DRS and umpire’s call: the rules, the numbers, and why it divides everyone

Few things in cricket generate more confusion than a review that shows the ball hitting the stumps and still returns “not out”. The logic is sound. The presentation is terrible. Both things are true.

How many reviews you get

Format	Unsuccessful reviews per team
Test cricket	3 per innings, per 80 overs
One-day internationals	2 per innings
T20 internationals	2 per innings

The crucial word is unsuccessful. A successful review is retained, so a side that reviews well never runs out. A side that burns reviews on hopeful appeals in the first ten overs is defenceless later, and that has cost matches far more often than any single bad decision.

Players have 15 seconds from the umpire’s decision to signal a review. The T sign must come from a batter at the crease or the fielding captain — not from the dressing room, which is why you see the frantic glances upstairs inside that window and why boards have been fined for signalling from the balcony.

The three questions, in order

For an lbw review, ball tracking answers three questions in sequence, and a “no” at any stage ends it:

1. Pitching. Did the ball pitch in line or outside off? Any part of the ball pitching in line counts. Pitching outside leg is not out, full stop.
2. Impact. Was the point of impact in line with the stumps? (Unless no shot was offered, in which case impact outside off can still be out.)
3. Wickets. Would the ball have gone on to hit the stumps?

Each stage returns one of three verdicts: out, not out, or umpire’s call. The last means the projection falls within the system’s margin of error, and the on-field decision therefore stands.

Why umpire’s call exists

Ball tracking is a prediction, not a measurement. It tracks the ball to the point of impact and then projects a path forward. Everything after impact is modelled, not observed.

Every projection carries an error range. Umpire’s call is the system saying: where the projection falls inside that range, the technology cannot honestly claim more precision than it has, so the human decision stands.

That is intellectually honest, and it is the same principle used everywhere else that measurement uncertainty matters. It is also a very hard sell at volume in a full stadium, because two identical-looking deliveries can produce opposite outcomes depending on what the standing umpire originally gave.

The strategic consequence

Because the on-field call carries through marginal cases, the umpire’s original decision has real value — and both sides know it.

This produces a genuine tactic. Bowling teams appeal hard and appeal often for anything remotely close, because getting the finger up first means every marginal review goes their way. Batting sides know the same thing in reverse, which is why you see batters set off for a run immediately after being struck on the pad, hoping to make the appeal look absurd.

It also changes review strategy. A side reviewing a “not out” needs the projection to be conclusively hitting; a side reviewing an “out” only needs it to be marginal. Those are different bars, and good captains review accordingly.

What DRS is unambiguously good at

The debate focuses on umpire’s call and skips the parts that work perfectly:

• Inside edges. Snicko and hot spot remove an entire category of terrible lbw decisions.
• Ball pitching outside leg. Now settled definitively rather than guessed.
• Height. Deliveries bouncing over the stumps, previously a matter of opinion, are now a matter of fact.
• Caught behind. The single largest improvement in decision quality of the last twenty years.

Overall decision accuracy has risen substantially. The argument is about the last few percent.

The alternatives

Removing umpire’s call would make outcomes feel cleaner and would hand every marginal decision entirely to a projection with a known error range. That is a trade, not an improvement.

The other proposal — publishing the margin visually, so viewers can see how close a call actually was — is a presentation fix rather than a rules change, and it would probably do more for public understanding than any rewrite of the law.

As it stands, the current rule keeps a human in the loop precisely where the machine is least certain. Stated plainly, that is exactly where you would want one.