James Nixon |
Russell Houldin has set out the basics of a framework and methodology for comparing the merit of individual players in two fairly recent articles in the ACS journal, The Cricket Statistician.
The ACS is a UK based organisation, titled The Association of Cricket Statisticians and Historians. Its total membership is, I gather, around 800-900 and extends to several countries.
He emphasises the novelty of it, integrating a player’s batting and bowling capabilities. The merit of specialist batsmen and specialist bowlers are compared with each other and also with all-rounders – though Houldin has found it somewhat problematic to adequately define those who should qualify in this dual role. And, looking ahead, hopefully incorporating their fielding/wicket-keeping capabilities as well.
His two articles feature in The Cricket Statistician journal: Issue 204, November 2023, pages 35-38; and Issue 207, August 2024, pages 35-39.
Headed, respectively: A Statistical Method for Comparing Cricketers; and Towards a Complete Theory for Cricketer Performance Comparisons.
While the fundamentals are in place, it is billed as an unfinished, evolving, piece of work and the latter article includes discussion of potential refinements and some thoughts on how to give value to fielding and wicket-keeping roles.
These articles are a mixture of large claims and modesty. To quote Houldin himself: “As far as I know, there is no accepted system for comparing the value of all cricketers that uses a common basis of the net run contributions of players. Using indices is rife with arbitrary judgements. The approach put forward here (August 2024) and previously (November 2023) I believe provides a simple way of launching into discussion.”
There is a terminological matter to address at the outset, as the ACS is keen for its members to use the term “batter” rather than “batsman.” I resist going along with this, much preferring the traditional “batsman”, to which I have added “batswoman” to my lexicon.
“Batter” – which is in vogue for those indoctrinated in the creeds of political correctness and gender-neutral terminology – finds support in the ACS guidelines for authors. Yet, while “batsman” is permissible, though not encouraged, the term “batswoman” is ruled out of court!
To quote from the ACS Style Guide (as last revised in May 2021):
“batsman, batter – either may be used, at editors’ discretion; but not batswoman”
My Aunt Dorothy, who passed away in April 2003 at Brentwood (Essex), would doubtless have called the person who decided on this “a silly arse,” this being a favourite comment of hers when well deserved.
And it perhaps goes to show how skin deep the ACS stance is on matters of equality and inclusion: merely, I surmise, an unthinking and half-baked nod to current fashion.
Moreover, I find use of the term “batter” to be more than faintly ridiculous. Because, for one thing, “batter” is redolent of fish and chips and, for another, it conjures up (for me, at least) a vision of the bat being applied to give the opposition bowlers a continual battering…sending them into oblivion and, perhaps, to their eventual death!
Also, by seeming a little longer to pronounce than “batsman”, the term “batswoman” tends to give a woman with bat in hand somewhat more weight…more gravitas…and so perhaps more importance than a counterpart man with bat in hand.
I think the term “batwoman” is a good alternative to “batswoman”, having a parallel with “Batman” who is associated with comic book heroes and is often called the “Caped Crusader” or the “Dark Knight”.
Either of these two epithets are in tune with the feminist movement that has swept, or is still sweeping, most of the world! This escapes the thinking of the ACS style guide authors. So high time for another revision to it!
For those readers having an interest with gender as it relates to cricketing terminology or, alternatively, feel it is an irrelevant distraction, I draw their attention to a recent piece by Megan Maurice: A Batsman by Any Other Name: Why Language Matters in Sport.
Posted on the internet site Substack, under Obstructing the Field, February 2025.
Summary of Houldin’s Approach to Rating Players
To summarise: Russell Houldin assesses the career performance of a bowler – or batsman – with reference to how superior, or inferior, they are in revealed ability relative to a benchmark for a relevant era. His benchmarks represent an estimate of average performance in a specified era, using conventional batting and bowling performance stats.
Batsmen are credited, or debited, with runs scored above or below the benchmark for their era; and bowlers likewise, though also taking account of the proportion of wickets they capture (assuming, for working purposes, that nine wickets are claimed by bowlers per innings). Hence, for a bowler, net runs conceded, or saved, per wicket captured are derived whilst also factoring in their wicket taking frequency (reflecting their strike rate).
Its Attraction
In a nutshell, this is a calculation of runs scored and runs conceded per innings after being standardised, for a given era, with reference to typical batting and bowling performance of those who participate – at whatever level of play is of interest. Accordingly, Houldin derives the value of a player to his team – the focus of attention – as being his average net runs contributed per innings. Hence the associated mantra: “A Run is a Run is a Run.”
This way of looking at a player’s worth has a basic and strong appeal – for, at end of the day, success at cricket is all about scoring more runs than the opposition. As he notes: “It is true, of course, that wickets need to be taken, but a team that takes more wickets yet scores fewer runs loses the match…capturing wickets is a secondary {instrumental} objective.”
This summary could fit neatly on a packet of breakfast cereal and so be easy to remember. And Albert Einstein would doubtless approve of its economy! For those unaware of his catch-phrase: “Explain things as simply as possible, but not simpler.”
Einstein: A Life, by Denis Brian, 1996.
So far, so good…in principle.
Note, in passing, that Houldin would have preferred to compare each player with the next best player of his era, in terms of runs created (for batsmen) or saved (for bowlers), although he has employed “collective” averages (ie those of a whole group of players) as an approximation to doing that. (August 2024 article, page 36)
On the same page of that article, Houldin also hankers after basing his comparisons on a convention that is applied by analysts of baseball players’ performance. “Ideally” he would compare each cricketer with “an equivalent” player from his era, “following the methodology of Major League Baseball.” By this term, it transpires that Houldin is referring to a specialist batsmen or specialist bowler, or an all-rounder, who is on the verge of being skilled/experienced enough to be of the same grade (or level) as the player being assessed.
In the baseball analytics literature, such a player is referred to as a “replacement-level player”: “one who is not a bench player, but rather a theoretical filler player such as a readily available Triple-A player. A team composed entirely of replacement-level players would win only about 45–50 games in a 162 game season” {ie around only 30% of the matches}. Hence these replacement players are not of the same standard as that of the players being assessed.
This notion has a similarity with a proposal put forward, a decade and a half ago, by Nicholas Rohde of Griffith University (Queensland). It is to base the value of specialist batsman (or bowler) on his performance in relation to counterparts in the other teams of the same competition, during his own playing time.
The underlying logic is as follows. The selectors’ decision to include any particular specialist batsman (or bowler) in a team means denying that place to an external counterpart, someone who could (in principle) have been secured from another team within the same competition. That alternative has been foregone, representing what economists term the “opportunity cost” involved. Hence, to justify his place in the team, the player in question must at least equal the contribution to be expected from a potential external replacement, which is denoted by the typical score per innings, or wicket taking performance, of such a counterpart player. Anything less than this benchmark level of scoring and the batsman (or bowler) becomes a liability to the team rather than an asset.
Rohde neatly combines the “intensity of performance” (output of net runs) and “longevity of career” into a single index to arrive at his 40 highest performing all time Test batsmen. See, his article, An Economic Ranking of Batters in Test Cricket. Economic Papers, Vol 30, December 2011, pp. 455–65.
Houldin’s ideal is, in effect, a watered down version of Rohde’s proposal, although he might not see it quite that way.
The Main Weakness
Turning to the chief drawback of Houldin’s approach – that is, in the form in which he has actually applied it – this constitutes a fatal one, it seems to me. It surfaces in a number of ways. Take the case of assessing a specialist batsman who doesn’t bowl (or so rarely it amounts to the same thing) – eg Crawley and Duckett in Tests: nil deliveries for each of them. Poor at bowling, presumably, otherwise they would have been put on before now. The only behavioural evidence to go on is their very limited spells in county matches: respectively, 11 overs of off spin for no wickets while conceding 33 runs, and 25 overs of off spin for 2 wickets at 49.5 runs apiece.
I can hear you, the reader, saying: neither of them has conceded a single run in a bowling role in Tests – one can’t beat that! Though, I emphasise: nor has either of them taken a single wicket.
The question therefore arises: is it correct, or even legitimate, to treat this non-functioning pair of bowlers in Tests on a par with a neutrally rated specialist or all-rounder bowler, so as to round them out as “a whole cricketer” ? (Leaving aside fielding matters, about which Houldin has, so far, had only preliminary thoughts – albeit some interesting ones.)
There is no compelling logic for such a treatment, in my view. This is demonstrated by Kapil Dev who gets a rating for his Test bowling of zero in Houldin’s latest table of findings – being rated neutrally for his runs per wicket and also for the frequency of his strike. Hence Dev is on a par with the representative “grand average” bowler for his era, who acts as the benchmark.
Such a convention, implicitly, equates Crawley/Duckett’s hypothetical bowling performance at Test level to that actually achieved by Dev who amassed 434 wickets at 29.6 runs apiece, striking at 1.9 wickets per innings – pretty handy! Yet there is no way of corroborating such an equality. Nor is there any valid way of grading Crawley/Duckett as Test bowlers.
Houldin’s latest table also has two other mainstream bowlers of modern times who emerge with a close to neutral performance value: Jacques Kallis at minus 3 (previously, minus 2.34) and Ian Botham at plus 4 (previously, plus 5.38). So we can say that as a duo these two are rated approximately neutral – and assume for illustration they get a net points score of 0. While Don Bradman – who isn’t treated as a bowler (though he occasionally got on) – stands on his batting alone. Bradman is credited with a net contribution of 64 runs per completed innings relative to his benchmark, which is an average of 36, and has a void for his bowling. In effect, Bradman is given a zero score for what bowling he actually did in Tests. Ipso facto, Bradman is implicitly treated as if his Test match bowling is on a par with the roughly “neutrally” valued Kallis/Botham duo.
It hardly needs be said that both Kallis and Botham were rather effective as bowlers for their respective Test teams, whereas Bradman rarely troubled the scorers as a bowler and could be thought of as a non-bowler for working purposes. Hence, equating them as bowlers is far-fetched, to say the least.
Likewise, Geoffrey Boycott gets a void for his bowling, though he was also an occasion contributor. I can visualise Boycs responding in this manner:
“I certainly don’t mind being put roughly on a par with Botham and Kallis for my crafty medium pacers: seven scalps I got with them in twenty innings, and with fair economy rate, just 2.4 runs an over – not bad eh? But if I came across Beefy in my local pub I’d have to buy him at least half a dozen pints before he’d stop ribbing me!”
In the end, Boycs surrendering: “Okay, Beefy, it’s oohtuhlee ridikyuhluhs, pon meh wud tis so. Boot abowt t’battin, nah ‘ere’s t’thing..…”
I have another illustration of this dead end. Houldin assesses Sobers’ bowling at 7 points below (ie inferior to) the grand average performance for specialist bowlers for his era, with 34 runs conceded per wicket taken against the chosen benchmark of 30, as well as a below average strike frequency. These negative points are debited against him when comparing his overall merit with Bradman (The Don), who gets a void for his Test bowling as he did little of it. The Don’s void equating to zero when the tallying is done, and so he ends up way ahead of Sobers in the final reckoning – with fully three times Sobers’ net merit points (64 versus 21).
Who, in their right mind – or even someone borderline bonkers – would consider The Don to be a better bowler than Sobers, or that Sobers wasn’t a positive asset to his side as a bowler – claiming 235 wickets at 34.0 runs apiece and striking at 1.48 victims per innings, delivering in his three different left arm styles of bowling: fast-medium, SLA orthodox and wrist spin. The Don, by comparison in earlier times, taking just 2 wickets in 9 innings at 36.0 runs apiece when sending down his leg spin mixed with an occasional googly.
(By the way, as to The Don’s two victims, one was a West Indian batting at number 8 at the Adelaide Test in December 1930: wicket-keeper Ivan Barrow, LBW to Bradman for 27. Although Barrow had a lowly test career batting average of 16.2 from his 19 innings, he became the first West Indian to score a Test century in England – making 105, opening the innings and putting on 100 runs with George Headley for the second wicket at Old Trafford in July 1933, when age 22 and 6 months. Bradman’s other scalp was England’s illustrious Wally Hammond who swatted at a full toss, missed it and got bowled when on 85. This occurring during the last over of the day’s play in the third Test of January 1933, also at Adelaide, with England then well on their way to a handsome victory and an eventual 4-1 series win.)
To sum up: Houldin’s approach, when applied to the kind of comparisons outlined above, could find a home in the Theatre of the Absurd. The stage plays of which being exemplified by that cricketer of first-class status and Nobel Prize winner in literature (awarded in October1969), Samuel Beckett.
Samuel Beckett (April 1906 – December 1989) was a left-handed opening batsman (possessing, in his own words, a gritty defence) and a medium pace bowler. He played two matches of first-class status for Dublin University against Northants, doing so in 1925 and 1926 – though he didn’t greatly trouble the scores on those occasions.
And the cricket nut cum lauded playwright Harold Pinter would no doubt have been eager to write the script and put on a production in London’s West End. The pity, of course, is that neither of them are still with us.
Implication
For the reasoning already set out, I consider that Houldin’s framework and associated method should be confined to three types of comparison of players’ relative merit, for a specified era:
- Comparing all specialist batsmen
- Comparing all specialist bowlers
- Comparing all genuine all-rounders: with degrees of freedom as to how they are defined.
In my view, his approach is potentially suitable for these three sorts of comparisons – that is, when finalised in its details (as are some rather different approaches to the same task).
One appealing, and accepted, alternative way of equating batting and bowling performances is that invented by the Tasmanian, Ric Finlay (a former maths teacher living in Hobart). This one is based on comparable levels of performance in batting and bowling roles, as given by their relative frequencies of occurrence in practice. In this way, identifying “equivalence frequencies” to determine performance similarities between batting and bowling milestones.
When this limitation, or confinement, is recognised, the looked-for holistic nature of the method – aiming to portray the “complete cricketer” – disintegrates! Apart from the genuine all-rounders, of which there seem to be relatively few, all other players are, necessarily, dismembered (in a figurative sense) for purposes of analysis and merit rating. Like the accident that befell Humpty Dumpty, Mr Houldin won’t be able to stick all the pieces back together again!
The nursery rhyme Humpty Dumpty describes a character who falls from a wall and cannot be repaired by the “king’s horses and men,” symbolizing the irrevocability of certain actions, losses, or accidents. The tale emphasises the futility of effort when trying to fix a broken situation.
The Latest Incarnation
At a recent gathering (via zoom) of ACS members, taking place at the end of March, Houldin, produced a revised merit rating table for Test players (as mentioned above) which embraces 101 players (rather than the initial 51). His talk having the title: “Evaluating the Whole Cricketer”.
The revised table of findings is a four-tiered creation – each tier, or division, being shown in a different colour shading: 25 players occupying the First Division, 28 in the Second Division, 22 in the Third, and 26 in the Fourth Division. Although each player gets an individual rating (again, a net points score), at this stage of its evolution Houldin doesn’t attach a lot importance to the differences in ratings within each of the divisions.
The boundaries to the divisions seem to be drawn to give rough numerical balance, as there aren’t decisive breaks between the individual players’ points scores. (A score of 20 points is bottom of the class in Division One, with the top of Division Two showing one player on 19 points followed by a clutch of them (five to be exact) on 18, and even more (seven) on 17 points. Bottom of that class gets 15 point. There being four players who head the next Division on 14 points, which goes down to 11 points. Four players head Division Four, all on 10 points.)
Poor Ted Dexter (who passed away in August 2021, at age 86) comes bottom of the whole shooting match, standing alone on minus 4 points. This implies he was something of a liability to England and not really worth his place, given the availability of other candidates for his position. This same applies to Bob Simpson and Jeff Thomson of Australia and to England’s Frank Woolley! (All ending up with negative net scores.)
Some Anomalies Surface
There appear to be a number of peculiar placings when certain players are singled out. For example:
– Of the Genuine All-Rounders: Hadlee, Khan and Davidson are in the top division, while Kapil Dev, Flintoff and Botham are in the bottom one. (Russell Houdin is unlikely, I feel, to be getting a Christmas Card from Beefy this year!)
– Of the Spinners: Ashwin is in the top division, while Laker and Grimmett are in the second one, and Underwood is placed in the bottom division (each being rated in bot batting and the bowling roles).
– Of the Fast Bowlers: Malcolm Marshall features in the top division, while Shaun Pollock and Lindwall – both very capable with the bat – feature in the second division.
– Of the Specialist Batsmen: Barrington is placed in the top division while Ponting, Tendulkar, Greg Chappell, Gavaskar and Viv Richards (all getting voids for the bowling role) are placed in the second division.
A Potential Refinement
A suggestion that I made in the February 2025 issue of the ACS journal was directed to Houldin’s quandary about how to identify “genuine” all-rounders, and where to draw the line for qualifying batting and bowling performances. This, inevitably, involves some subjective judgements.
Houldin ends up in dismay, his “general view being that all-rounder is not a useful concept. There are no cricketers who could be listed as a great bowler and a great batsman.” Later posing the question: Was Alan Davidson an all-rounder or a great bowler who was also a useful bat?” (Davidson averaged 20.5 runs per wicket with the ball and 24.6 runs per completed innings with bat.)
For the post-WW1 period, Houldin has chosen to use the long-run batting average of 30 as a minimum qualification for an all-rounder, combined with being more than just an occasional bowler.
What I suggested, to put an end to his dilemma, is that all-rounders should be defined either as those with better than the “reference average” performance for both bowlers and for specialist batters of their era, or – preferably and considerably more extensively – by setting a level of attainment within a specified distance below each of these two relevant “reference averages” – ie somewhat less demanding. Setting the reasonable upper and lower bounds could be based on the informed opinion of a selected panel. The precise levels for the all-rounder thresholds could be set as the mid-points of limits that are deemed reasonable. Each all-rounder would then get two (additive) net run contributions.
Other Potential Refinements & a Suggested Widening of Scope
- Specification of Eras
Houldin appears to work with only three eras: 1877-1889 (13 years), 1890-1914 (25 years) and 1920-2025 (106 years). Hence, after WW1 there is no differentiation at all for a period extending for more than a century. This era is given an overall batting average and an overall bowling average of 30 for the specialists and nominated all-rounders, and for all the other batsmen an average of 20.
That seems acceptable as a starting point, as the players’ collective averages haven’t, in the main, varied a lot from decade to decade – though varying by enough to warrant a refinement going forward. Each post-WW1 era would then be assigned its own overall batting average and bowling average plus strike frequency.
Such a disaggregation would bring out, more fully, the true relative merit of individuals as playing rules and regulations, the quality and mix of types of bowling, and the general quality of pitches played on have altered over time.
The researcher Geoff Dickson has led the way in his regard with a co-authored paper of 1998.
Geoff Dickson being a lecturer at Central Queensland University at Rockhampton from 1996-2004, then lecturing at NZ’s Auckland University of Technology from 2004-18, before taking up a position at La Trobe University’s Business School in Melbourne.
His co-authored paper is titled, A cricketer for the ages: Changes to the performance variation of Test cricket batting from 1877-1997.
He divides Test cricket into six eras: 1877-1914, 1920-39, 1946-66, 1967-79, the 1980s and the 1990s. To which can now be added: 2000-09 and 2010-25.
Each specified era has a defensible rationale. This is sometimes self-evident, such as the sizeable breaks for the two World Wars. The break made in the mid-1960s signals what is generally thought of as the beginning of modern times Test cricket. The 1980s and 1990s each underwent rapid change, including the influence of Packer’s World Series Cricket from 1977-79 and the effect of ODI matches that began in 1979. And the new millennium has witnessed a raft of new shots, spawned by T20 league matches.
In my own treatment, in cases where a batsman overlaps two (or more) of Dickson’s eras, his performance is established for each era in turn before combining them according to the proportion of his total innings in each.
See Chapter II (pages 3-42) of my book, Rescuing Don Bradman from Splendid Isolation, published in 2019 (80 pages).
In Houldin’s treatment, when a player’s Test career overlaps two adjacent eras then (as far as I can tell) he sometimes – though not always – pro rates the benchmark average to reflect the proportions of years or matches spent in each of them.
2. Houldin’s Selection Criteria
Adopting a minimum of 25 Tests for a player’s inclusion rules out too many stars to be satisfy many cricket enthusiasts. Having seen how a portion of them fare on his ratings, their appetite is likely to be whetted for further comparisons of highly prominent players. Houldin admits to omitting, among others: Larwood, Jessop, Tyson, Jardine and Colin Bland. A minimum of 20 Tests would be far more suitable and 17 would be highly desirable (netting, among others, Frank Tyson).
Adopting 48 runs per completed innings as the minimum admissible average for specialist batsmen is considerably too tough – omitting, Houldin acknowledges, Alastair Cook, Kevin Pietersen, Clive Lloyd, Arthur Morris and Lindsay Hassett. Rohan Kanhai just scrapes in with his average of 47.53.
Lowering the cut-off to a batting average of 44 would include all of those just mentioned, plus: Gordon Greenidge with 44.7 and Richie Richardson with 44.4. Also, Dean Jones (46.6) and Norman O’Neill (45.6); David Gower (44.3), Peter May (46.8), Colin Cowdrey (44.1), Tom Graveney (44.4) and John Edrich (43.5). In addition: Aravinda de Sliva (43.0), Hanif Mohammad (44.0); Virat Kohli (46.9), VVS Laxman (46.0), Mohammad Azharuddin (45.0), and Martin Crowe (45.4).
Adopting 25 runs per wicket as the maximum admissible average for specialist bowlers is felt to be substantially too low. The stellar Shane Warne only just enters with his average of 25.4.
This imposed cut-off rules out the following payers, as Houldin admits: Jimmy Anderson, Maurice Tate, Stuart Broad and Wes Hall; also Andy Roberts (25.6) and Roy Gilchrist (26.7) among the fast men. Omitted also are the following spinners: England’s Derek Underwood (25.8) and Graeme Swann (30.0); Australia’s Richie Benaud (27.0) and Stuart MacGill (29.0); India’s Bishan Bedi (28.7), Erapalli Prasanna (30.4) and Bhagwath Chandrasekhar (29.7); of the West Indies: Lance Gibbs (29), Sonny Ramadhin (29.0) and Alf Valentine (30.3); and of South Africa, Hugh Tayfield (25.9).
A maximum of 33 runs conceded per wicket would be suitable – netting many other high calibre spinners, such as Abdul Qadir (leg breaks/googlies) who kept wrist spin alive during his 1977-90 Test stint (13 hauls of 7 or more wickets in a match) – averaging 32.8 runs per wicket – before handing on the baton to Shane Warne; and Fred Titmus (off spin) averaging 32.2 runs per wicket during a Test career that spanned two decades, from 1955-75, finishing at the age of 42.
A maximum of 34 runs per wicket would also capture Garry Sobers.
Houldin refers to his cut-offs as being “controversial.” At this juncture, I’m inclined to bring tennis legend, John McEnroe into the conversation!
John McEnroe’s famous cry, “You cannot be serious” would be heard loud and clear down the telephone line.
In summary: Houldin’s present qualifying thresholds are too tough for full value. They should be relaxed in order to capture considerably more players who are of great interest to cricket enthusiasts. This worthwhile refinement will, inevitably lead to a call for quite a lot more work. Nil sine labore; labor omnia vincit.
I am indebted to the vaunted cricket writer Gideon Haigh for having the audacity to plonk down a couple of Latin phrases on the page without translation – ones that I’m sure few readers will have the foggiest idea of what is meant! That’s our Gideon for you (proudly proclaiming, “I didn’t go to university”) and residing in inner Melbourne (fairly near to me).
A Philosophical Matter
What is essentially a philosophical matter arises when Houldin emphasises that statistics are not the be-all-and-end-all for discussion about players relative merit. Rather, it provides a suitable starting point. That sounds fine and sensible. Yet there’s always the urge to refine…and then do more refining…until one reaches the stage where the “discussion” focusses on relatively minor matters and the methodology heavily dominates. Moreover, the discussion usually doesn’t lead anywhere conclusive. It tends to put people into sparring (or warring) camps.
Accordingly, I tend to think that a framework and method are what really matter, more or less exclusively. Though the findings (of relative merit or only rankings) need to undergo a sanity check. Are they really plausible? With any clear anomalies sparking further analytical refinement of some sort. That’s in line with the traditional scientific paradigm.
The Sobers Gap: A Veritable Migraine
It is interesting that Houldin embarked on devising his method because of wanting a quantitative assessment of Garry Sobers as an all-rounder. He reckons, quite understandably, that Sobers is the greatest all-rounder and the greatest player ever, and sought a way of demonstrating this:
“My motivation…a lifelong preoccupation with trying to find a way to show, from his record, that Sir Garfield Sobers was the greatest cricketer.” (Second article, page 35).
Yet the magnificently multi-talented Sobers comes only equal 18th (along with five others) when applying his method, as so far developed: Sobers being credited with 21 (net) points.
Sobers comes well behind Sir Donald Bradman who tops the list on 64 points, and also substantially behind the runner-up, Sir Richard Hadlee on 33 points who is not considered by Houldin to be a genuine an all-rounder. Alan Davidson – having a claim to be an all-rounder denied – comes third on 30 points – two-fifths ahead of Sobers. Davidson is followed, in fourth place, by the great all-rounder Imran Khan with 28 points.
Sobers is rated as one of a quartet of best all-rounders – including Imran Khan, Jacques Kallis and Keith Miller – but is ranked below the other three.
Houldin implies that his framework – as it currently stands, in the revised form – is unable to do justice to genuine all-rounders, whoever they may be. Within the basic framework of assessment he puts forward, I doubt that any amount of refinement per se – or plausible sensitivity analysis – could do enough to bridge the Sobers Gap in relation to other all-rounders. Let alone bridge the gap with all other players.
Houldin seems to put some hope in disaggregating the very lengthy post-WW1 era and delving into the relative quality of bowling that different batsmen have received – both of which won’t fix the ranking problem in my view.
For those who deliver their overs in a dual style or capacity, what would still be omitted from the contribution arithmetic, and resulting points scores, is that they enable a greater variety of bowling attack to be employed than if they were one-dimensional.
This is of positive value, as it gives a captain a greater ability to respond to changing circumstances as a game unfolds. Put another way, a dual capability allows for the inclusion of an additional bowler than otherwise. Hence the extra value might, in principle, be captured by the observed deeds of an additional bowler and his estimated points score.
Consistency of treatment would be required. That is, treating equally all those who deliver in two (or more) styles and qualify as genuine all-rounders, not only Sobers. Then one could devise a table devoted to displaying these dual/triple bowling all-rounders by themselves. My hunch is that Sobers would come top of this table by a comfortable margin.
Also, as Houldin points out in his August 2024 article (page 39), Sobers – and other all-rounders who bowl in two (or more) styles – have a pre-match value for selectors. “Test team selectors are often confronted with choice of an additional spinner or seamer depending on expectations about the state of the pitch: with Sobers, they could hedge their bets.”
The biggest problem, though, is the obdurate Bradman, who sits well ahead of all the rest. His points score has actually gone up from the previous table of Houldin’s November 2023 article. This is because of a change made to the associated batting benchmark, which is now put at 36 instead of 43.
Accordingly, if the preponderance of cricket followers’ opinion on the best ever cricketer is to be believed, we are left with a basically wonky approach even for genuine all-rounders – which are only a sub-set of Houldin’s overall scope of interest.
Some might go somewhat further than myself. From one eminent ACS member, whom I shall cloak in anonymity, I received this pithy comment:
“His approach is nonsense really. If there were a better way of doing cricket stats, Ashley-Cooper (shown below) would’ve worked it out in the Golden Age.
For those not acquainted with the work of Frederick Samuel Ashley-Cooper (b. March 1877, d. January 1932) take a look at Peter Wynne-Thomas’ book, Cricket’s Historians (published by the ACS in 2011).
A Tip for Narrowing the Bradman-Sobers Differential
As indicated, Bradman’s points score is way ahead of that estimated for Sobers: indeed, three times as great (64 versus 21). At this stage – sotto voce – I have a tip for Russell Houldin that would considerably reduce The Don’s batting average, and hence his points score. This comes in four additive parts:
(i) Deduct all of The Don’s “dead runs” which amount to 9% of his total runs scored – likewise for Sobers who had far fewer of them, only 0.7% – so reducing their respective averages to 91.0 and 57.4. Dead runs being those runs scored when the opposition had a negligible chance of winning the match.
(ii) Establish their respective dominance ratings, as given by the number of standard deviations their batting averages (net of dead runs) stand above the grand average for their own playing spans: coming out as Bradman 3.24 and Sobers 1.95. These degrees of dominance are then translated into the scoring context of the new millennium in order to show what batting averages are implied for each of them today: giving Bradman averaging 70.1, and Sobers averaging 53.0.
(iii) Allow for the advances made in batting technique between successive decades: resulting in a reduction of 4.5% for Bradman and 3.6% for Sobers.
(iv) Finally, allow for differences in the length of their respective Test careers. Implying a premium per completed innings of 1.6 runs for Bradman and a premium of 6.0 for Sobers.
The rationale behind this is explained in my book, referred to above (note xi), and in my article Standardising Test Batting Averages: A Career Length Overlay, appearing in The Cricket Statistician journal, May 2024 issue.
When both their batting averages are fully standardised in this way, Bradman’s batting average comes down to 68.6 (lowered by 31) while Sobers’ average becomes 57.1 (lowered by only 0.7) – giving Bradman a net advantage of 11.5 runs per completed innings.
Turning to their respective performance as bowlers, I have considered both runs conceded per wicket taken and number of wickets taken per innings in which they got on to bowl. On the former measure, Bradman is on 36.0 runs per wicket whilst Sobers is on 34.0; and on frequency of dismissing opponents, Bradman is on 0.22 wickets per innings whilst Sobers is on 1.48.
So now there is a judgement to be made. Does Sobers’ superiority in bowling outweigh Bradman’s superiority at batting? This might now work out to suit Houldin’s personal prejudice, depending on the trade-off applied. So over to you, Mr Houldin!
In any event, this portrayal is a big improvement – in both quantitative terms and in realism – on the 43 merit points differential in favour of Bradman.
A Heady Diversion
In further pursuing the comparative merit of individual players, Russell Houldin has been “Drawing inspiration from baseball’s sabermetrics and its influential Wins Above Replacement (WAR) model.”
His thinking in this regard formed some of the material presented to the recent ACS gathering, mentioned earlier. The so-called WAR approach is aimed at measuring a baseball player’s total contribution – summing, or consolidating, those made in the roles of offence, defence and base-running. The aggregate contribution is then translated into runs – as Houldin arrives at, analogously, with his net points for runs scored and runs saved in cricket.
Furthermore, with WAR the estimated number of runs a player gets are then translated into team wins, with around 10 runs generally considered to equate to one win. This step is somewhat tendentious.
Houldin believes that such a procedure has a potential application for merit rating players in white-ball forms of cricket. But he doesn’t see basketball’s WAR method as having an application to red-ball forms of cricket because of the presence of draws. WAR can’t cope, analytically, with the concept of drawn matches.
One cannot, of course, simply put all drawn Test matches to one side and press on regardless. Draws accounted for around 20% of Tests for the 2010-19 decade, falling to 13% subsequently: though with 30-45% being the norm for the 1970s through to the 2000s. So sizeable chunks of many players Test careers wouldn’t feature.
A way around this problem would be to project each drawn Test match to a positive conclusion, for one side or the other, by doing some common sense extrapolations. Okay, this would involve a mountain of work. As a short-cut, one could assess which team had the advantage when a final halt was called to the play. That would be quite easy to do after a couple of gin and tonics or chardonnays. Any tough-to-call matches could be decided by the toss of a coin!