DRS in cricket explained, from ball tracking to umpire's call

The captain makes a T with his forearms, the fielders stop halfway through the celebration, and a batter who has already started walking turns round to watch three replays that everybody in the ground has an opinion about. That gesture is the only part of the Decision Review System most spectators ever see, and it is the smallest part of the machine.

Most of what gets explained about DRS in cricket on a broadcast is the ball-tracking graphic, because it is the part with pictures. The system is wider than that, and it was built to answer a narrower question than people assume. It was never meant to make every decision correct. It was meant to stop a handful of decisions being catastrophically wrong, which is a different ambition and produces a very different set of rules.

Miss that and the rest looks arbitrary. Understand it and almost every complaint about the system turns into a complaint about something else.

Two different things are called a review, and conflating them causes most of the confusion

There are two entirely separate procedures running under the same banner, and they behave nothing alike.

The player review. A fielding captain or a dismissed batter disputes an out or not out decision. It costs the team something if they are wrong, it is limited in number, and it exists because a batter given out to a ball that pitched a foot outside leg used to have no remedy at all.

The umpire review. The on-field umpire, unable to see something from where he is standing, refers it upstairs himself. Run outs. Stumpings. Whether a catch carried. Whether a fielder was touching the rope. Whether the batters had crossed. It costs nothing, it is unlimited, and no team's allowance is touched by it.

Nearly every argument about a side "wasting reviews" on a run out is a person applying the rules of the first procedure to the second. The teams have no say in an umpire review and no budget for it. When the square-leg umpire draws a rectangle in the air after a diving batter, that is not the batting side spending anything, and the fielding captain could not have stopped it if he wanted to.

The two systems also answer to different standards, which matters more than the labels. An umpire referral usually asks a question with a physical answer visible in the pictures: the bat was grounded or it was not, the ball had touched the rope or it had not. A player review often asks a question no camera can see directly, because the ball never went where the question is about. That gap is where umpire's call lives, and it is the whole reason the concept exists.

One more distinction resolves a surprising number of pub disputes. None of this is in the Laws of Cricket. The review system is a playing condition, written by whoever runs the competition, sitting on top of the code kept at Lord's. The lbw law itself is a law, and it is unchanged by any of this. The system does not decide who is out. It decides who gets to look again, and it applies the same lbw sequence the Laws already contain with better pictures.

• 15Seconds to ask for a review
• 3lbw components the tracking reports
• 3Detection systems in a full kit
• 0Marginal projections that overturn an umpire

Structural counts from the review protocol. The thermal camera is not deployed at every venue, so a full three-technology kit is not universal.

The review budget, and why the number changes with the format

The player review is rationed. Each side gets a set number of unsuccessful reviews per innings, and the arithmetic of what counts as unsuccessful is where the design shows.

A review that overturns the decision costs nothing. That is the point of it. The side was right, the umpire was wrong, and charging a team for correcting an error would discourage exactly the behaviour the system was built to encourage.

A review that comes back as umpire's call also costs nothing under the current conditions. This has not always been true, and the change was one of the more consequential the International Cricket Council has made. While a marginal projection burned a review, sides became reluctant to use them on genuinely close calls, which meant the reviews sat unused while bad decisions stood. Returning the review on an umpire's call removed the penalty for asking a reasonable question.

A review that is simply wrong costs one from the allowance.

The size of that allowance is set by format, and the reasoning is length. A Test innings can run to a hundred and fifty overs across two days. A Twenty20 innings is twenty overs and might be finished inside ninety minutes. Giving both the same allowance would either flood the short game with stoppages or leave the long one defenceless by lunch on day two, so Tests carry the larger number and white-ball cricket the smaller. The exact figures live in the playing conditions of each competition, and they have been amended more than once, including a temporary increase during the period when neutral umpires could not travel.

The top-up rule is the part that has come and gone. In its usual form, a Test side's allowance resets at a fixed point deep in a long innings, most commonly tied to the over at which a new ball becomes available. The problem it solves is real. A captain who spends his reviews in the first hour, whether through bad luck or bad judgement, is then unprotected for another four sessions, and a howler in the ninetieth over is exactly as damaging as one in the ninth. The counter-argument is that a reset rewards profligacy and produces speculative reviews just before it arrives, which is the sort of behaviour that makes a system look silly. The ICC has both used and removed the top-up, and the retention of reviews on umpire's call reduced the pressure for it, because sides now lose far fewer reviews to close calls in the first place. Whether it is in force in the match in front of you is a question for that competition's conditions, not for the Laws.

Two smaller rules decide more matches than they should. A review is per innings, not per match, so an allowance untouched in the first innings does not carry over. And in a run chase, the batting side's reviews are the ones under real pressure, because a wrong decision late in a chase cannot be corrected by an equalising innings that no longer exists.

The three technologies, and what each one can honestly settle

The kit is usually described as one system. It is three, they answer different questions, and knowing which one is being consulted tells you how much weight to put on the answer.

Technology	What it does	What it cannot resolve
Ball tracking	Reconstructs the ball's path from multiple synchronised high-speed cameras, then projects it forward past the point of impact	Anything about the ball after it hit the pad, because that path never happened
Audio edge detection	Matches a sound waveform to the video frame in which the ball passes the bat, looking for the signature of a hard collision	Which object made the sound, when bat, pad and ground are all within a few centimetres
Thermal imaging	Films in infrared and shows the heat generated by friction where two surfaces meet	A contact too faint to raise a detectable temperature, or one outside the camera's angle

Ball tracking predicts as much as it measures

This is the one people treat as a measurement, and half of it is not.

Several fixed high-speed cameras watch the pitch from different angles. The system identifies the ball in each frame, triangulates its position in three dimensions, and builds the flight path. That part is measurement, and it is genuinely accurate, because the ball is a known size moving through a calibrated volume watched from several viewpoints at once.

Then the ball hits the pad and the measuring stops. The question the umpire needs answered is what the ball would have done next, and there is no footage of a thing that did not happen. So the system extrapolates: it takes the speed, the angle, the bounce off the surface and the deviation already observed, and models the rest of the flight to the stumps.

That extrapolation is very good and it is still a prediction. Its reliability depends on how much measured path it has to work with. A ball pitching on a length and travelling a decent distance before striking the pad gives the model plenty of evidence. A ball that pitches and hits the pad almost immediately gives it very little, and the playing conditions recognise this: below a defined minimum distance between the pitching point and the impact, the tracking output is not treated as sufficient to overturn anybody, and the on-field decision stands. Length matters in the other direction too. A batter struck a long way down the pitch leaves the model projecting a long way forward, and small errors in the modelled deviation grow with distance.

The surface itself is the awkward variable. The model has to assume the ball behaves normally after impact, and pitches do not always cooperate. A ball that struck a crack, gripped in a footmark or skidded off a bare patch has already behaved unusually once, and the projection cannot know whether it was about to do so again. Anyone who has watched a fifth-day surface knows that the character of the pitch changes what a delivery does between one over and the next.

Edge detection by sound

The audio system takes the stump microphone feed, or a dedicated microphone, and lines the waveform up frame by frame with the video. As the ball passes the bat, the operator looks at what the sound is doing in that precise frame.

What it is really looking for is a shape, not a volume. The collision of a hard ball with willow produces a short, sharp spike with a fast attack and a fast decay. Ball on pad produces something duller and broader. Bat on ground produces something different again. A trained eye reads the shape, and the placement of that shape against the frame is what makes the evidence usable.

Its limits are physical and unfixable. Bat, pad, ground and boot can all be within a hand's width of each other at the moment of contact, and at the frame rates involved two events a fraction apart can be indistinguishable. A batter who plays inside the line and clips the flap of the pad with the bat as the ball goes past the outside edge generates a real spike from a real collision that has nothing to do with the ball. This is the most common way an audio review misleads, and it is why the operator's question is never "was there a noise" but "was there a noise consistent with this ball striking this bat in this frame".

The keeper knows all of this instinctively, which is why the appeal from behind the stumps carries information the crowd does not have. Standing up or standing back, the wicketkeeper hears and feels the difference between an edge and a pad long before any waveform is drawn.

Edge detection by heat

The thermal camera measures infrared radiation, and it shows a bright mark where friction has warmed a surface. Ball on bat leaves a mark on the bat. Ball on pad leaves one on the pad. Bat on pad leaves one on both and none anywhere near the ball's line.

Its great virtue is that it locates the contact in space. Sound tells you something happened. Heat tells you where. A mark on the outside edge is not open to the interpretation a spike is.

Its weakness is sensitivity. The energy involved in a very fine edge, the sort that changes the ball's line by almost nothing, may simply not raise a temperature the camera can distinguish from the background. Absence of a mark is therefore weak evidence of absence of contact, and the third umpire is not entitled to treat it as proof. Add that the cameras cover specific angles, that a hand or an arm can block the view, and that the equipment is expensive enough for many series to be played without it, and thermal imaging becomes a strong confirming tool rather than a universal one.

Why they disagree, and what to do when they do

The two edge systems are not two measurements of the same thing. They are measurements of two different consequences of the same event, and the consequences are not equally reliable indicators.

A very fine edge: clear spike, no heat mark. The sound is real and the camera is not sensitive enough.

A bat clipping the pad: heat mark on the pad, spike in the audio, and no contact between ball and bat whatsoever. Both technologies are reporting truthfully about an event irrelevant to the appeal.

A ball brushing the pad on its way past the bat: dull sound, heat mark on the pad, and the question of whether the bat was involved at all still unanswered.

When they conflict, the protocol is not to average them. The third umpire has to decide whether the available evidence is conclusive, and where it is not, the on-field decision stands. That last clause does enormous work throughout the system and it is worth restating in plain terms. The technology does not have to prove the umpire right. It has to prove him wrong, and where it cannot, he wins by default.

Umpire's call in full, and why a margin has to exist at all

Umpire's call is the most complained-about phrase in cricket, and almost all of the complaints are about something the phrase does not say.

Start with what it is. When a review reaches the ball-tracking stage, the system reports three components separately: where the ball pitched, where it struck the batter, and whether it was going on to hit the stumps. Each of those has a zone drawn around it, defined in the playing conditions. If the ball's position relative to that zone is clear, the finding is conclusive and can overturn the on-field decision. If it is marginal, the finding is umpire's call, and the on-field decision stays exactly as it was.

For pitching and for impact, the test is about the centre of the ball against the line of the stumps. A ball whose centre pitched outside leg is not out with no further discussion. A ball straddling that line is umpire's call, which means it depends entirely on what the umpire said in the first place.

For the projection, the test is about a defined box around the stumps. A ball predicted to hit that box conclusively overturns a not out. A ball predicted only to clip the outside of it is umpire's call. The exact dimensions of that box are set by the conditions and have been adjusted, notably in how far up the stumps the zone extends, and each adjustment shifts the boundary between hitting and clipping for a whole class of deliveries.

Now the part people skip. The margin is not a courtesy extended to umpires. It is an admission about the evidence.

The projection is a model output, and every model output carries an uncertainty. If the model says the ball would have clipped the outer edge of leg stump, the honest reading is that the truth lies somewhere in a band around that answer, and the band includes both hitting and missing. There is no way to resolve it, because the actual ball is in the wicketkeeper's gloves and the delivery in question was interrupted by a pad.

Given that, a system obliged to produce a verdict has exactly three options. Give every marginal case out. Give every marginal case not out. Or leave it with whoever has already made a decision.

The first two are worse. Both apply a systematic bias to a category of deliveries where nobody knows the answer, and both produce identical treatment for a plumb-looking appeal and a shaved one. The third does not pretend to knowledge nobody has, and it preserves the principle the review system was built around: the umpire's decision is the default, and it is displaced only by evidence.

The unavoidable consequence is that two identical replays produce opposite results, because the two original decisions were opposite. That is not the system contradicting itself. It is the system saying "we cannot tell" twice, in a format that requires a batter to be either out or not out.

The argument about umpire's call, given properly on both sides

This is a real disagreement between serious people, and it is usually presented as one side being stupid. It is not.

The case for scrapping it. The system asks a binary question and the technology gives a binary answer. If the tracking says the ball hits the stumps, it hits the stumps, and the batter is out. Anything else means two players in the same match, facing the same delivery, get different outcomes for reasons entirely outside their control. A batting side is asked to accept a dismissal on a ball shown to be missing leg stump, because the umpire's finger went up first. Critics on this side add that the system already trusts ball tracking enough to overturn decisions outright, so treating the same output as unreliable when it is marginal is inconsistent: either you believe the model or you do not. There is also a simplicity argument that should not be waved away. A rule the crowd cannot explain to a newcomer in one sentence carries a cost, and the review system's public credibility is not improved by outcomes that look like coin flips.

The case for keeping it. A binary answer is not the same as a certain answer. The tracking output is a prediction with a genuine error band, and a system that treated the band's midpoint as fact would be manufacturing precision it does not have. Remove umpire's call and marginal deliveries stop being marginal by decree, which is a decision to be wrong in a specific direction rather than to admit uncertainty. There is a second argument, about the shape of the game rather than the accuracy of the machine. Umpire's call is what keeps the standing umpire the primary decision-maker. Abolish it and the on-field call becomes provisional in every close case, a placeholder awaiting confirmation. Cricket has spent two centuries building a game around the authority of the two people in the middle, and that authority is not a sentimental attachment. It is what keeps the game playable at every level below the ones with cameras.

The middle position, and it is where the actual reform has happened, is to keep the concept and argue about the boundaries. Widening or narrowing the zone changes how many deliveries fall into the uncertain band without pretending the band does not exist. The ICC has moved those boundaries, and each move produced a fresh round of complaint from whichever side it disadvantaged, which is roughly what you would expect from a compromise that is working.

Anyone who has followed the equivalent argument in football will find the shape of it familiar. The dispute over whether a marginal offside should be settled by a line drawn to the millimetre is the same dispute in different clothes, and the way video review changed football's relationship with its own referees is a useful comparison, because that sport chose a different answer and inherited a different set of problems.

The lbw tests, in the order a review actually applies them

The review does not assess an lbw appeal as a single judgement. It runs the law's tests in sequence, and a failure at any step ends the process. This is why a review can be turned down for a reason that has nothing to do with the graphic everybody was waiting for.

1. 1Was the delivery legal?A no ball ends it immediately, because lbw is not available off an illegal delivery. This is checked first precisely because it makes every later question pointless.
2. 2Did the ball hit the bat first?The audio and thermal evidence is examined here. Any contact with the bat, or the glove holding the bat, before the pad means not out regardless of where the ball was heading.
3. 3Where did the ball pitch?Outside the line of leg stump is not out, with no appeal to anything that follows. Marginal is umpire's call, which returns the original decision.
4. 4Where did it strike the batter?In line between wicket and wicket, the appeal survives whether or not a stroke was played. Outside off, it survives only if no genuine attempt was made to play the ball. Outside leg, it dies.
5. 5Was it going on to hit the stumps?Only now does the projection matter. Conclusively hitting overturns a not out. Conclusively missing overturns an out. Anything in between is umpire's call.
6. 6What was the original decision?On any umpire's call finding, the answer to this question is the answer to the review. Two identical tracking outputs produce opposite results because the two on-field decisions were opposite.

The sequence stops at the first failure. An appeal that dies on pitching is never assessed for whether the ball would have hit the stumps.

The sequence explains a specific piece of crowd anger. A ball that thuds into the pad in front of middle, with the graphic later showing it smashing into the top of leg, is not out if it pitched outside the line of leg stump. The review never reached the projection. It failed at the pitching question, and everything after that was for television.

It also explains why the impact question so often surprises the fielding side. A batter stretching forward to a spinner may be struck outside the line of off stump while playing a genuine shot, and no amount of ball tracking will help, because the law stops there. The forward press against spin is not an accident. It is a technique built around exactly this clause, and it works differently against the various kinds of spin a bowler can impart, because a ball turning in and a ball turning away meet the pad at different points relative to the stumps.

The protocol, and the captain's fifteen seconds

The clock is the part players get wrong most often, and it is unforgiving.

1. 1The umpire gives a decisionOut or not out. There is no option to refer a straightforward lbw or caught-behind decision upstairs, so the umpire must commit to an answer before anybody can review it.
2. 2Fifteen seconds startThe window runs from the moment the ball is dead. The request must come from the fielding captain or from the batter who was given out, and it must be a decision made on the field with no signal from the dressing room.
3. 3The T is signalledBoth forearms crossed above the shoulders. The umpire confirms the review to the third umpire and to the scorers, and play stops.
4. 4The third umpire works the sequenceLegality of the delivery first, then evidence of bat contact, then tracking if the appeal is lbw. Each component is called separately and in order.
5. 5The finding is communicatedThe third umpire reports what the evidence shows, component by component, and states whether it is conclusive. Where it is not, the original decision stands.
6. 6The on-field umpire signals the outcomeThe final signal is made in the middle. If the decision is reversed the umpire withdraws the original one and gives the other, and the review allowance is adjusted according to the finding.

The player review procedure in international cricket. An umpire referral for a run out or a stumping skips the first three steps entirely and costs no team anything.

Fifteen seconds sounds generous and is not. A fielding captain has to gather the opinions of the bowler and the keeper, both of whom are convinced, discount both, judge the impact point from thirty yards away, and commit. A batter has to decide whether what he felt through the handle was bat or thigh pad, without the benefit of any replay at all, while walking. Consulting the dressing room is prohibited, and sides have been warned for glancing at the balcony, because a team with a monitor and a delayed feed would hold information the process is designed to exclude.

The umpire is entitled to refuse a late request, and does. That is the whole enforcement mechanism, and it is why you sometimes see a captain signalling frantically while a bowler is still arguing with him.

Where the third umpire has more latitude than most viewers realise is on legality. Front-foot no balls are checked from the box on every delivery the technology allows, without anybody appealing, and a wicket is not confirmed until that check has been made. This is a quiet structural change to the game rather than a technical one. Before it, the standing umpire had to police the front line and watch the ball's contact with the batter in the same fraction of a second, which is a genuinely unreasonable request. Taking one job away from him improved his performance at the other.

What the review system has actually done to the game

The rules changed. The way the game is played changed more.

Spin bowling got its lbws back. For most of cricket's history, a ball striking a batter on the front pad and appearing to be going on to hit the stumps was routinely turned down, because giving it required an umpire to be confident about a low, slow trajectory over a long distance, and confidence like that was rare. Ball tracking demonstrated, delivery after delivery, that many of those balls were hitting. Umpires now give them. A finger spinner bowling into the pads of a batter pushing forward has a mode of dismissal that was, in practice, mostly unavailable to his predecessors.

Pad play stopped working. The old defensive technique of thrusting the front pad down the line of a turning ball and letting it hit you, secure in the knowledge that no umpire would give it, is now simply a way of getting out. Batters play with the bat closer to the pad, take fewer liberties about which line they can safely leave, and sweep with a clearer understanding that a top edge onto the pad in front of the stumps is a live appeal.

Front-foot discipline improved. When every delivery is checked from the box, a bowler who has been overstepping unnoticed for four overs finds out about it on the first ball, and a wicket taken off an illegal delivery is disallowed with no discussion. The old accommodation, in which a marginal front foot was called only when the umpire happened to be looking, disappeared.

The umpire's authority changed shape rather than vanishing. The obvious prediction, that technology would make standing umpires redundant, did not happen. What happened instead is more interesting. Umpires became more willing to give marginal decisions, because a genuine error can now be corrected rather than living forever, and the old convention of handing the benefit of every doubt to the batter softened. Umpire's call then gives a large category of close decisions straight back to them. The net effect is an umpire whose plain errors are removed and whose fine judgements are protected, which is arguably a stronger position than the one he held before, though it is not the one he is thanked for. Watching how the middle communicates through all this is worth the effort, since the vocabulary of umpire signals now has to carry the review procedure as well.

Reviews became a tactical resource. Captains manage their allowance the way they manage a bowler's overs. Speculative reviews late in an innings, when an unused allowance is about to expire with the innings, are rational and common. Refusing a review a bowler is begging for, because the allowance is needed for the recognised batter still to come, is a real decision made in real matches, and it is one of the few places in cricket where a captain's resource management is visible to the crowd as it happens. None of it shows up in the record. A wicket is a wicket, and the scorecard makes no distinction between a decision given on the field and one produced by a review three minutes later.

Where the system is genuinely fallible

An honest account has to include the failure modes, and there are more of them than the marketing suggests.

Prediction beyond the evidence. The further the ball has to be projected, the wider the real uncertainty, and the graphic does not visibly widen with it. A trajectory drawn on screen has the same confident appearance whether it was extrapolated over a short distance or a long one.

Unusual bounce. The model assumes standard behaviour off the surface. A ball that has already misbehaved once, off a crack or a footmark, is exactly the ball whose future behaviour is least predictable, and it is disproportionately likely to be the one hitting a pad.

Simultaneity in audio. Bat, pad, ground and glove within a few centimetres of each other, at the frame rates involved, can produce evidence no operator can separate. The system's answer in that case is to leave the decision alone, which is correct and still means the wrong batter sometimes walks off.

Insensitivity in thermal. No mark is not proof of no contact. Everybody involved knows this, and the pictures on screen are still persuasive in a way the caveat is not.

Calibration and coverage. The cameras must be positioned, calibrated and unobstructed. A fielder, an arm or a piece of ground equipment in the wrong place removes an angle. Rain, light and the changeover to floodlights all affect what the cameras see.

Uneven availability. The kit is expensive. Full systems appear in high-profile series and major tournaments, partial ones elsewhere, and none at all across a great deal of first-class and domestic cricket. A batter's chance of having a bad decision corrected depends on the commercial value of the fixture he is playing in, which is not a technical problem and is a real one.

The human layer on top. Reviews are requested by people under pressure, on incomplete information, in fifteen seconds. A side can be entirely right about a decision and still lose the argument because the captain hesitated, or because the allowance was spent an hour earlier on a delivery the bowler felt strongly about.

None of that makes the system worse than what preceded it. It makes it a system that reduces error rather than eliminating it, which is what it was designed to do and rather less than it is asked to be.

How to judge a review while it is happening

Six things to watch, roughly in the order the screen will show them.

Establish which review this is. Umpire referral or player review. If the umpire drew the rectangle himself, no team is spending anything and the answer will usually be visible in the pictures.

Check the delivery's legality before anything else. If the front-foot check comes back against the bowler, the rest of the sequence is decoration and the appeal is finished.

On a caught-behind, watch the frame, not the noise. The question is whether a spike or a mark coincides with the ball passing the bat, and whether anything else was in contact at that instant. A loud sound in the wrong frame proves nothing.

On an lbw, look at pitching and impact before the projection. Most failed reviews die at one of those two, and the projection graphic everyone waits for is often irrelevant by the time it appears.

Remember what the on-field decision was. Umpire's call means the original answer survives, so the same graphic produces opposite outcomes depending on a decision made before any of this started. If you have forgotten what the umpire did, you cannot predict the result.

Note how far down the pitch the impact was. A long way forward means a long projection, which means a wider real margin, which means umpire's call is more likely than the confident line on the screen suggests.

Do those six and you will call the outcome correctly more often than the commentary box, which is a low bar cleared more rarely than it should be. The rest of the game's rules and mechanics are collected in the cricket section, and the review system makes far more sense once it is read as what it is: not a machine that knows the answer, but a procedure for deciding when the umpire clearly did not.