SA’s best Test batsman? For sports aficionados, it is not a hard question.
The trouble is, the right answer — Graeme Pollock — turns out to be wrong.
Researchers at two universities in the UK have rewritten the rankings after discovering that batting averages tell only part of the story.
Based on their findings, Pollock — with an average of 60.97 in 41 innings from 1963 to 1970 — is deposed by Jacques Kallis as SA’s greatest Test batsman.
And from being fourth-best in the world in the official rankings, Pollock slumps to 15th when he is compared to all 2,856 batsmen who played Test cricket for 10 countries between the first match in 1877 and the 2,259th in August 2017. Kallis, who played 280 innings between 1995 and 2013 and retired with an average score of 55.37, rises from 16th to 10th.
Richard Boys from Newcastle University and Peter Philipson from Northumbria University took numerous factors into account when they built a model to rank the best Test batsmen, and have published their results in the Journal of the Royal Statistical Society.
They found that a ranking can easily be too high if, for example, a batsman played only against weaker Test nations or only in home matches.
Their model aimed to account for numerous factors that influence performance, such as era, number of innings played, player’s age, home advantage and the opposition team. Even when they corrected for these factors, there is considerable variation in the innings-by-innings performance of a batsman.
What we have shown is that it doesn’t make sense to rank batsmen solely on the basis of what they have achieved in Test matches because this simply doesn’t give an accurate picture.
"Current rankings based on averages don’t take account of all the different influencing factors or the uncertainty every time you step out to bat," said Boys, who is professor of applied statistics and head of pure mathematics and statistics at Newcastle.
"And because the numbers of innings we are looking at are relatively small, you can’t say what their true worth is — a batsman would have to take millions of innings for us to be able to rank them accurately."
Boys and Philipson modelled the inning-by-innings variation to determine what might have happened if the batsmen had played many more matches, and found significant uncertainty in current player rankings.
The one constant for all ranking systems is Donald Bradman, whose status as the best batsman in history is confirmed by the study.
In 80 innings for Australia between 1928 and 1948 he scored an average of 99.94
"What we have shown is that it doesn’t make sense to rank batsmen solely on the basis of what they have achieved in Test matches because this simply doesn’t give an accurate picture," said Boys.
"There are too many factors influencing their success: who they’re playing, is it at home or away, and how experienced they are. More importantly, batsmen play in relatively few Test matches to give an accurate picture of their ability and so there can be considerable uncertainty on a player’s rank.
"While the top spot remains fairly static, when you get nearer the middle of the table the uncertainty is huge."
Philipson said the batting average method of ranking players assumed they were equally likely to lose their wicket at any time, but this was not true for most batsmen, who displayed more frailty at the beginning of an innings. "Many players have an increased chance of getting out for a duck. We also see that there is very little practical difference between an average of, say, 55 and one of 50 bearing in mind the large uncertainty."
While the study might be the subject of friendly banter the next time Pollock, 74, and Kallis, 42, meet, they still have something in common: they are the only South African batsmen in the top 20 on both "official" and "revised" rankings.