Cricket abounds with statistics of all shapes, sizes and forms.
In this article I will add to the maze by trying to delve in more detail into Test batting averages, most notably for the current big four: Steve Smith, Virat Kohli, Kane Williamson and Joe Root.
The one irksome thing about batting averages is that it is not contextualised by the quality of bowling. I have painstakingly attempted to do so here on an innings-by-innings basis and have come up with two types of batting averages: premium batting average (PBA) and semi-premium batting average (SPBA).
The former means the batting average is against the best bowlers, whom I have classified as having bowling averages of 29 or less, while a semi-premium batting average is against bowlers with bowling averages of 30-39. I have also included separately the batsman’s normal average as is usually quoted.
I have drawn on the treasure trove of statistics from ESPN Cricinfo, and the data is right up to date and includes the very recently concluded India-England Test series.
However, I freely admit there are still significant drawbacks with my developed approach.
(Photo by Ryan Pierse/Getty Images)
The first is that the statistics I drew from attribute all runs scored by the batsman to the bowler who ultimately dismissed him. For example, if a batsman is dismissed by a bowler for 100, then it goes onto the bowler’s wicket tally at a cost of 100 even though clearly in scoring a century the batsman has scored runs against multiple bowlers. If the data were more available, a more forensic analysis would involve runs scored against each bowler innings by innings.
The second is that I have included not-outs and run-outs and attributed these runs to the best bowlers – average less than 29 – on the grounds that not even the best bowlers were able to get the batsmen out. I could have excluded not-outs and run-outs, but that would skew the batting performance significantly.
The third is that I have not taken into account other variables, such as whether a game is at home or away, first or second innings, state of the match and so on.
Fourthly, the bowling averages used are those at the conclusion of the bowler’s career, when the player is retired, not the bowling average at the time that the bowler dismissed the batsman. The current latest bowling average is used for bowlers still playing.
Nonetheless, despite all this I think the analysis is interesting and potentially informative.
With a knot in my stomach tightening and at the risk of incurring the wrath of the cricketing gods I have also performed the exercise for the greatest of them all, Sir Donald Bradman, as the ultimate benchmark, hoping against hope that the analysis does not diminish his performance.
The results are shown in the following table.
| Batsman | Premium batting average (PBA) | Semi-premium batting average (SPBA) | Normal batting average | Dismissals against premium bowlers |
|---|---|---|---|---|
| Steve Smith | 61.17 | 55.52 | 61.80 | 56.7 per cent |
| Virat Kohli | 48.11 | 56.9 | 52.37 | 46.1 per cent |
| Joe Root | 45.5 | 53.97 | 49.24 | 51.9 per cent |
| Kane Williamson | 49.9 | 48.5 | 54.31 | 45.3 per cent |
| Don Bradman | 86.3 | 95.6 | 99.94 | 62.3 per cent |
So first things first. Steve Smith and Kane Williamson have premium batting averages higher than semi-premium batting averages – much more apparently in Smith’s case. This means that Smith and to a lesser extent Williamson have higher batting averages against the very best bowlers – that is, those with bowling averages less than 29 – than the next tier of bowlers, who have a 30-39 bowling average, noting all the caveats I discussed earlier.
This is remarkable, considering that even the Don’s premium batting average is below his semi-premium batting average.
As an example, in Steve Smith’s case the PBA is influenced significantly by his batting average of 77.5 against Jimmy Anderson, who has a bowling average of 26.46, thus making him a premium bowler. Further, and unlike the others, Steve Smith’s PBA is almost identical to his normal batting average, indicating the strength of performance against the best bowlers. All the others have PBAs quite a bit below the normal batting average.
Further, the gap between Smith ad Donald Bradman’s PBAs is smaller than the gap between these two for the normally quoted batting averages, although it is still very significant. There are only two batsmen, Kohli and Root, whose SPBA exceed their normal batting average, perhaps indicating they are not maximising their performance against all types of bowlers.
The final column shows the percentage of dismissals against the best bowlers. It demonstrates that for Don Bradman and Steve Smith it takes the very best bowlers to dismiss them, indicating in part that these two do not or have not surrendered their wickets to the lesser bowlers, unlike the others, especially Williamson and Kohli.
I cannot conclude this article being accused of heresy. It is clear even with reformulated numbers that Sir Donald Bradman stands head and shoulders above the rest. Further examples illuminate the greatness of the Don. Don Bradman’s Test batting average was 77.6 against Harold Larwood compared to Larwood’s overall Test bowling average of 28.35. I often found this kind of discrepancy, sometimes by an ever greater magnitude, when looking at Bradman’s performance against the other great names of English cricket, like Alec Bedser, Hedley Verity, Maurice Tate and Bruce Yardley.
Further, unlike the other contemporary batsmen analysed, Bradman was up against a number of bowlers whose averages were 25 or less. In addition, uncovered pitches would have skewed things in the bowler’s favour.
Finally, there is only one premium bowler, Bill Voce, who performed well against Bradman from my observations.
Notwithstanding the limitations and caveats of data, this has been an interesting exercise from my point of view and one I hope could lead to further development of the game’s statistics by the experts.
Bobbo7
Guest
An interesting read. Well done. The issue with these types of pure average articles is it does not and cannot account for attitude. Williamson often throws his wicket away going for quick runs in a second dig - always the team first. We saw this against the WI and Pakistan.. Not to say others on this list don't, but it is quite apparent Williamson is not interested in protecting his average. Accounting for difficult of pitches is also a tough one. I always though deVilliers was some player to average over 50 in SA given some low scores there.
Just Nuisance
Roar Rookie
By the way I got those stats from a site called cricmetric. Gives a really good analysis on any major cricket ground in the world.. Super interesting.
Just Nuisance
Roar Rookie
What did Darryl Cullinan average against Shane Warne? I doubt I'll need a calculator. :laughing:
Just Nuisance
Roar Rookie
Took you up a bit on that Micko and did a bit of homework on The Wanderers Cricket Ground and it made for interesting reading.. 1. Avg batting total overall all 1st class and Test matches 247. 2. Average winning percentage team batting first 60%., batting second 30%. Draws 10%. By comparison, SCG winning percentage team batting first 30%, 2nd 30%, draws 40%. 3.Bowling avg economy, offbreak 55, Legbreak 38, orthodox 32, Medium pace 28, fast 23.. No wonder SA produces so many great quicks and no spinners. What this reinforces is that the type of pitch and its location has to have a fundamental impact on an individuals averages..
Chip
Roar Guru
Thanks Matth yes you could well be completely correct about bowlers dismissing bunnies. There are so many caveats about all of this. The spin is mixed of the current lot at any rate. Lyon has an average over 30 and Ashwin just over 26
Micko
Roar Rookie
Maybe we can analyse world cricket grounds and see what test venues have the most difficult pitches for batting & bowling. So who were the most effective bowlers on roads (McGrath is up there, no?), and who were the batsmen who stood up on the raging seamers/raging turners?
matth
Roar Guru
Interesting stuff. One unfortunate problem, and I have no idea how to solve it, is that you are using the bowlers’ general bowling averages when deciding the batsmen’s premium and semi-premium averages. But what if the bowlers in question have a significantly worse “premium bowling average” and basically dismissed bunnies? Are they still premium even with their sub 29 average? The problem is that this becomes circular so I’ll just shut up now and say this was very interesting. Except one more thing. Do more spin bowlers average over 30 overall? Does this skew the model in some way for players who spend more time batting in spin friendly conditions? No idea :silly:
The Late News
Roar Rookie
Chip...very interesting. thanks.
Chip
Roar Guru
Thanks- that is useful
Chip
Roar Guru
Thanks !
Chip
Roar Guru
Thanks- yes I have in mind to look at the reverse at some stage
Once Upon a Time on the Roar
Roar Guru
I suppose 1999 in the Caribbean and Adelaide 2000 and 2005 helped him there. McGrath wasn't playing in India 1998 nor when they came here in 2003-04.
All day Roseville all day
Roar Guru
Narayanan states that Lara averaged 45.00 against McGrath.
Once Upon a Time on the Roar
Roar Guru
What did Lara average against McGrath?
All day Roseville all day
Roar Guru
Hi Chip, Direct head-to-head match-ups seem to be a thing in T20 tactics nowadays. For Test cricket, Charles Davis and Anantha Narayanan are the oracles. To save you trawling through their articles, here's a few figures- McGrath averaged 22.6 against all top-order batsmen, followed by Ambrose with 22.9. Bradman averaged only 49 against Verity, who dismissed him 8 times. Tendulkar averaged a massive 107 against Warne, 83 against Johnson, 60 against Vettori, 55 against Lee, 48 against Murali, 35 against Donald, 24 against Gillespie, 23 against Anderson, and only 14 against McGrath. Lara averaged 124 against Murali, 110 against MacGill, and 71 against Warne, Which clearly suggests that Lara was better than Tendulkar, and McGrath was superior to Warne. Dravid scored 223 runs against MacGill, without being dismissed.
Colin Turner
Guest
Always difficult to compare across different era’s. Changes over the years have vastly favoured recent batting statistics. Some examples: Boundary ropes, massive sight screens, helmets, fast outfields, protected pitches, drop-in pitches, bat technology etc. Imagine facing a West Indies pace attack without a helmet, on a post AFL wet season Melbourne pitch resembling a sheet of corrugated iron, and an outfield resembling a cow paddock, no boundary rope and a tiny sight screen by todays standard. How do you compare the batting performance then to now, when the MCG has been resurfaced with modern drainage, pitches are drop-in to avoid the AFL carnage of olden days, sight screens are gigantic, boundaries and sixes are much easier to score with boundary ropes, and crowd behaviour is so much better. I know which era I would prefer to bat in.
Once Upon a Time on the Roar
Roar Guru
I know that when Brian Lara made a run a ball unbeaten 197 in a county match for Warwickshire against Northamptonshire in 1994, he scored 12 runs from the 43 balls he faced from Curtly Ambrose.
Peter85
Roar Rookie
While I don't know where to find the data publicly, I understand that with ball by ball statistics most of what you are proposing can be achieved. I remember seeing on the coverage a batsmans particular average against a bowler and type of bowler (left arm spin etc). If you had the data set you could conceivable use a bowlers ICC ranking/rating at the beginning of a test match to assess a batsman against strength of bowler - maybe this is partly how the ICC ranking already works. With any statistic, the more granular you look, the noisier the data is due to sample size. I also have no idea how to deal with not-outs in calculating how good a batsman is. It is unfair to call it a completed innings because their team mates put them in a position to not complete an innings (declaration, winning a match or all-out) and I feel that the times a batsman is dismissed chasing runs probably offsets the not outs. Does the reverse also need to be considered? A bowlers average against strength of batsman?
Clay
Roar Pro
Poor Norman Yardley, being supplanted by old mate Bruce. Wouldn't have really called him a great of English cricket either, but a good presence in those early post WWII series. I like the analysis here. Doing it for the likes of Tendulkar vs Ponting would shed some extra light on their careers as well. I wonder what the bowling equivalent would be - averages between 20-40 for semi premium and over 40 for premium? Worth consideration.