Death bowling in T20 cricket explained: the last four overs

Nineteenth over. The batter is set, has faced twenty-something deliveries, knows what the bowler has, and has already decided he is going to hit the next ball into the second tier whether it deserves it or not. Five fielders are on the boundary and every single is free. The bowler has one plan, one back-up plan, and six deliveries.

Death bowling in T20 explained as "bowl yorkers" is the sporting equivalent of explaining surgery as "cut carefully". The instruction is right and useless. What makes the last four overs a distinct trade, coached separately and paid for separately, is not that the deliveries are harder to bowl. It is that the person at the other end has stopped playing the same game.

Everything else follows from that.

The batter's incentive inverts, and that is the whole reason this is a separate skill

A batting innings is a resource-allocation problem with two resources: balls and wickets. In the first over of a T20 innings a batting side has 120 balls and ten wickets, and the wickets are worth a great deal, because losing them early forecloses everything that comes later. A batter in the third over who takes a wild risk is spending an asset the innings will need for another hour.

In the eighteenth over the arithmetic has flipped. There are perhaps a dozen balls left and eight wickets in hand. A wicket in that situation costs almost nothing, because the batter it protects will face two balls at most and the innings ends before the loss can compound. The only thing that has value is the ball itself.

That inversion is not a mood. It is a calculation, and good batters make it explicitly. If a conventional shot has a high chance of producing a single and a slight chance of a boundary, and a premeditated slog has a much better chance of a boundary and a much worse chance of a dismissal, the second option is correct at the death and incorrect at the start. The batter does not become reckless. His definition of a good outcome changes, and with it the entire risk calculus that a bowler has spent his career relying on.

This is why every conventional bowling weapon degrades in the final overs. A bowler beats the bat and the batter does not care. A bowler forces a false shot and the ball goes over the ring for four anyway, because the false shot was played at full swing. A bowler builds pressure with two dot balls and the batter answers with a shot he would never attempt in the seventh over, because the cost of failing has collapsed.

Bowling to a batter who is indifferent to dismissal is a fundamentally different problem, and it has one solution: take away the ability to hit the ball hard, rather than the ability to hit it safely. That is what every death delivery is actually trying to do. Not deceive. Constrain.

17th over concedes 418.7

17th over concedes 917

17th over concedes 1515

17th over concedes 2113

Constructed arithmetic on stated assumptions, not measured data or any real match. The chasing side is assumed to need 60 runs from the last four overs, an even 15 per over, with wickets in hand. The bars show the rate required from the remaining three overs after the 17th over goes for the amount shown.

Show the numbers

Worked example: how one over changes the rest of a chase

Item	Runs per over required from the last three overs
17th over concedes 4	18.7
17th over concedes 9	17
17th over concedes 15	15
17th over concedes 21	13

Read the constructed example in both directions and the pressure becomes legible. A four-run over does not merely save eleven runs against the par figure. It raises the required rate by nearly four an over for the rest of the chase, which pushes the batting side into shot selection it did not want, which is where the wickets come from. A twenty-one-run over does the reverse and effectively ends the contest, because it drops the required rate below what a set batter needs to take any risk at all.

Death bowling is therefore not a marginal skill applied to a decided outcome. It is a compounding one, and the reason a good death over is worth so much more than the runs it saves is that it changes what the batting side is forced to attempt next.

The yorker: the best delivery in cricket, and the one with almost no margin

The pitch is 22 yards, which is 20.12 metres between the bowling creases, and the popping crease is 1.22 metres in front of the stumps at each end. A bowler releasing the ball from somewhere around head height near his own popping crease is aiming at a target zone at the base of the batter's stumps that is, in practical terms, a few centimetres deep.

That is the whole difficulty. The yorker is not a hard ball to conceive and it is not a hard ball to bowl in the nets. It is hard to bowl inside a window measured in centimetres, from twenty metres away, at high speed, at a target that moves.

Why it works is worth stating precisely, because the usual explanation is imprecise. A yorker is not effective because it is difficult to hit. It is effective because it removes the batter's ability to generate power. Power in a cricket shot comes from bat speed through a long arc, and a long arc requires the ball to be far enough from the batter's body that he can swing through it. A ball arriving at the base of the stumps gives him no room to complete a swing, forces him to jam the bat down rather than through, and converts what would have been a six into a squirt to mid-on. The delivery works on geometry, not on deception.

The two ways to miss it are punished in opposite directions, and this asymmetry is the source of most death-over disasters.

Bowl it fractionally short and it becomes a low full toss or a half volley on a length the batter has been waiting his whole innings for. He is already committed to a full swing. The ball arrives in the perfect place for that swing. This is the single most expensive delivery in T20 cricket, and it is not a bad ball in the ordinary sense; it is a good ball that missed by a hand's width.

Bowl it with the batter advancing and it becomes a full toss, and if that full toss passes above the batter's waist height without pitching, Law 41.7 makes it a no ball regardless of intent. In T20 that means a free hit as well, so a single error of judgement about where the batter's feet will be costs a run, a free delivery on which the batter cannot be bowled or caught, and the psychological ground that goes with both.

What actually fails under pressure is mechanical and predictable. The yorker depends on a late release, which depends on the front arm pulling down at the right moment, which depends on a stride length the bowler has grooved over years. Adrenaline lengthens the stride and drops the shoulder. A longer stride and a lower release point both push the ball fuller and flatter, which is why the classic failure mode at the death is not the short ball but the full toss, and why a bowler who has bowled forty perfect yorkers in training will bowl a waist-high one in the twentieth over. His body did something four per cent different and four per cent is the entire margin.

The other thing that has changed, and this is the honest part that the coaching literature is slow to say, is that batters have made the yorker a worse ball than it was. A batter who premeditates a low full toss and pre-moves to a position from which he can scoop or ramp does not need the bowler to miss; he has changed the geometry himself by moving his body before release. The bowler is aiming at where the feet were. The mechanics of executing the delivery itself, and the drills built around it, are worked through separately in the piece on how a yorker is actually bowled. The point for the death over is that the yorker is now a high-variance ball rather than a safe one, and treating it as the default is a choice with a real downside.

The wide yorker and the slower ball: two options, two different bills when they go wrong

A death bowler's second and third choices are not interchangeable, because what each one concedes on a bad day is completely different.

The wide yorker aims full and outside off stump, away from the batter's swing. It works because it makes the batter reach, which shortens his arc in the other direction, and because a ball moving away from the body cannot be hit with the whole of the bat. Against a batter who has walked across his stumps to open up the leg side, it is the punishment for having moved.

Its failure is a wide. In limited-overs cricket the wide is judged strictly, with a guideline marking on the popping crease, and a wide costs a run and a re-bowled delivery. That is not a disaster on its own. What makes it dangerous is that the bowler now has to bowl the ball again with one fewer delivery of margin, knowing that the next attempt will be a fraction straighter out of caution, and a fraction straighter is exactly the width the batter wanted.

The other failure is worse and less discussed. A wide yorker that is not quite wide enough and not quite full enough is a wide half volley, which is width on a hittable length, which is a free four through the off side against any batter with a functioning arm.

The slower ball works on time rather than space. Everything about the delivery has to look identical until release: the run-up, the arm speed, the wrist through the crease. The variation happens in the last fraction of the action, through the fingers coming across the ball for an off-cutter, the wrist turning over for a back-of-the-hand ball, the knuckles pushing the seam for a knuckleball, or the ball held deep in a split grip. What all of them share is that they arrive later than the batter's swing has been timed for.

Its failure comes in three flavours and each has a distinct cost. Telegraphed, through a visible grip change or a shortened arm, and the batter simply waits and hits it further than he could have hit the quick one, because he has more time to select the shot. Too short, and it sits up as a slow long hop, which is the easiest ball in cricket to hit for six. Too full, and it is a slow low full toss, which is the second easiest.

There is a subtler cost that only appears across an over. Every slower ball a bowler shows is information. A batter who has seen the off-cutter twice will pick it the third time, so the delivery is a depleting asset within a single over, and a bowler who leads with his best variation has less left for the balls that matter most. The good ones sequence deliberately: show the least valuable variation early, keep the one the batter has not seen for the nineteenth over.

The other thing a slower ball concedes, which nobody warns young bowlers about, is that it is a gift to a batter who has already decided to hit through the line. A slower ball taken on the full swing goes further than a quick one, because the batter is providing all of the power and none of it is being wasted deflecting pace. A slower ball only works against a batter who is swinging on the bowler's timing. Against one who is swinging on his own, it is a slower ball to hit.

The bouncer at the death, and the rules that decide how much of it you get

The short ball at the death does something none of the other options do: it attacks the batter's assumption rather than his technique. A batter who has set himself for a full delivery has moved his weight forward and lowered his hands. A ball at chest height arrives at a body that has already committed in the wrong direction, and the resulting shot is either a top edge or nothing.

It is also the only death delivery whose availability is capped by regulation, which changes how it is used.

ICC men's T20I playing conditions have limited a bowler to one short-pitched delivery passing above shoulder height per over, with the second called a no ball. Domestic competitions differ: the IPL moved to two per over from the 2024 season, after a trial in Indian domestic T20 cricket, on the argument that the balance between bat and ball at the death had tilted too far. Whether the international game follows is published in the current playing conditions rather than assumed here, and the number does vary by competition, which is one of the small ways in which a franchise league and an international match are genuinely different games. The general framework, including how the umpire judges shoulder and head height and why the limit exists at all, is set out in the article on the short-ball limits and how they are policed.

The cap matters tactically rather than merely legally. A bowler with one bouncer per over holds an option he cannot use twice, which means the threat is worth more than the delivery. Bowling it in the first ball of the over spends the threat immediately. Holding it to the fifth means the batter has to account for it on every ball in between, which is why experienced death bowlers so often do not bowl their bouncer at all and still profit from having one.

The cost of misjudging it is the steepest in the phase. A second short ball above shoulder height is a no ball, and every no ball in T20 brings a free hit. A bowler who has just conceded a boundary, wants to change the picture, and bowls a short ball that gets away from him has converted a bad over into a ruinous one in a single delivery: extra run, free hit, and a batter who now knows exactly which delivery is coming next because the bowler has run out of options.

• 24Balls in the last four overs
• 5Fielders permitted outside the circle
• 4Overs any one bowler may bowl in a T20 innings
• 60Seconds between overs under the stop clock

Counts and limits set by the format and by ICC playing conditions rather than measured data. The bowler limit follows from the rule that no bowler may deliver more than one fifth of the innings.

The field is five deep, four in, and it tells the batter what you are about to bowl

After the powerplay, a maximum of five fielders may be outside the thirty-yard circle, and no more than two may be behind square on the leg side. Take out the bowler and the wicketkeeper and that leaves four fielders inside the circle. Those four are the entire infield, and they cannot simultaneously save the single, cover the inside edge and protect the areas a mishit lands in.

45%36%9%9%

• On the boundary, protecting the areas the plan concedes5
• Inside the circle, saving one and covering the mishit4
• Wicketkeeper1
• Bowler1

A constructed illustration of one common arrangement, not a prescribed field and not a measurement. The counts are constrained by ICC playing conditions: a maximum of five fielders outside the circle after the powerplay, and no more than two behind square on the leg side.

Show the numbers

Worked example: how the eleven are distributed in a death over

Item	Value
On the boundary, protecting the areas the plan concedes	5
Inside the circle, saving one and covering the mishit	4
Wicketkeeper	1
Bowler	1

The first thing to understand about a death field is that it is a set of decisions about which boundaries you will allow. Five fielders cannot cover a full circumference. Somewhere on that boundary there is a gap, and the batter can see it before the ball is bowled.

The second thing is that the field announces the plan. A wide yorker plan requires a deep point and a deep third, because the ball is going that way if the batter connects. A leg-side plan requires a deep square leg and a long leg. A batter looking at the field before the ball is released is reading a partially completed sentence, and good batters read it correctly most of the time. This is why the best death bowlers and captains build fields that cover two plans rather than one, accept a slightly worse fit for each, and preserve the ambiguity. It is also why the field is set before the batter has taken guard rather than after, and why a captain moving a fielder late in an over is often not adjusting the plan but disguising it.

The third thing is the single. In a death over the single is essentially free, because the infield is four men covering the whole circle. A bowler cannot save singles and protect boundaries at the same time, so the conventional wisdom is to concede the single and deny the boundary, on the arithmetic that six singles is a six-run over and one six plus three singles is a nine-run over. That arithmetic is right and incomplete. Six singles also means both batters keep the strike rotating, and if one of them is the dangerous one, six singles has handed him three more deliveries. Which is why the good captains do sometimes plug the single, and why the choice is a matchup decision rather than a rule. The full set of arrangements and the reasoning behind each is worked through in the piece on how death fields are actually built.

The fourth thing is what a fielding restriction penalty does to all of this. Some competitions apply an in-game over-rate sanction that requires an extra fielder inside the circle for the remaining overs of the innings. A side that has bowled slowly through the middle overs therefore arrives at the death with four boundary riders instead of five, which removes an entire boundary option from the plan. This is the most severe punishment in white-ball cricket that has nothing to do with the delivery, and it is levied precisely when it costs the most.

The mental problem: no help, no cover, and failure that everyone can see

There is a specific psychological demand in this phase that has no equivalent elsewhere in cricket, and it is not simply pressure.

A bowler in the seventh over of a T20 innings is part of a plan. There are catchers, there is a partner at the other end applying the same squeeze, and a good ball is rewarded by the structure around him. A bowler in the nineteenth over is alone. Everybody is on the boundary. There is no slip, no short cover, no pressure being built from the other end because the other end is somebody else's problem. Every delivery is a solo decision, executed in isolation, and the field cannot save him from a bad one.

The failure is also uniquely legible. A batting collapse is a collective event and a dropped catch is over in a second. A death over going wrong takes four minutes, happens in front of everyone, and is scored in a column that will be recited on television for the rest of the bowler's career. Bowling figures are a blunt instrument in this phase for exactly that reason: an economy rate in the death overs is not comparable to the same number in the powerplay, and the different things bowling averages, strike rates and economy rates each fail to capture are set out in the piece on what the three bowling numbers actually measure.

The load is heaviest in the middle of an over that has already gone wrong. Two balls in, the over has cost fourteen, and the bowler now has to choose between continuing with a plan that is not working and switching to a plan he has practised less. Both are defensible and both feel like the wrong answer while the crowd is making the noise it makes. What separates a specialist here is not courage. It is having decided in advance, in the calm of the changing room, what the response to that specific situation will be, so that the decision does not have to be taken while the heart rate is at its peak.

There is a quieter demand as well, which is tolerating a good ball being hit for six. It happens. A perfectly executed yorker can be scooped over the keeper's head. A bowler who responds to a well-bowled ball being hit by changing his method has learned the wrong lesson from a single outcome, and doing this repeatedly is the most common way a promising death bowler stops being one. The discipline is to separate the execution from the result, which is easy to say in a coaching manual and very difficult in the nineteenth over of a match that is being watched by a full ground.

Matchups, the left-arm angle, and why the crease is a tactical resource

A captain choosing a bowler for the death is not choosing the best bowler. He is choosing the bowler whose method is least convenient for the specific batter on strike, which is a different question and often produces a different answer.

The most reliable structural matchup is angle. A left-arm seamer bowling over the wicket to a right-handed batter is delivering the ball from outside the batter's off stump and slanting it across him. That does three things at the death. It makes the wide yorker the natural extension of the angle rather than a deliberate deviation, so it is easier to execute and harder to read. It takes the ball away from the leg side, which is where most premeditated death hitting is aimed. And it means the batter's favourite area requires him to hit against the angle, which is the hardest way to generate power.

The same bowler to a left-handed batter is a completely different and much friendlier proposition, which is why a batting side that has a left-hander and a right-hander at the crease in the final overs has already won a small tactical argument. This is a large part of why lower-order pairs are constructed the way they are, and the mechanics of that angle in both directions are worked through in the article on what the left-arm angle actually does.

The crease itself is an underused resource, and the measurements make the point. The return crease is 1.32 metres either side of the middle stump, so a bowler may legally release the ball anywhere across a corridor more than two and a half metres wide. Moving from tight to the stumps to wide on the crease changes the release point by over a metre, and over twenty metres of pitch that is a materially different line of approach to the same target. A bowler who uses the crease has doubled his apparent variation without changing anything about his delivery, and the batter has to re-read the angle every ball. Going around the wicket does the same thing more dramatically and costs the bowler the lbw angle in return.

Matchups also run the other way, and a good batting side plans for them. If a captain has one left-arm seamer and holds him for the nineteenth over, the batting side's job is to make sure the right-hander is not on strike for it, and a well-drilled pair will spend the eighteenth over manufacturing exactly that. The final overs of a close T20 match contain an entire negotiation about who faces whom that is invisible unless you are watching the strike rotation rather than the runs.

Saving overs for the death, and the hole it leaves in the middle

Every bowler in a T20 innings may bowl a maximum of four overs, which follows from the rule that no bowler may deliver more than one fifth of the innings. Twenty overs must therefore be covered by at least five bowlers, and a captain has to allocate them across four distinct phases with the same limited stock.

Holding overs back for the death is not free, and the cost is precisely calculable.

Take a constructed example. A side has two genuine death specialists and three other bowling options. If the captain wants both specialists to bowl two overs each in the last four, then across the first sixteen overs those two bowlers can supply only four overs between them, and the other twelve must come from the three remaining options. That means the middle overs, seven through sixteen, are being bowled almost entirely by the bowlers the captain trusts least, against batters who are set, on a pitch that is at its easiest.

A batting side that has done its homework knows this. Its plan is to survive the powerplay, attack the sixth-choice bowler in overs eleven to fifteen, and arrive at the death with wickets in hand and a lower required rate than the fielding captain wanted. Holding your best bowlers back is therefore an invitation, and it is a perfectly reasonable one to accept if the alternative is losing the last four overs outright.

The opposite policy has a matching failure. A captain who uses his best bowlers when the game is live, in the middle overs, and leaves the death to whoever is left, will often find the match already decided by the sixteenth over. There is no clean answer. The trade is between losing the middle and losing the end, and different captains resolve it differently, which is one of the few genuinely open tactical questions left in the format. The broader problem of allocating a fixed stock of overs across an innings sits inside the wider subject of how a captain actually makes decisions.

Two practical constraints tighten it further. A bowler who has been held back for four overs and is then asked to bowl the seventeenth and nineteenth is bowling cold, in front of a batter who is not cold. And a bowler whose two remaining overs are both at the death is a known quantity: the batting side can see the same arithmetic and knows precisely who is coming.

A death over, ball by ball

1. 1Before the over: the two plansThe bowler and captain agree a primary delivery and a fallback, and set a field that fits both imperfectly rather than one perfectly. A field that fits one plan exactly has told the batter which plan it is.
2. 2Read the batter's set-upWhere is he standing, has he moved across his stumps, is he opening the leg side, is he deep in his crease or advancing? The answer changes which delivery is safe, not just which is best.
3. 3Ball one: establish, do not surpriseThe first ball is usually the plan rather than the variation. It sets a reference the later variations will be measured against, and a variation bowled first has nothing to vary from.
4. 4Ball two: read the responseIf ball one produced a premeditated movement, the batter has shown his hand for the over. If it produced a conventional shot, he is still watching the ball and the variation is worth less.
5. 5Ball three: the first variationShow the least valuable of the available variations, not the best one. Information spent now is information unavailable in the last two deliveries, which are the ones that decide the over.
6. 6The over goes wrongTwo boundaries in three balls. The prepared answer matters more than the correct one, because the decision is now being taken with a peak heart rate in front of a loud ground.
7. 7Damage control, or the resetEither bowl the plan again on the reasoning that it was well executed and unluckily treated, or move to the fallback. Switching to a third option that has not been rehearsed is how a fourteen-run over becomes a twenty-two-run over.
8. 8Ball five: the withheld optionThe bouncer, the wide yorker or the slowest ball, whichever the batter has not seen. This is the delivery the whole over has been protecting, and it is worth most when the batter has run out of information.
9. 9Ball six: the batter is committedWith one ball left the batter will almost always attempt maximum value, which makes his movement predictable even when his shot is not. Bowl to where he is going rather than where he is.
10. 10Between overs: sixty secondsThe stop clock gives the fielding side sixty seconds to be ready for the next over. Two warnings, then five penalty runs per breach, which is the entire time available to move fielders and change the plan.

The decision sequence a coached death bowler and captain run through. Ball-by-ball choices are illustrative of the reasoning rather than a prescription, and every good bowler's sequence differs.

What separates a specialist from a good bowler bowling late

Most bowlers can bowl a yorker. Very few can be relied on to bowl one in the twentieth over of a match that is level, and the difference is not the delivery.

Repeatability under fatigue beats range. The instinct is to collect variations, and a bowler with six of them who executes each one seventy per cent of the time is worse than a bowler with three he executes ninety per cent of the time. Variations are only valuable if they land where they were meant to; a mis-executed knuckleball is a long hop. The specialists tend to have fewer options than people assume and much better control of them.

A genuine second-choice ball. Every death bowler's primary delivery will be read at some point in a season. What matters is whether the fallback is a real weapon or a hopeful one. A bowler whose plan B is "try the yorker again but harder" does not have a plan B.

Bowling the same ball twice. Having been hit for six off a well-executed delivery, can the bowler execute it again immediately? This is the rarest quality in the discipline and the most predictive one. It requires a bowler to trust a process over an outcome at the precise moment when the outcome is loudest.

Awareness of the field as part of the delivery. A specialist is not bowling at the stumps, he is bowling at a field. Knowing that the gap is at deep midwicket changes which side of the yorker to miss on, and a bowler who does not know where his fielders are is bowling a good ball into the wrong half of the ground.

Accepting the boring over. A six-run over with no wicket is an excellent outcome and produces no highlight. Bowlers who need the wicket to feel successful take risks that a specialist declines, and the risk-taker's average over is worse even though his best over is better.

There is a cost side to all of this that the game has been slow to acknowledge. The action that produces a repeatable yorker at high pace is a demanding one, and bowlers who specialise in it tend to accumulate a particular kind of workload in a particular set of joints. The role is also concentrated: a side that has one death specialist will use him at the death in every match, in every competition, for a season that now runs almost year round. Scarcity plus concentration is a reliable formula for breakdown, and it is why the market for these bowlers in franchise cricket, including at the IPL auction, prices them well above what their raw figures suggest.

What has changed recently, and it favours the bowler for once

Three regulatory changes in the last few seasons bear directly on this phase, and they do not all point the same way.

The most significant for the death bowler is the white-ball wide-ball trial the ICC began in October 2025 for an initial six-month period. Under it, the reference point for judging a wide is the position of the batter's legs at the moment of delivery, even if the batter subsequently moves across towards the off side, and on the leg side a ball passing the popping crease between leg stump and the protected area marker is not a wide, with that marker line extended to the popping crease as a guide for the umpire. Both changes address the same complaint: a batter who moves before release was effectively able to manufacture wides out of good deliveries, which punished the bowler for the batter's movement. Whether the trial has been made permanent is a matter for the current playing conditions rather than an assumption, but the direction of travel is clear, and it makes the wide yorker a materially safer option than it was.

The stop clock is the second. Made a permanent fixture in ODIs and T20Is, it gives the fielding side sixty seconds between overs, with two warnings and then five penalty runs per subsequent breach. For the death overs, where field changes are constant and conversations between captain and bowler are where the plans actually get made, this is a real constraint. The tactical thinking has been pushed backwards into the innings break and the drinks break, because there is no longer time to do it between overs.

The third is the in-game over-rate penalty applied in some competitions, which brings an extra fielder inside the circle for the remainder of the innings when the fielding side falls behind schedule. It is the harshest sanction in the format, because it hits at the exact point where an outfielder is worth most.

Running underneath all three is a change in the batting side rather than the rules. Domestic competitions that allow a side to introduce a substitute batter deepen the batting order, which means the batter facing the nineteenth over is more likely to be a genuine hitter than a tail-ender. The death bowler's job has become harder for a reason that has nothing to do with bowling. And the equipment and boundary dimensions have moved in the batter's favour over a long period, which is the background against which the two-bouncer trials should be read: they are attempts to give something back.

What to watch in the last four overs

Four things are visible on any broadcast and none of them are mentioned in commentary, which tends to describe what the ball did rather than what was intended.

Look at the field before the ball, not after it. Count the boundary riders and find the gap. That gap is the bowler's plan expressed in fielders, and it tells you which delivery is coming more reliably than the bowler's grip does. If the field covers two plans imperfectly rather than one perfectly, the captain is a good one.

Watch the batter's feet at the moment of release, not the shot. Whether he has pre-moved, and in which direction, decides everything about whether the delivery that follows was good or lucky. A yorker bowled at a batter who has already backed away was aimed at a target that had gone.

Note which variation the bowler shows first. Leading with the best variation is a sign of a bowler under pressure. Holding it back until the fifth ball is a sign of one who has a plan for the whole over rather than for the next delivery.

Count the overs each bowler has left going into the seventeenth. The whole endgame is contained in that number. A captain with two overs of his best bowler in hand is running a different match from one who has four overs of his fifth choice, and you can know which before a ball of the death is bowled.

Watch those four across a few matches and the last four overs stop looking like chaos and start looking like what they are: a constrained problem with a small number of legal answers, being solved at speed by somebody who knows he will be judged on the outcome rather than the choice. The rest of the format is covered in the wider cricket section, but this is the phase where the gap between a good plan and a good result is widest, and where the two are confused most often.