Nightwatchman in cricket explained, and whether it works

Five twenty in the afternoon on the second day. Four overs to be bowled, the light is going the way it always goes in an English September, and the number three has just edged one to second slip. The number four is padded up. He has been padded up for two hours. And somebody in the dressing room says the thing that has been said in cricket dressing rooms for over a century: send the bowler.

That is the nightwatchman in cricket explained in one sentence, and the sentence hides everything interesting about it. A side that has spent the whole day trying to keep its best batters at the crease now deliberately sends its worst one out to face the hardest few overs available. It is one of the very few tactics in any sport that puts a weaker performer into a harder situation on purpose, and after more than a hundred years of use nobody has established beyond argument that it is a good idea.

The people who defend it are not stupid. Neither are the ones who have abandoned it. They are disagreeing about a number that cannot be measured directly, and the disagreement is worth understanding properly, because once you can see the trade you can watch the last half hour of a day's play as a decision rather than as a formality.

What a nightwatchman actually is, and who gets the job

A nightwatchman is a lower-order batter promoted up the order when a wicket falls close to stumps, so that the specialist who was next in does not have to begin an innings in the final overs of a day.

Nothing about it is a formal position. The batting order in cricket is not fixed and does not have to be declared. A captain may send anybody in at any fall of a wicket, may change the order between wickets, and owes the opposition no explanation. So a nightwatchman is not a rule, nor an exception to one. It is a batting order chosen for a particular ten minutes. The only law that bears on it at all is timed out, which requires the incoming batter to be ready to receive the next delivery within the permitted period, and which is the reason a side that dithers over whether to send the nightwatchman is dithering against a clock.

The job usually goes to a bowler who bats somewhere between eight and ten, and the selection criteria within that group are more specific than they look.

He needs a defensive technique that holds up. A tail-ender who plays every ball with hard hands and a bat coming down from the direction of third slip is no use, because the point is survival, not runs.

He should not be the bowler you most need fresh the following morning. Forty minutes of concentration at the end of a long day in the field, followed by an early start with the bat, is a real physical cost, and it lands on a player whose actual job begins when the innings ends.

He should be somebody who does not mind. This sounds soft and is not. A bowler sent in as a nightwatchman is being told, in front of his teammates, that his wicket is worth less than somebody else's, and asked to walk out and prove it. Some players take that badly enough that captains stop asking them.

• 4Overs left in the day
• 24Deliveries left to survive
• 3Wickets already down
• 5Specialist batters still to come

Invented figures, used throughout this article to show the shape of the comparison. Nothing here is a measured or reported number.

The nightwatchman in cricket explained as a trade, not a favour

The mistake almost every explanation makes is to describe the tactic as protection, as though the specialist is being shielded from harm at no cost. He is not being shielded. Somebody is being sent to take the hit for him, and that somebody is a wicket.

Set it out as the exchange it actually is.

Without a nightwatchman, the side risks the specialist being dismissed in the remaining overs. Call that the primary risk. It is the risk the captain is trying to remove.

With a nightwatchman, the side accepts three things instead. A high chance that the promoted bowler is dismissed, tonight or early tomorrow, which costs a wicket the side would otherwise still hold. A meaningful chance that he is dismissed quickly enough that the specialist has to come in anyway, in which case the side has paid the price and received nothing. And a set of consequences the following morning that are almost never priced in at all.

Those three are the whole argument. The first is obvious, the second is what makes the tactic look foolish when it fails, and the third is where the interesting analysis lives.

Notice what the trade is not. It is not a comparison of how good the two batters are. Everybody agrees the specialist is the better player. The claim is about marginal value: that a wicket taken from the eighth batter costs the innings less than a wicket taken from the fourth, and that the difference is large enough to pay for the higher chance of losing the eighth.

That claim is defensible. Whether it is true on any given evening depends almost entirely on the two variables the standard explanations skip over, which are how many overs are left and how many wickets are in hand.

Why the last half hour is the most hostile part of a batting day

The argument for the nightwatchman rests on the assertion that the final overs are unusually dangerous. If they are no more dangerous than the middle of an afternoon session, the whole tactic collapses, because you would just be sending a worse batter in for no reason.

There are four separate mechanisms, and they stack.

A new innings is the most dangerous part of any innings. This is the one everybody underrates, and it is the load-bearing fact of the entire debate. A batter's chance of dismissal per delivery is not constant. It is highest in the first few overs, before the eyes have adjusted to the pace off the pitch, before the feet are moving, and while the bowler still has a field set for exactly this moment. A specialist walking out at 5.20pm is not merely facing four difficult overs. He is facing the four most difficult overs of his own innings, and if he survives them he will resume the next morning at something close to the vulnerability of a new batter all over again.

The light. In a Test played without floodlights, the last hour is often the worst light of the day, and a batter who cannot pick the seam early is playing a different game to the one he played at two o'clock. The playing conditions give umpires discretion to take the players off, but the standard is safety and playability, not comfort, so a great deal of cricket is played in light that is legal and horrible. What the umpires do when they intervene is a small procedure with a large effect on the last hour, because it also determines how many overs are actually left to survive.

Fatigue. A side batting late on day two has been on the field for most of two days. Concentration is a finite resource and it is at its lowest at the end of a session that has already contained two others.

The bowler is fresh and knows it. A captain with four overs left and a new batter walking out will bring back his quickest bowler for a short, hostile burst, because that bowler can empty himself knowing the day ends shortly. A specialist arriving at that moment gets the most aggressive four overs the fielding side is capable of producing.

Put those together and the last half hour is genuinely a spike in risk rather than an ordinary passage of play. That is the strongest part of the case for the nightwatchman, and it is a good case.

The case against: you are spending a wicket to save a wicket

Now the other side, which has been winning the argument in recent years.

The first objection is the simplest. A wicket is the currency of a Test innings. A side batting with seven wickets in hand and a session to bat has a range of options that a side with five wickets in hand does not, including scoring quickly, batting through, and eventually declaring at a moment of its own choosing. Spending one of them on a bowler who blocked twelve balls and edged the thirteenth is not a neutral act. It is a purchase, and the thing purchased is the removal of one specialist's exposure to about four overs.

The second objection is sharper, and it turns the tactic's own logic against it. If new batters are unusually vulnerable, which is the premise of the whole tactic, then sending a new batter into the most hostile period of the day is precisely the wrong response. The nightwatchman is not a hardened survivor being asked to do a hard job. He is a poor batter at his most fragile moment, in the worst conditions of the day, against the fastest bowling the opposition has left.

The third objection is the one nobody thinks about until it happens, and it is the reason the tactic can be worse than doing nothing. Send the nightwatchman with four overs left, and if he is dismissed in the second over the specialist walks out anyway. He now faces the same hostile conditions, against a bowler with his tail up who has just taken a wicket, with the field crowding him, and the side is one wicket worse off than if it had never bothered. That outcome is not rare. It is a substantial fraction of all nightwatchman innings, and every time it happens the wicket has bought nothing whatsoever.

The fourth objection is about the man at the other end. A specialist who is set, who has faced sixty balls and is seeing it well, is the most valuable asset the batting side owns. Send a nightwatchman in and that set batter spends the last four overs mostly at the non-striker's end, watching. Then the day ends and his innings pauses anyway. The tactic protects one batter by half-wasting another.

2 overs left12

4 overs left24

6 overs left36

8 overs left48

Arithmetic only, six deliveries per over. The nightwatchman will face some fraction of these rather than all of them, depending on how the strike rotates.

Show the numbers

Deliveries a nightwatchman is being asked to survive

Item	Deliveries remaining
2 overs left	12
4 overs left	24
6 overs left	36
8 overs left	48

How many overs are left changes the answer more than anything else

Almost every argument about the nightwatchman is conducted as though it were one decision. It is not. It is a family of decisions that share a name, and the number of overs remaining moves the answer from clearly right to clearly wrong.

With two overs left, the case is at its strongest. Twelve deliveries, of which the promoted batter might face seven or eight, is a small enough exposure that even a poor player has a decent chance of getting through, and the specialist is spared a genuinely nasty ten minutes. The cost of failure is lowest here too, because if the nightwatchman goes in the first over only a handful of deliveries remain for the specialist to survive.

With eight overs left, the case has inverted. Forty-eight deliveries is close to a third of a session's exposure for a tail-ender. The chance he survives it is poor, the chance he is dismissed early enough to send the specialist in anyway is high, and the specialist would in any case have had time to get himself in, which is exactly the thing that reduces his risk. Sending a nightwatchman with that long left is not protection. It is an early surrender that costs a wicket.

Somewhere between those two the answer flips, and where it flips is a matter of judgement rather than calculation. Captains who use the tactic tend to work with a private threshold, expressed in overs rather than minutes, and to be rigid about it once they have chosen one. The rigidity is deliberate. A decision taken at speed, by a group of tired people, in the two minutes the timed out law allows, is better made by a rule agreed in advance than by an argument on a balcony.

Here is a constructed worked example to show why the threshold exists rather than what it should be. Every figure in it is invented.

Suppose a side judges that its number four has a one in ten chance of being dismissed in the twenty-four deliveries left, and that the nightwatchman has a one in three chance in the same period. Suppose it also judges that losing the number four early costs the innings forty runs of expected total, while losing the nightwatchman costs fifteen.

Without the nightwatchman, the expected cost of the last four overs is one tenth of forty, which is four runs.

With the nightwatchman, the expected cost is one third of fifteen, which is five runs, plus a further amount for all the cases where he is dismissed early and the number four has to bat anyway. Even before that second term is added, the tactic is behind.

Now halve the exposure to two overs. The number four's risk drops to something like one in twenty and the nightwatchman's to one in six. The first option costs two runs, the second costs two and a half, and the gap has narrowed to almost nothing while the variance has fallen. Halve it again and the two are indistinguishable.

That is the shape of it, and it holds regardless of what the true numbers are. The nightwatchman's disadvantage scales with exposure, because he is worse per ball, so the fewer balls remain the less the difference matters. The tactic is a short-exposure instrument that was historically used at medium exposure.

Wickets in hand decide whether you can afford the trade

The second variable gets almost no attention and probably deserves the most.

At three wickets down, a wicket is a cheap thing to spend. Seven remain, five of them belonging to specialists, and the innings has plenty of structure left. The nightwatchman's dismissal removes one of the two or three wickets the side was never going to use anyway.

At six wickets down, everything changes, and it changes in both directions at once. The wicket is now expensive, because the innings is nearly over and each remaining partnership matters more. But the side also has fewer candidates: promoting the number nine means the number ten and the number eleven are all that stand behind the remaining specialists, and if the nightwatchman fails, the tail is exposed with two sessions still to bat.

There is a specific trap that catches sides at five or six down. Promote a nightwatchman there and you have not protected a specialist so much as reordered the tail. The batter you spared may be a number six who is himself not much better than the man you sent, in which case you have bought almost nothing at full price. The tactic is only worth its cost while the gap between the protected batter and the promoted one is large, and that gap narrows every time a wicket falls.

The same logic explains why the tactic never appears in the fourth innings of a close chase. A side needing runs against the clock cannot afford a batter whose instruction is to score none, and the arithmetic of pacing a chase makes a blocking tail-ender actively harmful rather than merely expensive. Protection is a first-innings idea.

What the nightwatchman is actually told to do

The instructions are more specific than "block", and they contain a contradiction the player has to resolve on his own in the middle.

He is told to survive. That is the whole point, and it means leaving anything he can leave, playing as late as he can, and treating a maiden over as a success rather than a failure. A nightwatchman who is beaten four times in an over and scores nothing has had an excellent over.

He is told, in most sides, to take the strike where he safely can, because leaving the set batter to face the quick bowler defeats the purpose of the exercise. This is where the contradiction lives. Taking strike late in an over means running singles, running singles late in the day against a fielding side throwing at the stumps is a way of being run out, and a run out in this situation is the worst available outcome because it can take either batter. Some captains solve it by telling the nightwatchman to run nothing after the fourth ball of an over. Others tell him to run everything and accept the risk. Both are defensible and they are not the same instruction.

He is told what to do about the last ball of the day, which is a small thing that matters. A nightwatchman on strike for the final delivery will be on strike the next morning, which is what the side wants if it intends him to keep going, and exactly what it does not want if it intends to reverse the order at the first opportunity.

And he is told, quietly, that nobody will blame him. This is not sentiment. A tail-ender who believes his place is on the line if he gets out will play a nervous, half-forward, edge-prone innings, which is the innings the tactic is designed to avoid.

1. 1The wicket falls and the clock is checkedThe only number that matters first is how many overs remain, including whether the umpires are likely to take the players off for light and whether the extra half hour is in play.
2. 2Is the exposure short enough to be worth buying?A handful of overs makes the tactic cheap, because a poor batter's disadvantage scales with the number of deliveries he has to face. A long remainder makes it expensive, which is why most sides fix a threshold in overs before the match rather than argue about it on the balcony.
3. 3Can the side afford to spend a wicket?Early in an innings a wicket is cheap and the gap between the protected batter and the promoted one is wide. Deep in an innings both of those reverse, and the promotion starts to look like exposing the tail rather than shielding the top.
4. 4Is there a suitable body?The candidate needs a defensive technique, a reasonable temperament, and no bowling responsibility so urgent the next morning that forty minutes of batting tonight becomes a bad trade.
5. 5The instruction is given, and it is specificSurvive, leave what can be left, take the strike where it is safe to take it, and be told in advance what to do about running singles late in an over and about the final delivery of the day.
6. 6The incoming batter has to be ready in timeThe timed out law is already running from the moment the wicket fell, which is why a side that has not settled this before the day started is settling it badly.

The sequence a dressing room runs through in the two minutes the timed out law allows. It is a decision about time and wickets, not about ability.

The morning after is the part nobody plans for

Everything up to stumps is the easy half of the decision. The awkward half arrives at eleven the next morning, and it is where the tactic quietly gives back much of what it gained.

Suppose the nightwatchman survives. The side now begins a fresh day with a set specialist at one end and a bowler at the other, against an attack that has slept, on a pitch that is often at its most helpful in the first hour. The specialist the tactic was designed to protect is still at the non-striker's end, still not facing, and the very conditions the promotion was meant to shield him from are being negotiated by the man least equipped to negotiate them.

The batting side has to choose between two bad options, and the choice is rarely made explicitly.

It can tell the nightwatchman to keep blocking. He then occupies the crease for another hour, denies the set batter the strike, scores at close to nothing, and burns overs from the most dangerous period of the day. The side survives the hour and has spent a great deal of scoring time doing it. In a first innings where the plan involves batting long, that may be acceptable. In an innings that needs a rate, it is a slow disaster.

It can tell the nightwatchman to play his shots and get on with it, which is really telling him to get out in a useful way, and which is the honest modern position. The problem is that a tail-ender playing shots against a hard ball is very likely to get out immediately, at which point the specialist walks in to face exactly the conditions the entire manoeuvre was designed to avoid, one wicket down on the deal, with the added indignity of having spent the previous evening padded up for nothing.

There is a third possibility that sides occasionally use and almost nobody explains. The nightwatchman can be retired out, which the Laws permit without anybody's consent provided the batting side accepts that the innings is recorded as a dismissal, or he can simply be left in and instructed to bat as a genuine number eight would. The first is legal, wastes the wicket formally rather than informally, and causes a fuss every time. The second is the quiet answer that good sides have been drifting towards.

1. 1Play resumes with a bowler at the creaseA hard ball, refreshed seamers, and often the most helpful hour of the day for bowling. The set batter is at the other end and may not even be on strike.
2. 2The side has to decide what the nightwatchman is now forNothing about the previous evening's instruction carries over automatically. Survival was the plan for four overs. It is a completely different plan for a session.
3. 3If he keeps blockingThe dangerous hour is survived, but the set batter is starved of strike, the scoring rate collapses, and the side spends its best batting conditions of the day defending.
4. 4If he plays normallyHe is likely to be dismissed within a few overs, which brings the specialist in to face precisely the conditions the promotion was meant to avoid, with one fewer wicket in hand than before.
5. 5If he keeps going and starts scoringThe batting order behind him is compressed, the genuine number four may not bat until after lunch, and the bowler occupying the crease is spending energy he will need later with the ball.
6. 6Either way the wicket has been spentThe only question left is what the side received for it, and the honest answer is usually a quieter twenty minutes on the previous evening.

The sequence when a nightwatchman survives to stumps. Each branch carries a cost, which is why the decision is better taken the night before than at the ground.

When the nightwatchman survives and starts scoring, the problem changes shape

Every so often the tactic produces the outcome nobody planned for. The nightwatchman does not get out. He gets in.

A handful of nightwatchmen have made Test hundreds, and one has made a double century, which remains the highest score anybody has made in the role. Those innings are treasured precisely because they are absurd, and they create a genuine tactical headache that the celebration tends to obscure.

The batting side now has a problem of order. Its number four has been sitting with his pads on since the previous afternoon. Its number five is behind him. If the nightwatchman bats for two hours, the specialists are all pushed back by two hours, and a side that intended its top order to bat on a first-day pitch finds them batting on a second-afternoon one instead. The pitch does not wait for the batting order, and the way a batting order distributes risk across an innings assumes the order will roughly hold.

There is a workload problem too. The bowler at the crease is accumulating fatigue in a way that shows up two days later. A seamer who bats for ninety minutes on the morning of day three, then has to bowl twenty overs across days four and five on a flat surface, has been asked to do two jobs by a decision that was framed as a small one.

And there is an odd psychological effect that sides genuinely worry about. A nightwatchman who scores fifty has, in the dressing room's collective imagination, been promoted. Doing it again is now expected of him, which is a great deal of pressure to put on somebody who was chosen for the role because it was thought his wicket did not matter.

None of which is a reason not to enjoy it. A tail-ender defending his stumps in fading light and still being there at tea the next day is one of the better things the format produces, and the fact that it is tactically inconvenient is part of the joke.

Why modern sides use the nightwatchman in cricket less than they used to

The tactic has not disappeared, but its frequency has clearly fallen, and four separate changes explain it.

Lower orders bat better than they used to. For most of the twentieth century a number nine was a genuine passenger with the bat. Professional batting coaching now reaches the whole side, and a modern tail contains players who can defend properly and score at a rate. That cuts the tactic at both ends: the gap between the protected batter and the promoted one has narrowed, and the side is more willing to let its actual number eight bat when the time comes.

Scoring rates have risen, so wickets are worth more relative to time. In an era where a side expects to score at four an over rather than under three, the binding constraint on an innings is wickets rather than overs, and anything that spends a wicket to buy time is running against the grain of the format. That change in tempo is one of the broader shifts separating the modern game from the one the tactic was invented for, and it shows up across the differences between the three international formats.

The aggressive alternative arrived. Rather than send a blocker, some sides now send a hitter with instructions to attack whatever comes. The reasoning is that a wicket is being spent either way, so it may as well be spent trying to score, unsettle the field and take the bowlers out of their rhythm. This is usually called sending a nighthawk. It carries the same wicket cost, a much shorter expected occupation of the crease, and none of the following-morning problem, because the player almost always resolves the question himself before lunch.

Captains stopped treating it as automatic. For a long time the nightwatchman was the default and not sending one required a reason. The default has flipped in a lot of dressing rooms, and sending one now requires a reason. The reasons that survive are specific: a very short remainder, a pitch doing a great deal, a batter who is genuinely unwell or has been on the field for a long time, and above all the twilight period in a day-night Test.

That last one is the strongest surviving case. Under lights with a pink ball, twilight is not a mild inconvenience, it is the single most hostile passage in the match, and everything about how a pink ball behaves under floodlights argues for keeping specialists away from it. The nightwatchman survives best in exactly the situation that most resembles the one it was invented for.

What the modelling says, and what it cannot see

There is real mathematics on this, and it is worth reporting honestly, including the parts that weaken it.

The published treatments work in roughly the same way. Take a batter's historical scoring rate and his average number of balls faced per innings. From the second of those you can derive a per-ball probability of dismissal, treat each delivery as an independent trial, and compute the expected number of runs the innings produces under two orders: with the nightwatchman promoted, and without. Tim Trudgian's treatment of the problem, published in The Conversation alongside a paper, reaches the conclusion that the expected total is always lower with a nightwatchman, because the better batter both survives more deliveries and scores more from the ones he survives. Statistical work of the other kind, comparing actual totals in innings where the tactic was used against those where it was not, has pointed the same way.

The models are clean and the conclusion is not obviously wrong. But three assumptions inside them can reasonably be attacked, and it is worth being precise about which.

Independence. Treating each delivery as an independent trial with a fixed dismissal probability is what makes the arithmetic tractable, and it is the assumption most clearly at odds with how batting works. Batters are far more likely to be dismissed in their opening overs than later, which means the risk is front-loaded rather than spread evenly. That front-loading is the exact effect the nightwatchman exists to manage, and a model that averages it away has assumed the problem out of existence. Treatments that do allow a higher early hazard soften the result without reversing it.

Conditions. The models use career-long averages, computed across all conditions, most of them better than the last four overs of a day in poor light against a fast bowler in a short burst. If the final overs really are more hostile than average, both batters' true dismissal probabilities are higher than their career rates suggest, and the gap between them may widen or narrow in ways the model cannot see.

What a run is worth. Expected runs is not the only objective. A side batting first on day two is often trying to maximise something closer to the probability of reaching a total that makes the match safe, and protecting the top order against a catastrophic collapse can be worth more than a few expected runs. Risk-averse play is not irrational simply because it lowers the mean.

The honest summary is that the burden of proof has shifted. The tactic was the default for a century without ever being demonstrated to work, the formal analysis that exists points against it, and the counter-arguments are plausible but have not been quantified. That is a fair description of a genuinely open question, and it is very different from saying the nightwatchman has been debunked.

How to judge the call live

The next time a wicket falls late in a day and a bowler walks out, you can assess the decision in about ten seconds, before anybody on commentary has finished explaining what a nightwatchman is.

Count the overs, not the minutes. Two or three overs left is a defensible promotion. Six or more is not, and the reason is that the tail-ender's disadvantage grows with every delivery he has to face while the specialist's advantage does not.

Check the score for wickets, not runs. Three or four down means the wicket is cheap and the gap between the two batters is wide, which is when the trade works. Six down means the side is reordering its tail and calling it protection.

Look at who walks out. A fourth seamer who bats sensibly is a proper nightwatchman. The frontline spinner who has to bowl thirty overs on days four and five is a decision somebody will regret on Sunday.

Watch the strike in the last two overs. If the nightwatchman is farming the strike and the set batter is standing at the non-striker's end for four consecutive deliveries, the tactic is working exactly as intended. If the set batter ends up facing the quick bowler anyway, the side has paid for something it is not receiving.

Come back at eleven the next morning and check what he was told. A nightwatchman who blocks for an hour is being used as a shield. A nightwatchman who plays a drive at the third ball of the day is being used as a gate. Both are legitimate, and knowing which one you are watching tells you more about the batting side's plan for the day than anything the captain will say afterwards.

Get into the habit of asking those five questions and the last half hour of a Test day stops being an anticlimax. It becomes the one passage of a five-day match where a captain has to decide something under a two-minute clock, with a stadium watching, on the basis of an argument that has never been settled. The rest of the format's tactical furniture, from the follow-on to the second new ball, is collected in the cricket section, and almost all of it is more comfortably resolved than this.