Criconomics May 2, 2015

When the best is not the greatest

A new book brings science to bear on the cricket ratings game - and largely succeeds

Yes, Viv Richards is found to be head and shoulders above everyone else on the batting front © Getty Images

Which has been the most dominant ODI team ever? Who has been the best, and the greatest (the two terms mean different things in this book, which is addressed later) ODI batsman and bowler ever? Criconomics takes on these weighty topics and does a fair job of making sense of them. There are some contentious outcomes, but it is in the very nature of such exercises to produce those: lists that proclaim to rank the best or greatest are invariably provocative, without necessarily meaning to be so.

To reach a wider audience, a book of this nature, which is numbers-driven and based on a reasonably complex rating methodology, must avoid certain pitfalls. Too many pages of complicated numbers-based explanations can lose the readers. And yet, since the book is all about ratings, the numbers and stats need to be at the forefront, with explanations for what seem like anomalies, or even limitations of the methodology. The book does a good job on the first count, but it could have done better on the second.

The introduction to the concepts on which the ratings are based is pretty crisp. There's the pitch index, which gives a value to the degree of difficulty of run-scoring at a venue, based on past results. Each team has a batting, bowling and team index based on past performances - to which the latest scores are added, to give an up-to-date new value of team strength.

The authors, Surjit S Bhalla and Ankur Choudhary, introduce us to the Kalman filter, an econometric algorithm that is used in rocket science for real-time course correction. In the more humble realm of cricket, it has been used to scale the batting value of a team, and the bowling value of the opposing team, up or down, based on every new performance. Since these ratings start from the first ODI ever played by each team, they're built organically, taking into account each performance, with recent matches weighted higher. These team numbers (batting and bowling indices, and the team index) are then used to not only rank teams and predict scores and results but also to evaluate batting and bowling performances against teams.

When comparing players, Bhalla and Choudhary have, quite appropriately, differentiated between best (the highest overall score), and the greatest (the most dominant over the longest period of time)

Justifiably, the authors play up the robustness and objectivity of their system: "What has been set up is a completely objective framework that leaves no room for post-fact subjective elements. Untouched by human hands would be an apt description."

They also put plenty of emphasis on the ability of the methodology to predict results. ("The art of prediction is important and not just for entertainment, because it gives us a way to evaluate the evaluators.") Bhalla and Choudhary's CricketX model has an accuracy of 65% in predicting the result in all ODIs played between the top ten teams. It is up for debate how good that number is, but the authors try to play it up by pointing out that the ODI format inherently tends to produce upsets. Bangladesh's win percent of 23 against the other nine teams is presented in this regard. "In how many of those matches would you have put your money on Bangladesh? Close to none." (The authors forget, though, that a huge chunk of those wins came against Zimbabwe after 2004, when Bangladesh would have been favourites in most match-ups between the two sides. Excluding matches against Zimbabwe, Bangladesh's win percentage drops to around 13%.) There's an interesting section, though, comparing Australia's all-conquering team of the 2000s and West Indies of the 1970s and '80s; Australia come out on top, but only just.

When comparing players, Bhalla and Choudhary have also, quite appropriately, differentiated between best (the highest overall score), and the greatest (the most dominant over the longest period of time). Viv Richards is way ahead of everyone else among batsmen on both parameters - as indeed he is in any ODI rating exercise - but the bowling numbers aren't so clear-cut: Joel Garner is the best but Glenn McGrath the greatest. Similarly Kapil Dev emerges as the best allrounder, but Richard Hadlee as the greatest. And there's a section on captains, where, quite cleverly, captains are ranked by the number of "extra" wins they fetched their teams, compared to the expected number given the overall strength of the team and the oppositions they played.

Throughout, the basic concepts and the thinking behind them have been explained clearly, which gives readers a good idea of what the authors are attempting to do.

A few interesting tidbits add flavour to the book: there's Don Bradman's handwritten letter in response to Bhalla's first book, Between the Wickets: The Who and Why of the Best in Cricket; John Arlott's review of that book; a quick, back-of-the-hand calculator to estimate during a team's innings what their final score will be; and a method to guess whether a chase will be successful or not.

However, these nuggets are overpowered by other, less insightful, passages that move away from numbers but do little to add value to the reading experience. The section on bowlers, for example, starts with three pages on why bowlers were traditionally the lesser cousins of batsmen, and which bowlers have tried to redress the balance; it has no relevance to the analysis that follows later. Similarly, the section on captaincy has, quite inexplicably, a long paragraph on the Mike Gatting-Shakoor Rana fiasco, sourced from Wikipedia. There are several other passages from Wikipedia that could have easily been omitted. There's also a three-page recommendation for the future of the game, which can be dispensed with.

Much of that space could have been used for further insights on the methodology and ratings. How does Kapil find a place as the ninth-greatest ODI batsman of all time, ahead of, among others, Adam Gilchrist, Mark Waugh and Javed Miandad? What makes Lonwabo Tsotsobe the second best ODI bowler of all time, next only to Garner? And why is James Anderson the best among current ODI bowlers? (It doesn't help that he is at one point referred to as "Jim" Anderson.) The answers to all these questions would have made for interesting reading; instead, there's an unnecessary reference to two successful offspinners having questionable bowling actions. ("Is it in the nature of competition that very successful offspinners are questioned for their action, or is it that a questionable action makes the bowlers better?")

The writing isn't the most consistent, but the book is interesting for the concepts and stats it puts forward, and for that alone it's worth a read.

Criconomics: Everything You Wanted to Know about ODI Cricket and More
By Surjit S Bhalla and Ankur Choudhary
Rupa Publications
216 pages

S Rajesh is stats editor of ESPNcricinfo. @rajeshstats