Rating Shane Warne and Sydney Barnes

As I mentioned in this post a good use of your compulsory CV - 19 lock down is to read https://redballdata.blog/  

It's a bit geeky, which I like, and Mr RedBallData has some interesting ideas on how to predict and select, cricket and cricketers, an area which I find fascinating, although I don't have the maths to do anything other than cheer on from the sidelines.

Something that really appealed to me was his list of the best bowlers of the last 50 years.  The approach used to compile the list was, rather than  look at average, to order the bowlers by their impact on the averages of the batsmen they played against (you can get a bowler's career broken down this way on cricinfos stats guru.).  So, for instance,  let's say we have a bowler who has only ever bowled to two batsmen, A&B.  A has a career average of 50 but our bowler gets him out for 25; B has a career average of 10, our guy though has a bowling average of 15 against him.  The reduction in average is (-25+5)/(50+10) = 33.33% reduction.

I've taken that approach and used it on a couple of players who weren't on the list of the best bowlers of the last 50 years, Shane Warne and Sydney Barnes.

Shane Warne

In terms of the RedBallData list Shane is the dog that didn't bark, the Aussie that didn't sledge, he's not on it all!  This is because the list only includes bowlers with an average of under 25 and Warne's career average was a touch over.  So I thought I'd see whether his career was better using the impact on opposing batsmen's average metric rather than raw average.  And the answer seems to be well yes, but only a bit.  Warne's impact is to reduce an opponents average by 13.  That is better than Wasim Akram, 19th on the list of best bowlers of the last 50 years but not anybody else on the list.  Warne on the impact measurement would be the third best spinner of the last fifty years, behind Muttiah Muralitharan and, Ravi Jadeja.

There is a basis to argue both average and reduction in opponents average underestimate Warne.  First he took wickets, lots of them, over 700 in total.  That's evidence of a certain persistence and it shows Warne to be have been Australia's first choice spinner over a protracted period.  Playing as part of what was, generally, a four man attack, Warne would always have to bowl his overs, even if the pitch wasn't helpful.  By contrast a bowler like Ravi Jadeja isn't an automatic choice for India and he's, perhaps, tended to miss games when conditions aren't in his favour.  So his average of 24.62 should be seen against his 213 wickets.

On a related but I think distinct point, spinners as a class have an advantage over fast bowlers because they can bowl more overs.  A fast bowler can create havoc in five overs but then the captain has to weigh up the benefit of squeezing out a few more overs at a vital point in the game against resting the bowler to ensure he continues to be effective.  With a spinner there is no dilemma, if the spinner is taking wickets he can keep bowling until there are no more wickets left to take.

This table shows Warne and Muttiah Muralitharan took more wickets an innings than Malcolm Marshall.  Ravi Jadeja has quite a low per innings wicket figure, perhaps another indication that his average based statistics slightly over value his contribution.

Bowler	Innings bowled	Wickets	Wickets per innings
Malcolm Marshall	151	376	2.5
Ravi Jadeja	94	213	2.3
Shane Warne	273	708	2.6
Muttiah Muralitharan	230	800	3.5

But Shane Warne revisionism should only go so far.  Yes he was an excellent bowler, one of the best leg spinners ever, but to claim " to many he remains the greatest spinner - if not bowler - of them all" Is taking things a bit far.  Generally in trying to come up with a "best ever" list we are left to struggle with matching the apples of one era with the oranges of another.  With Warne there is an immediate and contemporary point of reference in Muralitharan.  Murali took more wickets than Warne, his raw average was lower than Warne's and his impact on opposing batsmen's average was higher. Murali was the king of wickets per innings, taking almost a whole extra wicket when compared to Warne.  We don't know everything, we may not know much at all, but there is no way Warne can come any higher than second on the all time list of spinners because Muralitharan was undoubtedly superior.

Sydney Barnes

If Warne looms large in modern cricket Sydney Barnes is an ephemeral character. He played his 27 tests between 1901  - 1914 and although he was still playing professionally well into his fifties there can be few, if any, living today who saw him bowl.  Statistically though he he has one of the best test records, 189 wickets at 16.43.  But was he any good?  Batting averages were low pre - World War 1 and Barnes isn't the only bowler from test cricket's early days to have remarkable figures.  George Lohmann, playing in the late nineteenth century, took over a hundred test wickets at a remarkable 10.75.  By rating on the basis of reduction in batsman's average rather than raw average it should be possible to distinguish how much of Barnes' success was the result of him being exceptional and how much due to bowlers, in general, having a good time of it in the early twentieth century.

I calculate Barnes' impact on the average of opposing batsmen was a reduction of  29%.  That's impressive, don't forget Shane Warne reduced the average of opposing batsmen by 13%, but it's not quite as stand out as Barnes' career average.  In the last 50 years Malcolm Marshall, Curtly Ambrose and Glen McGrath all have slightly larger negative impacts on a batsman's average than Barnes achieved.  So Barnes was indubitably a great bowler but not quite the "freak" his raw average makes him appear.   

And even the enhanced metric of reduction in batsman's average only takes us so far.  Barnes' figure isn't really comparable with McGrath et al because he played in a different era.  The standard of cricket might have improved (or declined) in both batting and bowling between the two eras which wouldn't show up in any average based method. I'm not sure if there's any way to use statistics to comprehensively rate cricketers from different eras.  Especially when we are comparing the early twentieth with the early twenty first century.