It is a constant debate in cricket – who is the best batsman of all time/1990s/2000s etc? Similarly for bowlers as well. Given the question and the interest it generates, how do we measure performance> The standard measure is averages which experts believe is unfair as people sho stay not out tend to have higher averages. This measure is unfair especially to openers who face the most hostile bowling conditions and have much greater possibilities of getting out. For an economics student, similarity to discussions on GDP is too obvious.
Anantha Narayana looks at the measures and suggests runs per test is a pretty useful indicator as well.
There are quite a few measures used while analysing Test batsmen’s performances. Let us first get out of the way the measures such as runs and centuries scored, which are basically accumulated over a long number of years and Tests. There is a lot to admire in a batsman who has played over 150 Tests, but then we must accept that he is likely to accumulate over 12,000 runs. Scoring rate is a non-starter since it is not available for over two-thirds of the matches.
The Batting Average is the most popular one but has a few shortcomings. The variations in averages because of not-outs is the major one. A small analysis highlights this problem. In batting positions 1-2, there are 560 not-outs (4.5%) out of 12,558 innings. In batting positions 3-5, there are 1319 not-outs (7.3%) out of 18,107 innings. In batting positions 6-8, there are 1822 not-outs (11.0%) out of 16,603 innings. That is one heck of a variation. Intrinsically this measure is unfair to the openers.
Runs per Innings (RpI) is quite good. However, the problem of low not-out scores remains. I have analysed this extensively and come out with solutions fine-tuning the innings count with adjustments based on the size of innings. These work very well. However, it has to be accepted that these are only tweaks to iron out a basic problem. This measure is unfair to middle-order batsmen.
That leaves us with the Runs per Test (RpT) measure. When we sit down and looks at all these measures we can easily see that the RpT measure is an excellent one. The benefits are outlined below.
1. The Test is the most important playing unit in Test cricket. An innings is only a means to an end – to win the Test, and if not possible, to draw the Test. As such it is correct to measure the performance by players in one Test. Without any doubt, the batting and bowling performances in the Test go towards getting a favourable result.
2. The batsman is allowed to make up for failure in the first innings with a compensating performance in the second.
3. Alternatively, if a batsman scores reasonably well in the first innings, he would not mind failing in the second innings, maybe with a specific objective of speeding up the scoring.
4. The effect of the not-out and batting late in the innings are neutralised, to a certain extent.
Some readers may ask: Why not go for a series as the unit? The fundamental problem is that a series is of completely uncertain length. It could vary from one to six Tests. It is played on different grounds and along with and against differing sets of players.
Let me clarify that I am not advocating the replacement of the Batting Average with RpT. The average is an excellent measure, ingrained in our collective Test-cricket-following mindsets. Many of us have spent decades working with the average. I am only suggesting that the RpT measure is an underrated and under-used one and should be given more exposure.
Based on this, he ranks batters and there are some obvious players entering and some surprising ones..
There is lot more in the article. I wish statistics was indeed taught using cricket examples. Would have been so much easier to grasp the fundas..