Australian selectors are going against history and maths by selecting Marnus Labuschagne and Marcus Stoinis for the national Test team.
By the time you read this the second Test against Sri Lanka may well be underway – Marnus may have scored another 50 runs or even a ton and Marcus may have done enough on debut to be pencilled in for the Ashes.
If so, some of you may have a good laugh reading what I have to say and think I’ve got mud on my face. If you do, you’ll have missed the point.
The point is that numbers in cricket are very stubborn things that reflect deep underlying patterns in how players perform over time. In picking Labuschagne and Stoinis for Test cricket the selectors are defying what the numbers tell us about cricket performance over the last century. Who knows, they may be proved right – nothing is impossible.
In theory you could sit the proverbial monkey down at the typewriter and he could bang out a Shakespearean play, but they are attempting to achieve the extremely improbable.
A lot of cricket fans were perplexed by these selections, mainly because there seemed to be more qualified players in terms of performance or because the search for that elusive all-rounder is nonsense.
Many were moved to ask: Aren’t their respective first class batting averages – around 33 – just a bit too far below the normal standard?
So I wanted to find out whether we’re missing something here. Have any of the greats or even a few of the solid but not stellar Test batsmen made their Test debut with a first-class average of 33 or lower?
In pursuit of this Holy Grail we cranked up the Cray computer at Nerdsville Central and dissected the stats for Aussie batsmen over the last 100 years – essentially since World War I. Stats from the 1920s and 1930s are pretty much the same as those of today in terms of the kinds of scores and records regarded as being very good.
On the other hand, the pitches before WWI were much dodgier. Australia’s two greatest batsmen of the period, Victor Trumper and Clem Hill, averaged only 39 in Tests. I can’t see any huge differences in conditions sufficient to say things are much harder today than 50 or 100 years ago – there might be more varied surfaces and bowlers to face around the world today, but that is counteracted by the advantage of bigger bats, smaller grounds, helmets and playing a lower percentage of games against weaker opposition.
(Ryan Pierse/Getty Images)
What bar to set for a successful Test batting career? Most would agree that an average over 50 is great; 40 to 50 is good or very, very good; and in the high 30s is solid but presumably inconsistent. But to cater for a very wide definition of success we kept the bar reasonably low – a Test average of more than 35 achieved over a minimum of ten Tests or three years. We then worked out what was the first class average of the batsman at the time of his Test debut.
The results paint a very clear picture. Of the 64 Australian batsmen making their Test and first-class debuts since 1918 who averaged over 35 across their career only two made their debuts with first-class debuts under 35 and only one with an average of 33 or lower: Keith Stackpole.
Keith Stackpole, however, didn’t even debut in the top six. Averaging only 31.9, but with some claims as a part-time spinner, Stackpole actually batted at No.8 on debut in 1966 in a late middle order that included two other batsmen-cum-spinners or vice versa in Tom Veivers and Ian Chappell. Stacky went on to become a pugnacious opening batsman with a couple of really good series in the early 1970s and finished with a solid if not stellar average of 37 over eight years with seven tons.
The other batsmen debuting with a first-class average below 35 was, curiously, Kim Hughes on 34.6, a player renowned for flare and some great innings for Australia. Even so, despite a few great innings for Australia, Hughes’s Test average finished on a somewhat disappointing 37.
The observant among you will point out that this exercise is not a randomly based trial of how batsmen averaging 33 or fewer fare in Test cricket. Most of the 33-and-belows were not given a chance, given the selectors’ general wisdom of favouring higher averages over lower ones, and usually having players with much better records demanding to be selected.
We ran out of grease for the Cray computer crankshaft so couldn’t go through the myriad of batsmen who averaged less than 35 to see whether was a strong sprinkling of low averages on debut among the poorer performers. Among the few we sampled there was a decent sprinkling of very promising players prior who didn’t fulfil that promise: Phil Hughes (55.6 first-class pre-debut, 32.7 Test average), David Hookes (52.2 first-class pre-debut, 34.4 average) and Greg Blewett (45.1 first-class pre-debut, 34 average). If you suggest that some of these players were successful enough in their own way, fair enough, but it only supports the argument that a higher first-class performance prior to debut is more likely to result in success.
Among the successful batsmen with pre-debut averages under 40 there are a few surprises. They’re some of our greatest-ever batsmen – Greg Chappell (38), Allan Border (37), Michael Clarke (38) and Adam Gilchrist (39). But these could be explained in different ways. Chappell and Clarke faced more seamer-friendly pitches in county cricket stints compared to contemporaries confined to the Sheffield Shield and Border arrived a fraction early due to the World Series Cricket exodus.
The successful batsmen with the best records on debut are as follows, shown along with their final Test averages: Ponsford (96/48), Woodfull (75/46), Rick McCosker (72/40), Bradman (71/100), Bob Cowper (68/47), Hayden (63/51), Norm O’Neill (62/46), David Warner (62/48), Steve Smith (56/64) and Mark Waugh (56/42).
Apart from a couple like Bradman and Smith and a few like Chappell, Border and Clarke noted above, the pattern for the majority of successful batsmen was to end up with a Test average close to or a few points below their pre-debut first-class average. You might be interested to know the company Joe Burns keeps in terms of a pre-debut average of 42, for example, is Steve Waugh, Colin McDonald, Jack Ryder, Andrew Symonds, Colin McDonald and Ross Edwards. A case for patience rather than panicky axings as happened in 2016 and 2018?
Of course a higher first-class average does not guarantee success in Tests, but it is a lot better predictor of the probability of success than a lower one. There is a big difference between an average in the high 30s and one in the low 30s. It shouldn’t be a big surprise that Mitchell Marsh debuted with a first-class average of 29 and has finished, for the time being at least, on 25.
The reasons for these underlying patterns are because cricket scores are ultimately about probability and chance. We like to create stories about individual scores – Travis Head ‘battled hard’ for 81 or Joe Burns ‘failed’ with a mere 15.
(Jono Searle – CA/Cricket Australia/Getty Images)
But what shapes scores and averages and the frequency of big scores is a batsman’s underlying ‘fatal error rate’ that sooner or later comes back to bite them as well as the natural strike rates of bowlers they come up against and the times they get the unplayable ball. Think of all the plays and misses, the balls in the air and the near-strangles down the leg side that accompany many successful innings. There is a great degree of chance that combines with players’ techniques, skill, temperament and shot selection to determine individual scores. It’s only in the long run that luck and chance tend to even out and the underlying quality shows in overall averages.
So if Labuschagne or Stoinis do well in Canberra, I will be concerned, but not because it in any way disproves my little rant about numbers. A sample of one proves very little and doing well against this Sri Lankan attack proves even less. But if these two, great blokes they may be, end up in the Ashes squad, I will feel our chances of success in the medium and long-term are diminished compared to having players like Burns, Patterson or Maxwell with first-class averages over 40.
The current selection panel and coach appear to be on a different wavelength when it comes to numbers. These are apparently boring things not to be given much heed compared to feel-good factors about a player’s personality. And they have gone into a completely different astral plane in picking Stoinis on the basis of T20 form – if you think it worked with Warner a few years back, it was his pre-debut first-class average of 62 that was the real clincher.
The Stoinis selection might best be summed up by the following dialogue: “Hey Usain, you’ve been getting some great times in the sprints – we reckon you’ve got potential at the longer distances and we’re going to enter you in the marathon. I know your marathon times are pretty crap, but Warnie reckons you’ve got something about you and Justin Langer says you’re an upstanding fine young Jamaican.
“In any case, the Kenyans and Ugandans aren’t running in this one, so you might do okay. Then you’ll be on the plane for the Boston Marathon and who knows?”
DR
Guest
Maybe Greg Chappell read this article, realised what a mess he had made and decided it was time to let someone start the rebuild in Australian cricket!! Seriously great article, thank you for sharing. To me, averages tell a story, there is always a few variables as you point out but averages over a decent sample period are a pointer to the probability of success. If I may offer an opinion..... a batter has a low average for one of 2 reasons. They are either techinically inept and get dismissed cheaply too often or they lose concentration/don’t have the temperament to play long innings and thus get out for lots of pretty 30’s. Either one could be described as a fatal flaw and will not disappear at Test level. The events of the Second Test fit with this. Burns failed in the second innings but in the first innings he got to 20 and turned it into 180, Patterson got an early let off but once set turned that chance into an unbeaten hundred in his second Test match innings. Compare that with Head and Labuschagne, Head took 13 innings to get his 1st century and Labuschagne has batted 8 times without a century. Both have FC averages in the low-mid 30’s and their past performance fits with this. With Head I think it’s temperament and with Labuschagne more technical.
Jon Richardson
Roar Pro
Thanks Dad... I mean, Pete..
Jon Richardson
Roar Pro
Thanks Bush, I probably wasn’t super clear about Chappell Border and Clarke. I wasn’t suggesting below 40 was mediocre, just lower than you might have expected for these guys. Not sure it was because they were young though. Quite a few others - Hookes, Phil Hughes, Ponting, Martyn, Smith and obviously Bradman - were picked at a similar age with already stellar FC averages. I think playing county cricket for Chappell and Clarke lowered the average a bit as well as increasing the sample size, ie a lot more games under the belt. I can’t imagine Hooked or Hughes coming out of a couple of seasons of county cricket aged 21 with averages in the 50s or 60s .
Paul
Roar Guru
makes sense Jon. I also forgot to say, I enjoyed your article
Jon Richardson
Roar Pro
Thanks MJ . Numbers aren’t everything and they can be crude ciomparing one era against another. But often they tell you SOMEthing.
Jon Richardson
Roar Pro
My point was that the best of the pre WWI era averaged under 40, as an example of why not to start the analysis until after 1918 when conditions were apparently comparable to subsequent decades.
Paul
Roar Guru
especially as Boon was either opening or batting 3
George
Guest
"They are stuck in some 1970’s view of the cricket and world." Spot on. Explains the all-pervading mateship claptrap.
Peter Warrington
Guest
This is my most favourite article in near-centuries. Great stuff!
Peter Warrington
Guest
yes, those are the worst selections, but make the most compelling case. your point re duration of scoring is timely. Hookes was the best/worst example of someone "banging down the door". Essentially he a purple patch in a few weeks, on dry tracks incl Adelaide, when the test players were away in NZ - Victoria minus Walker and Hurst; Qld minus the injured in the Thommo; NSW minus Gilmour and O'Keefe. The tracks were easy - Richie Robinson made 185 out of 500 for the Vics; Queensland chased down 262 to tie on the 4th day in 220 minutes; a guy names Handrickan scored a ton for SA in the game against NSW, in which NSW made 400 without a ton, despite being without Davis, McCosker, Turner, Walters and Gilmour (Beatle Watson a solid 70 batting at 3, for his 3rd state!), Stuey Webster getting 98* in the second dig at almost a run a minute. So there is now doubt Hookes' average was super-inflated by recency, and context. Then the Cent Test 50 was blown out of proportion - it was a dead track, 600 runs scored for the last 14 wickets of the match (after Hookes' dismissal). Hookes actually did OK in england in 77 and was one of few shining lights in WSC and was shafted after that, especially when he was needed (post 84). He was a big part of the 82-3 Ashes win, many solid partnerships with Hughes, despite their past rivalry and future enmity. But a good batsman is all he ever was, and the coaches are to blame for that. Hughes is the other interesting one. Not just because Hookes vaulted over him into the Australian team, with his softer runs in the Shield comparing favourably with Hughes' travails as the spare bat in the NZ touring team. Hughes then didn't get to debut until the end of the Ashes tour, when his form and spirits were sunk. The difference was that Hughes had played well crafted innings over a couple of seasons. On Shield debut against a test quality attack. Against the Windies. Against Pakistan, in an unfeasible chase. His average was clearly weighed down by concentration issues, which bedevilled him until his later years. His average when picked was weighed down by recency - about 38 was his level. But he managed to elevate his test average over the years to 41, Ian Chappell level, until the final fall, when his country needed him the most but he was gawn. The collapse from 41 to 37 was brutal. Arguably, neither should have been picked when they were. But that was the Australian way. We had not had youth into the team under Chappelli except for Davis, so Yallop, Cosier, Hookes, Serjeant, Hughes were all given a shot, often at the worst time for them. But renewal is eternal.
MJ
Guest
This is the kind of article that keeps me coming back to the Roar. Great work, wish this evidence based analysis was driving the selection discussions on TV & elsewhere. Meanwhile the national selectors keep betting against the odds and the cricket fans gnash their teeth.
The Bush
Roar Guru
I actually think the author was implying exactly what you're staying - that Hill and Trumper were actually very, very good, despite "only" averaging 39. But the statistical analysis he's undertaken doesn't work if you're including statistics that vary so widely. Obviously there were numerous successful batsmen pre-WWI who debuted with averages below 35 and became key members of the team - because averages overall were much lower. The point of picking post-WWI as the starting point is that averages have been remarkably stable since that time - i.e. an average of 50 or above is very, very good, averages between 40 and 50 are good and averages between 35 and 40 are serviceable. But yes, you're point is valid - you always need to compare averages of batsmen with their contemporaries before you do so across generations. David Boon "only" averaged 43 across 100 tests, but I'd still rank him above guys like Martyn, Hussey and Langer, who all had much higher averages, because batting in the late '80s and early '90s was much more difficult than it was circa 2000 to 2013.
The Bush
Roar Guru
Great article Jon, really interesting stuff. It's one of the reasons why I really don't like picking guys off the back of extremely recent form. Cameron Bancroft may yet prove to be a decent first class bat, but he debuted with an average well under 40, based on a small vein of form in five FC games. He then went on to average 30 across is 8 tests. As you say, statistically he had to come back to reality. M Marsh was consistently picked based on either potential or one off centuries. But he's never actually put together a body of FC work for any prolonged period of time. If he can't do it at that level, I'm not sure why anyone thinks he can do it at test level. One thing that is important to understand when looking at these stats though, is the age of the batsmen and how many games they've played. It's not necessarily surprising that Border, Clarke and Chappell (I'm ignoring Gilly as he was a 'keeper') all debuted with sub-40 averages and then went on to be very successful. Chappell was only 22 when he was selected and Border and Clarke were only 23. An average of 37 or 38 at such a young age is actually not that bad. Labs is 34 and he's 4 to 5 runs worse than that (and older). The reality is they were all extremely talented batsmen and would have demonstrated this at certain points (under age tournaments, second XI comps etc). They were also years away from the peak period for a batsmen when you expect them to score heavily. In other words, they had years of potential improvement ahead of them. It's one of the reasons why I'm not prepared to give up on Head or even Labs for the moment - they're young and could yet improve into decent batsmen. What blows my mind is when the selectors pick guys like Stoinis. He's 29. The last few years should have been the time when he really stepped up his game and he should be at his peak now. Instead he's have two very average summers, though is currently having a decent enough one. He averages 33 after nearly 10 years of professional cricket. The reality is he is never going to turn into a test match standard batsman. To some degree he's been hampered by things like limited overs cricket and being an "all rounder", but it doesn't matter - the current Australian side simply cannot afford to carry a guy that'll average 30 at number 6. I'm not just criticising the Stoinis selection though. The selectors have done this for years; Doolan, Quiney, Finch. These guys were all selected at the ages of between 28 and 32 and all had FC averages in the mid-30s - they were never, ever, going to turn into viable test players.
Paul
Roar Guru
"Australia’s two greatest batsmen of the period, Victor Trumper and Clem Hill, averaged only 39 in Tests." This is a simple example of numbers not telling the real story. I've read a number of articles over the years that suggest an average of 39 in that era, is equal to an average, somewhere between the high '40's & mid '50s today. The other test is to compare thee averages with their contemporaries and by that measure as well, Trumper and Hill are standouts as Labuschagne & Stoinis are (in reverse) with their FC averages.
Vicboy
Roar Rookie
Excellent article The next step is the schedule, so we can pick a test team based on first class form, or the selections will continue to be based on good bloke, gut feel, 20/20 form etc
anon
Roar Pro
The selectors just aren't that bright. They probably don't pay attention to other sports, other trends. They are stuck in some 1970's view of the cricket and world. The got their jobs on the back of cronyism, brown nosing the right people rather than for having sharp, analytical minds. Really, selecting the best players isn't hard. It could be done using an algorithm. These selectors are the equivalent of mug punters at the track. Langer and the selectors like to justify their positions by making these "hero picks" that time and again blow up in their faces. Like the mug punter backing the 20-1 horse because it finished well in the last start, or because he thinks it's a mudlark, or because it has a good trainer (that he's mates with). The selectors are the equivalent of mug punters when they punt on Mitch Marsh for 31 Tests, turn Finch into a Test opener despite not being good enough to open for Victoria, parachuting Stoinis into the Test side despite no real red ball credentials, etc. They want to roll the dice, take a punt so that when the "hero pick" works out 1 in 10 times, they can turn around and tell everyone I picked him, I identified the talent in him early, how good am I? I think Langer is now 5 wins out of 29 games against legitimate cricketing nations. It's appalling, it's a mess, but no-one will hold him accountable.
Sportstragic
Guest
A bit harsh on Bradman. 99.94 rounds to 100.
Dexter The Hamster
Roar Rookie
Agreed. And one of my biggest gripes is people mentioning Warner as an example of a T20 player who made it in test, forgetting that he proved himself in FC cricket first. Feel Stoinis should have the opportunity to do so as well before getting the call up. On an inconsequential matter though, I understand you rounded those batting averages to the nearest whole number, but you cannot give Bradman an ave of 100. I know it goes against the "maths", but you needed to give him a 99.