runs, total wickets and a handful of eye-catching innings. Phase, matchup, ball age, field setting, bowler type — these are almost absent. I have spent years watching matches from the small grounds of Mymensingh, Bogura and Dhaka. The best calls there were made from handwritten scorebooks and a coach's memory. Those notes deserve better than the bin. They were my first database, just written in ink. When the stadiums emptied in 2026, my model kept counting ghosts. After the pandemic break I audited 306 empty-stadium matches, and the home-advantage coefficient fell from 0.41 goals to 0.17. I refused to update the model until a twenty-match sample accumulated. I spent six weeks re-watching Project Restart fixtures and tagging crowd noise. That stretch taught me that when context shifts, data shifts too — and cricket's context shifts fastest of all. What the franchise market needs most is the phase-adjusted strike rate. A plain strike rate tells you how many runs a batter scores per hundred balls, but not when. A batter who strikes at 140 in the powerplay and 110 through the middle is really a different role. In T20 the three phases of an innings are not equal in value. The powerplay has the field up and boundaries are easier; the middle overs bring spinners and squeezed runs; the last four overs turn everyone boundary-hunting. A batter's true worth becomes visible only when his runs are split by phase and compared to the league average for each. In my own ledger I log every ball of every domestic match. Across thirty-four domestic matches in the last two seasons I found one batter striking at 138 in the powerplay, 104 in the middle and 162 in the last four. His overall strike rate of 131 reads superb. Yet if he is sent in at number four, his real skill is that middle-overs 104, which sits below the league average. The auction still priced him on the overall 131. That is where the market and the model walk apart. The death overs are crueller. Outcomes per ball are least predictable in the last four overs, and that is exactly where matches are decided. To price a death bowler I read three numbers together: death-over economy, his yorker ratio in the death, and how often he loses his wicket-taking threat under pressure. A bowler holding an economy of 8.2 in the death but conceding two full tosses an over has a handsome number and high risk. Without field setting and a bowling plan, economy alone says little. The matchup model is decisive here. In T20 a left-arm spinner is far more effective against a right-handed batter than a right-arm spinner is. In my own database I have found specific pairings where a batter's strike rate swings by more than forty points depending on the bowler's type. The auction sheet has no room for that pairing. Yet a team's composition is built precisely on it. Expected wickets is a harder idea in cricket than expected goals is in football, because a wicket sits largely outside the bowler's control — dropped catches, run-outs, umpiring calls. Even so, line, length, pace and the batter's shot selection can yield an expected wicket probability. I build it carefully, never on a small sample. I trust numbers, but only after they have survived a cold night of rechecking. Everything folds into a simple valuation frame: a player's price should combine his phase-based contribution, his matchup dependence and the scarcity of his role. Total runs and total wickets are none of those three. A team that buys a batter on total runs alone is buying the most overpriced mistake in the market. Take an example. An opener made 420 runs last season at a strike rate of 138. Striking. But 240 of those runs came in the powerplay, where the field is up. Against spin in the middle overs his strike rate was 98, and in the death he barely batted. If the team plays him at number three, he loses the benefit of those 240 runs. The market priced him as an opener, but the team's need is middle-overs output. That is the widest gap of all. The reverse also holds. A slow-ball middle-order batter with an overall strike rate of just 118 can contribute more to team success — if most of his runs come in the middle overs, under pressure, against spin. If the league average in the middle is 110, his 118 is actually league-leading. But the auction sheet prints 118 beside his name, and 140 beside a fast-scoring opener's. The market reads the two differently, though by role their value may invert. Retention economics folds in here too. When a team releases a veteran batter for a youngster, it is not only saving money, it is changing roles. The veteran's retention fee is higher, but his replaceability is lower, because his phase profile is known. The youngster's phase profile is unknown, so the risk is higher but the price lower. A team that can measure that risk with data profits in this market. A team that cannot is gambling blind. I sat through a domestic final in 2026 and watched a side chase 180, needing 32 from the last four overs, with its best death batter already out. The man at the crease had a death strike rate of 119; the batter on the bench had 158. The side lost by eight runs. On paper it was the strongest team. By role-based accounting it was the weakest. Since that night I have kept a separate death index for every side. Here I have to be careful. The phase-adjusted strike rate is a strong tool, but it does not stand alone. If I read only phase, I ignore field setting, ball age, pitch behaviour and weather. On a dry pitch spin is worth more; on a dew-soaked pitch death bowling is harder. One number is never the whole picture. Correlation is not causation — a batter with more phase-based runs is not automatically more valuable to his side, and that leap cannot be made directly. Sample size is another trap. Measuring a batter's true skill from one season's three hundred balls is risky. Luck weighs heavier in small samples, and the franchise market turns over its largest sums in precisely those small samples. Teams buy on one season's flash, just as transfer rumours and esports upsets are both variables waiting for sample size. Until the sample grows, any claim is only a hypothesis. The broken model taught me more than the accurate one ever did. In 2026 I had not assumed the home-advantage would collapse, because my sample was large but old. Every database carries an expiry date that nobody prints on it. In domestic cricket that expiry comes faster, because pitch, ball and players all change quickly. An analyst who will not admit the expiry is walking a new ground with an old map. In the next auction I will watch three signals. Which teams buy batters by phase role, and which buy on total runs. Which teams retain specialist death bowling, and which fill the death with middle-overs bowlers. How closely a young player's price tracks his data profile, and how much of it is pure scent of promise. The day those three signals align, the market and the model will walk the same road. Until then, my ledger stays open.