Dead Runs – Responding to the Critics

Peter Kettle | 12:09am BST 21 June 2024

The concept of “Dead Runs” – those deemed to have no real value to a team in seeking to win a match – and its application was introduced in my book on Don Bradman’s scoring (published in 2019).[i]This endeavoured to put Test batting averages of the various eras on an equal footing and thereby produce a merit ordering of prominent players.The results of this “standardisation” exercise have been refined and updated in two recent ACS journal articles (The Cricket Statistician, February and May 2024). 

In standardising the raw averages, Dead Runs is one of four factors taken into account, along with use of dominance ratings, long-run improvement in batting expertise and career length. This factor has been especially important in the case of Bradman (assessed at around 10% of his total career runs), and ranging between 2.0% and 5.0% of total runs scored for three-quarters of the other 172 leading Test batsmen selected across the eras.[ii]

The innovation has come in for scepticism and criticism from a number of participants in a CW Forum in November 2023 on my standardisation exercise and also some correspondence in the ACS journal. After recapping on the concept and how it is applied, the intent is to summarise the critics’ points, discuss their validity, and consider if my treatment of Dead Runs needs to be improved on or perhaps even be abandoned. It is thought that the debate may be of general interest to those who engage (in some way or other) with Cricket Web.

There are, of course, many interesting ways in which batting performance has been evaluated – incorporating factors such as strike rate, consistency of scoring, proportion of fifties and centuries made as well as contributions to team wins. Yet the traditional “runs scored per completed innings” retains its interest as a readily accessible single, and frequently quoted, statistic. 

Identifying Dead Runs

The considerations applied in pronouncing runs to be dead:

• The focus is on those runs that are scored after a stage is reached when the opposition is considered to have only a remote possibility of winning the contest; and so the team in question is deemed to have only a remote possibility of losing, with a victory or perhaps a draw being the only realistic outcomes for itself.
  
  
• I have sought to make a “remote possibility” operational by applying a likelihood of the opposition winning (by a slender margin) of around 1 in 20, based on the observed capabilities of their players. When matters reach this stage, Dead Runs then set in, making no material enhancement to the prospect of securing victory.
  
  
• I contend that when the odds of being defeated fall to around this level, the captain cannot reasonably be accused of behaving in an unduly risky or cavalier way, at least in normal circumstances.
  
  
• (It might be rational to declare well before that stage is reached if, by then, a draw is most likely – ie risk losing in order to try to force a win, depending on the position reached in the series. But this complication doesn’t affect the rest of this discussion.)

Runs of negligible worth, in my sense, occur due to the captain being excessively cautious, to the point of being unsound or irrational (such as Bill Woodfull’s marked tendency during Bradman’s time), or the captain has simply let matters drift without sufficient attention.

Why Discarded

Dead Runs are wholly discounted – eliminated – in standardising batting averages and arriving at an indicative merit ordering, because:

• Not only are they irrelevant to the team’s cause, which is the rationale for all scoring. They can be, and often are, counter-productive as their accumulation makes it increasingly possible that the opposition will be able escape with a draw as remaining playing time shrinks and maybe bad weather lurks.
  
  
• One contention concerns the case of timeless Tests (to be played to a finish) – routinely staged in Australia prior to WW2 and elsewhere as an occasional series tie-breaker – which is that excessive runs in the confines of a specific match actually had a positive value in the context of the series as a whole as they served to exhaust and demoralise the opposition. But this point has little force as these matches were generally well spaced, with typical intervals between them of ten to twenty days. Dead Runs were then simply without any purpose. In the case of Bradman’s six timeless matches in which he struck Dead Runs, only one exceeded four days (extending to five and a half days); so there were no prolonged exertions.[iii]
  
  
• Whilst it is the captain who lets Dead Runs accumulate, they are fully deducted from the totals of those who do the scoring. It is true that from the batsman’s perspective, with the psychological pressure to perform well being lifted, further runs become that much easier to make; and the opposition bowlers by this time are likely to be tiring and perhaps losing their focus. However, these runs are deducted from their scores on the “no help to the team” premise.

Probing the Criticisms

It is widely agreed that runs far beyond their need do occur in some Test matches, and not only when batting third and setting a target. Striking examples are Australia’s first innings compilation of 674 against India at Adelaide in January 1948 (4^th Test), 380 being the visitors’ highest total in the entire series; and Australia’s 469 in reply to South Africa’s first innings of 153 at Sydney in December 1931, the visitors arriving with only two high quality batsmen.

(i) A fundamental point made at the CW Forum is that, in constructing a merit rating/ordering, whether or not to exclude Dead Runs is matter of personal opinion, and that individual tastes will differ on this matter. I fully agree, though without explicitly acknowledging this until now, and elaborate by saying that what ought to constitute, or qualify, as “merit” in a role is not something that can be proved or shown to be correct.

Similarly, personal views of this kind are embedded in various, widely accepted, batsmen (and bowler) rating schemes – though often claiming to be wholly objective because, once set, only hard verifiable data are fed in to produce the finding. Yet what factors are to be included, and excluded, and the relative importance to be assigned to them are matters of personal opinion of an individual or panel – as for instance with Wisden and ICC ratings of batsmen’s innings. This is unavoidable – inherent in any merit rating scheme.

In Wisden’s case, it has determined the best individual innings in Test matches, with a ranking of the top 100 men, each being accorded as a merit rating denoted by a specified number of points – maximum possible being 300 (released in July 2001).

Its scheme has 12 indicators of merit such as – in addition to number of runs scored – proportion of team runs scored, pitch quality, strength of opposition bowling, pressure faced  at the start and end of his innings, wickets falling during his stay, support received from colleagues during his innings and support given to tail-enders. Unfortunately, we are not told about their relative significance and the resulting trade-offs. For instance, is poor pitch quality deemed to be as significant as strong bowling faced, or rated as twice as important or what? Does a good batting pitch offset strong bowling? And so on. But whatever are the trade-offs actually applied, there will be plenty of scope for differing personal opinions among knowledgeable people about what they ought to be, implying a different merit ordering of the very same innings.

Steve Ferrier, a well regarded researcher, has constructed his own greatest 100 Test innings using only 8 indicators, three of which differ from those of Wisden (Cricket Web, September 2014). So there’s clearly room for debate also on what factors are worthy of inclusion and don’t measure essentially the same thing.

(ii) The Forum also produced some opposition to completely discounting Dead Runs. This would fit the unconstrained perspective of a batsman’s demonstrated ability, recognising that the pressure is off once the opposition cannot win the match (cut a third or a half of these runs?). Also, perhaps, factor in that as Dead Runs start to be accumulated things tend to become dull for spectators – the contest itself petering-out into a one-way street with a potential dead end.

(iii) The Forum discussion rightly emphasised that the task of determining the scale of Dead Runs struck in any given match is certainly a difficult one, and inevitably involves a good deal of judgement. Yet Test captains frequently have to weigh the odds of the opposition being able to meet potential targets as the match progresses – if only intuitively. They also have to decide what odds are acceptable. I try to mimic the captain in this respect, with the 1 in 20 odds mentioned earlier, and judge at what stage of the team’s innings these have applied. This judgement is informed by:

• • • examining the recent form of the opposition players in the current series, or prior matches if at the start of the series,
    • knowledge of highest fourth innings chases in general when a stiff target is presented,
    • specifying an upper and lower range for the number of Dead Runs struck and taking the mid-point as the best estimate (with a panel of assessors, taking the average of the individuals’ best estimates).

In this way, I try to minimise the problem of getting the resulting estimate of Dead Runs accepted. Even so, Martin Chandler (for one) struggles with how such runs could ever be quantified satisfactorily – some runs are unarguably dead, but then there are areas that get greyer and greyer.

A test of this proposition could be to have a panel of 3 or 5 “knowledgeable” individuals – armed with information just touched on – make estimates of Dead Runs for a series of specified cases and then see if the divergence between their estimates is small or large. “Large” would indicate the task is too fraught with uncertainty to be acceptable. For material cases (Dead Runs of 20 or more in an innings), a large difference could be denoted by a spread around the average finding of at least 60%.

Looking externally, I note that personal judgements also enters decisions taken, for example, with Wisden’s and the ICC’s batsmen ratings about how best to measure factors such as bowling strength faced (should runs per dismissal, economy rate and strike rate all enter?), and pitch quality (rely on runs scored/wickets taken during a match or amount of seam movement and turn extracted?).

(iv) In commenting on an earlier draft, Martin Chandler was conscious of the fact that some runs are worth more than others, citing Graham Gooch’s (undefeated) 154 second innings runs against West Indies (Leeds, 1991) as being more valuable than his mammoth 333 against India (Lord’s, 1990). He is inclined to the view that, in an overall sense, conventional batting averages smooth these variations in value to the team in as good a way as any single indicator can. Don Bradman, though, is regarded as a special case – to be looked into.

Perhaps what might be termed Bradman’s “decisive contributions” to Test wins could be weighed against his “unarguably” Dead Runs on a one-to-one basis. I have in mind identifying those matches that Australia would not have won without his overall contribution where he was also the team’s highest scorer (both innings added if Australia batted twice). In these cases, there would be a shortfall between the combined runs made by all other members of the team and the total that was just sufficient to produce the win. This shortfall would then count as his decisive contribution. If this is thought of as too harsh, the contrasting alternative would be to take his own total after deducting both the average score made by the other specialist batsmen in the team and his “unarguably” Dead Runs.

Where the win is by a certain number of wickets and Bradman was the team’s highest scorer, one could take the excess of his runs per innings over those of the other batting specialists.  

(v) Some commentators express a dislike of mixing subjective factors (individual perception and interpretation) with analysis of a purely statistical nature, sometimes expressed as “individual opinion in the midst of otherwise statistical analysis” – though the reason for this position isn’t clear in what I’ve read.

In contrast, I find this mixing of different animals to be acceptable practice so long as the subjective factors are thought significant for a rounded evaluation of performance. I believe that relevance should be the over-riding consideration. 

(vi) In a letter published in The Cricket Statistician journal (May 2024 Issue), Russell Houldin asserts that runs scored which are of little or no value to a team can be identified only after the match has been concluded, whereas I start counting these runs as a match unfolds. His reason being that “during an actual match one never knows what the outcome will be or the corresponding value of scoring runs at a particular time.”

Deducing if, and how many, Dead Runs have been accumulated from examining a completed scorecard sounds straightforward. Yet this is deceptive. A golfing tale provides a way into the problem. When Bryson DeChambeau struck the “bunker shot of my life” onto the 18^th green in the recent US Open to make victory over Rory McIlroy a formality, instead of blaming the Gods, the moral was simply to accept it could have gone either way; and next time to strive to be ahead by a substantial safety margin before the closing stages.

Transferring this to cricket, it is the “safety margin” that gives rise to a problem. How is the required margin to be established? The aim is to be sufficiently well placed during the later stages of an opposition’s fourth innings (or one’s own chasing innings) to provide insurance against potential bad luck – or potential lack of good luck. If a victory is by a single run, Dead Runs are nil by definition. What about a Test match win by 11 runs or 25 runs – both seem too close for comfort, too small to entertain Dead Runs being scored. But what about a winning margin of 60 runs: after a high scoring match, this might not be regarded as comfortably safe yet after a low scoring match it might be regarded as ample. 

In looking back after winning by 60 runs, and thinking about whether this happens to be greater than a required safety margin – and so contains some Dead Runs – one is driven to pose “what if…” questions about events during play. That is, to speculate about particular events which could easily have turned out differently and favoured the opposition.

How many more runs would the opposition have scored if, say, their in-form number six hadn’t run himself out with a foolish call, or their number seven hadn’t played his only rash shot to be caught when well set, or the umpire hadn’t ruled in favour of a dubious catch to make them eight wickets down. The likely scoring consequences if any one of these specific possibilities had materialised is not susceptible to estimation and is all the more uncertain when all three hypotheticals are taken together. Even informed guesses are fraught with difficulty as subsequent events could turn out in many alternative and unpredictable ways. This in turn means that the required safety margin is indeterminate, even after the match is finished and won. Hence whether any Dead Runs have been scored – and if so, how many – is also inherently indeterminate.

Nor can any excess of runs beyond some specified safety margin be attributed to particular batsmen in the order. A win by 90 runs with a safety margin of 40 yields 50 Dead Runs. As there is no specified stage when Dead Runs start being struck, only an end of match reckoning, these 50 runs can be trimmed off batsmen’s scores anywhere in the order. There will be a large number of possible combinations that exhaust the sum of 50 (10 or 15 runs off an opener who has made 30; 20 or 25 runs off the number four who has made 45…). One could spread it across all batsmen on a neutral – equal proportional – basis. But this is hardly satisfactory.

With my approach, the focus is on what would have been a sound decision on a declaration in real time, ie as the match unfolds. The judgements taken on this can, of course, sometimes turn out to be incorrect after the event; or a loss eventuates because the odds of 1 in 20 carried some, albeit very low, risk. On checking the dead runs estimated for Bradman, plus those of his partners, which occurred in eight Test matches, these were far outweighed by the eventual winning margins. These margins were: more than a team innings in four cases, 10 wickets, 560 runs and 360 runs. One match was drawn due to rain saving England on the final day.

Conclusions

• Runs struck seemingly beyond their need should not necessarily be excluded from a batsman’s merit rating, though I happen to feel they should be.
  
  
• This preference is held despite a lot of judgement being involved, which is also present at a fundamental level in other
  widely accepted rating schemes, most notably those of Wisden and the ICC.
  
  
• The onus should be on the assessor to provide a reasoned explanation of the quantum of estimated Dead Runs. Unless this is done, it is unlikely that the result will be accepted other than by just a few.
  
  
• A test of whether Dead Runs can be established satisfactorily could be to examine the findings of the individuals comprising a panel of, say, 3 or 5 assessors. A small dispersion of findings would confirm this could be achieved, whilst a large dispersion would be to the contrary.
  
  
• Establishing Dead Runs after a match is completed is considered unsatisfactory as it is, in logic, an indeterminate exercise.

Please visit this thread on the CricketWeb Forum where you can join the community discussion surrounding this article

---

 

End Notes

[i] Rescuing Don Bradman from Splendid Isolation, published by PK Associates, February 2019.

[ii] Based on a comprehensive analysis for 25 of the 172 batsmen.

[iii] Occasionally, even timeless Tests resulted in an abandoned draw. As with the final match of West Indies versus England in April 1930, to resolve the series: after seven days play and two rained-off, England’s booked voyage home intervened! Similarly, the final match of South Africa versus England in March 1939: this deal breaker came to an indeterminate ending after twelve days (three lost to poor weather) so as to enable the visitors to catch their booked ship home.