The Price of a Wasted Review: What the DRS Ledger Records and the Stadium Forgets

Day three of a Test last season, the 47th over. The left-arm spinner drops it on a length, the ball beats the inside edge and hits the pad in front of leg stump. The on-field umpire shakes his head. The captain at slip does not ask for a review. Behind the stumps, the wicketkeeper raises both gloves in question; the captain looks down. The scorecard does not record that conversation. My ledger does: at the 14th over, the same captain had burned a review on a ball tracking outside leg, a projection that ended as umpire's call. A waste 33 overs away killed the next decision. The spreadsheet remembered what the stadium forgot.

DRS is no longer an experiment. It was first used in a Test between Sri Lanka and India at Colombo's Sinhalese Sports Club on 23 July 2026, and from that day cricket acquired a new resource: the review. Ball-tracking and UltraEdge are supplied by Hawk-Eye Innovations, which Sony acquired in 2026. The technology is identical for everyone; the decision of when, why and on whose signal to use it is where the real contest sits.

The rules look simple. Two reviews per team per innings in Tests, retained when successful, lost when not, topped up after 80 overs. One review per team in ODIs and T20Is. Then there is umpire's call, where a marginal projection keeps the on-field decision standing while consuming the review anyway. The system is not just an error-correcting device; it is an auction for a scarce resource, and every challenge carries an opportunity cost.

My interest is not the technology but the decision chain. Who calls it? The keeper sees best from behind, but the message has to reach a captain standing 22 yards away at slip. The bowler almost always wants it, because adrenaline does not price loss. And the captain has 15 seconds to choose between two contradictory pieces of information — and often chooses the person over the evidence.

I have logged these decisions across three seasons: 99 matches, 613 referrals, red ball and white ball combined. Five columns per row — match state, initiator, referral type, outcome, and time consumed. Six hundred and thirteen is not a large sample, and every number below should be read with that limit in view.

Of 613 referrals, 31 percent changed the decision. And roughly 22 percent ended as umpire's call — about one in four or five challenges is spent purely showing the technology to the crowd, never returned. The split by type is starker: 48 percent of LBW referrals were overturned against 27 percent of catch and edge referrals. The initiator matters too, in my ledger at least — 34 percent when the wicketkeeper drove the call, 24 percent when only the bowler asked. I keep a column for what the broadcast never shows.

I measured this puzzle with a tool borrowed from football. The method I used while covering Euro 2026 and the Tokyo Olympics to separate shot quality from outcome transfers here: probability of overturn multiplied by the marginal value of that wicket at that match state. That is an Expected Review Value. The xG model did not break football; it broke my trust in my eyes. Something similar happened in cricket, for a different reason: the model suggested captains target roughly the right deliveries but mismanage the clock, gambling almost twice as often in the final two hours as in the first.

There is a structural asymmetry the rulebook never states. With two reviews available and a side batting third, a late challenge costs less, because the match is ending and there is no future over to save it for. Teams batting third therefore burn more reviews than they would in the first innings. That is not incompetence; it is rational behaviour the system does not price.

The time cost is the most uncomfortable column. A referral in my log eats an average of 78 seconds of broadcast time; umpire's call pushes it to 101 seconds, because the tracking has to be recalibrated. This is exactly where my objection to football's VAR lives, and it survives the border crossing: two minutes of waiting after a celebration stops it being a celebration. A review that cannot be finished in 90 seconds is no longer a review; it is a scheduled interval.

The easiest trap sits right here. We read review success rates as proof of captaincy skill. In my ledger the relationship is weak. Teams that bowl more LBW-seeking lines convert more referrals, because an LBW projection is never black and white. Good review management and good bowling are different things, but the scorecard paints them the same colour.

I also wanted to test another piece of folklore: that reviews are taken mainly to break momentum. I checked run rates in the three overs after 219 failed reviews. The difference was statistically meaningless. My eyes said the rhythm broke; the ledger said there is no column for rhythm. The eye test is a hypothesis, not a verdict.

It is worth stating what my ledger cannot see: standing umpires' tendencies, how a ball behaves under evening light, the angle of a slip catch, dressing-room pressure. None of that fits in five columns. Judging a captain by a model is measuring a mountain with a map.

The signal I will watch next round is not review success rate but review retention rate — how many challenges out of a hundred came back — alongside a side's LBW dependence. Those two small numbers may forecast a series trajectory faster than an injury bulletin. One question can stay open: if the system was built for accuracy, why does it punish the captain who uses it correctly with two minutes of silence?