February 28, 2017

It might be time to undo some of our theories about the game

Michael Lewis' new book suggests we often make judgements based on flawed reasoning. Let's learn from it
14

In 2013, who would have foreseen that in four years Steven Smith (left) would be Australia's captain and one of the world's all-time leading batsmen and Shane Watson would be retired with only four Test hundreds? © Getty Images

As of this week, Steven Smith is the sixth-highest-ranked Test batsman of all time (in terms of highest rating achieved), and the man in possession of the sixth-best average in the history of Test cricket (having batted more times than everyone above him in the list). He makes a century once every 5.2 innings.

It is all a long way from his first Test in 2010, when he batted No. 8, and also from England in 2013, when, 11 Test matches into his career and after 22 innings, he had an average of 29.52 and a highest score of 92.

In his next match, at The Oval, in the dead rubber of a thankless tour, he made a century, much of it in partnership with another enigma, Shane Watson, who scored his third and highest Test hundred, 176. Watson had gone more than 45 innings without a ton until that knock, and his average, after 83 innings as a Test batsman, stood at 34.51.

An impartial observer beamed down from planet Zarg might have struggled to tell which of the two players might go on to hit the heights that Smith has hit. The stats would not enlighten him - there was nothing in either player's numbers that suggested such a surge. Perhaps they might have opted for Watson simply because he looked more like a successful batsman - technically correct, powerful, domineering. Smith had an oddball method that reinforced the perception that he might struggle for consistency against the world's best bowlers.

None of which is intended to decry Smith, rather to show how difficult it can be to judge a player's potential. I thought about this while reading Michael Lewis' remarkable new book, The Undoing Project. Lewis is best known (certainly in cricket) as the author of Moneyball, a book that sent every sport looking for its hidden statistical edges.

Lewis writes in his introduction to The Undoing Project about a review that Moneyball received when it was first published: "… the author did not seem to realise the deeper reasons for the inefficiency in the market for baseball players: they sprang directly from the inner workings of the human mind."

We are an age beyond the first wave of statistical analysis brought on by Moneyball. Some of the game has been undressed, but a deeper relationship between the immutability of numbers and more refined but fallible opinions of coaches and players exists

Intrigued, Lewis decided to find out what those workings might be. It was a short journey to the story of Daniel Kahneman and Amos Tversky, the Israeli psychologists whose "breathtakingly original" studies revealed that human beings made predictable systematic mistakes in judgement again and again.

Lewis has written a wonderful book about the friendship between these two amazing men - almost every chapter has one or two revelatory moments. Some of their early work was deconstructing a study of pilots in the Israeli air force. During training exercises, fliers were divided into two groups. If pilots in one group made an error during a training flight, they were criticised by the instructors. In the other, they were encouraged. It found that those who were criticised responded better and flew well the next day - something the study made great play of.

Kahneman realised that neither praise nor criticism was affecting the pilots. What was happening was that they were simply returning to their "mean" performance - that is, if they flew badly one day, they were more likely, statistically, to fly better the next, whatever anyone said to them. It was a natural variation that the study hadn't allowed for. It's fun to think of the implication for the over-coaching of players in the light of that information, and also to perceive more deeply what "form" in cricket might really be.

Kahneman and Tversky discovered more anomalies in judgement. They called these "heuristics". There was the availability heuristic, where reasoning was skewed by the most easily recalled piece of information (a good one for the psychology of batting collapses); the anchoring heuristic, a reliance on the first piece of information offered (batting or bowling average, perhaps); and the representative heuristic, making judgements about uncertain things by their similarity to standard procedures ("this wicket looks like it will turn"). These cognitive biases, defined in the 1970s, went on to have far-reaching implications in economics, healthcare and the military. The archaic world of sport took longer to catch on.

The Undoing Project is way richer than those few examples, and the reasoning is much more subtle and useful. Post Moneyball, Lewis takes these concepts and shows how human judgement is subverted, controlled and directed, especially by existing paradigms and conventional wisdom.

There's nothing in the book about cricket, and yet it feels essential. We are an age beyond the first wave of statistical analysis brought on by Moneyball. Some of the game has been undressed, but a deeper relationship between the immutability of numbers and more refined but fallible opinions of coaches and players exists.

The organisations that surround international teams and franchises are trying to provide some kind of certainty of judgement, and through it control the inherently uncertain world of professional sport. In knowing better how and why their decisions can be wrong, by studying the actual decision-making process rather than its outcomes, they can perhaps gain a greater understanding of the game and the people that play it. Lewis has done it again, but bigger and better this time.

Jon Hotten blogs at The Old Batsman. @theoldbatsman

Comments have now been closed for this article

  • Peter on March 3, 2017, 10:00 GMT

    I can still remember vividly, as a 7 year old at my first cricket practice, being told to pick up a cricket bat "like an axe", as if chopping wood. Thank goodness the game has moved on since then!

  • Chris on March 2, 2017, 22:48 GMT

    It was easy to predict simply by taking into account Smith's shield performances whose FC average was well above 50 at a time when almost all other shield players averaged 30-45. I backed him at the time while everyone was complaining about how his strange technique was not up to test standard. Warner was the same. Also the same reason why it was not a surprise that O'Keefe with his low shield average single-handedly won Australia their first test on Indian soil in over a decade. O'Keefe's test average now mirrors his outstanding FC average. If the book talks about moneyball but fails to talk about the pros and cons of stats in the shield competition then the author missed the point and it's probably a good one to skip. Otherwise sounds interesting.

  • Clifford on March 2, 2017, 12:33 GMT

    My favorite rhubarb in cricket is picking players on looks, as in "he's such a classy looking batsman" or "he has such an elegant bowling action"...

  • Thomas on March 2, 2017, 3:21 GMT

    If anyone's style of batting sums up the phrase: "Do what works best for you," it is probably Steven Smith. Not really pretty to watch or textbook in approach but usually finds a way to get the job done.

  • rob on March 2, 2017, 2:08 GMT

    God knows I'm a complete dummy when it comes to stats but there's something in here that I'm really struggling to understand. It says that the pilots who were criticised generally flew better on the next outing but then it goes on to say that it had nothing to do with what others said to them, it was just that they returned to their mean performance levels. .. The only way that makes any sense to me is if the group that was criticised were naturally better pilots than the other group. If what was said to them truly made no difference, wouldn't both groups tend to fly better the next day? .. More than that, if the groups were not evenly skilled in the first place, what was the point of the study.

  • ravi.narla on March 1, 2017, 13:17 GMT

    @PYOLAB, Spot on. Not only that, many a times there is always something to say when you are in position to convey those kind of messages. It is just that people want to have a say, be it their kid or a pupil or a coach. The best coaches are the ones that coach less and make subtle changes to the technique only when required. If you are player dont go to the coach. Figure it out yourself. The ones that are self taught are the ones that come out on top. It takes time and effort but the best way. IMO Coach is just a manual on what not to do or not in a situation, basically a mentor for your mental aspect of the game. Bradman learnt most of the things on his own and so was Akram. Leave the players alone they will be better off which only coaches that were former players understand it well especially with the Indian team.

  • John on March 1, 2017, 13:01 GMT

    Never run to a mis-field they used to say. Now everyone runs for mis-fields.

  • Adam on March 1, 2017, 12:21 GMT

    In cricket more than any other sport, "wisdoms" are passed on from coach to player, and from commentator to viewer, who then passes on those same insights to the players they coach, or to the viewers the commentate for.

    Unfortunately, most of these wisdoms are complete nonsense, but they've been around long enough that no-one questions them, even though they are often directly contradictory to what is obviously visible on the screen. "Never hit against the spin", "Change the angle by going around the wicket". "Batsman should never play front-on". All rubbish, yet how often do you hear them?

  • Frank on March 1, 2017, 12:19 GMT

    I thought the same as you at first Layerroadboy. Upon re-reading though you see the first conclusion was made by the original study and the second conclusion was made by Kahneman and Tversky after their deconstruction of the original study. So yes the two conclusions conflict but this is not a contradiction within one piece of work if you see what I mean.

    I agree about Smith, I was one who thought Smith, Phil Hughes and Watson would all end up with very strong test batting records, when the consensus seemed to be none of them would, due to technical deficiencies. I have to admit defeat on Watson, claim victory on Smith and Hughes, sadly we will never know.

  • pete on March 1, 2017, 10:06 GMT

    Interesting, and a reinforcement of much that we instinctively 'know' about stats ... but perhaps more interesting is the impression given of novelty (if the 70s can be accused of such). let's not fall prey to the belief that these types of cognitive bias are recent discoveries. The first two 'heuristics' quoted here sound remarkably like the Primary and Recency effects beloved by psychologists and go back to the fifties, while the third sounds an awful lot like confirmation bias, and this can be traced to the early seventeenth century at least: 'The human intellect is not a dry light but is contaminated by the will and affections [...] For man would rather believe what he wishes to be true' (Francis Bacon). To the best of my knowledge, however, Bacon wasn't a keen cricketer.

  • No featured comments at the moment.