The case for a non-systematic understanding of cricket

Sport is designed to be a complex system

Beginning to write this article gives me a weighty feeling. It's likely that many of my readers on this Substack are fans of data-oriented cricket analysis. If you've ever followed my writing, you know that I am one too. Nevertheless, the "limits" of data-oriented cricket analysis do not elude me. These limits, and the consequent case for a non-systematic understanding of cricket, are what I plan to explore in this short piece. Thus I ask: why might a non-systematic analysis of cricket be better than a systematic one?

Unfortunately, this is an all-too-familiar question for me. I study economics, which is the most systemized of the social sciences. Keep in mind that I say this with all the wisdom and credentials of an opinionated grad student, but it has long been my understanding that all the social sciences and the humanities fundamentally seek to study the same thing - society - but it is in their approach towards this task that they diverge. Take a field like sociology, which argues that it is by understanding subjective experience that you best understand society. In literature, the contention is that stories and works of art comprise the most fascinating kaleidoscope of society.

The claim in economics is that the best way to understand society is through brute systemization; that by filtering out the minor and unimportant details and by studying what we are left with we put our best foot forward in trying to understand what the heck is going on. This, I think, is what has united economists throughout history, not that they study the economy (paradoxically) or that they study tradeoffs (as some elementary textbooks say). Consider, after all, that economists today work in fields as wide-ranging as criminal law and evolutionary biology.

So that's the claim. But is it true? As an economics person and a believer in the economics method in general, I am obliged to argue that it is. But my liberal-arts background has had the effect of sensitizing me towards the possibility that it may not be so. (More on that later.) The point of smothering you with this academic guff is that at the end of the day what cricket analysis aims to do is also precisely this: systemization. It is systematic to record the actors constituting every ball and the outcomes those actors produce. It is systematic to analyze this data and to generate conclusions and to assume that every ball is almost equally significant. Modelling, and data, are the language of systemization. A few examples of non-systematic ways of understanding sport might help make this distinction clear: the highlights package, which shows you only the ‘excellent’ and ‘terrible’ outcomes are an obvious example, as are match reports and color features which ruminate over the deciding final few seconds of a game.

The stage is now set for my case for non-systematic analyses. It is this. Some systems are unfortunately more complex than others. People can debate all they want about why the physical sciences are in better shape than the social sciences, but part of the answer is likely that individuals are more complex to study than objects. Take our understanding of incentives. All human beings are born with a natural goodness of heart? But hold on, this vanishes when we put them in a market. So are incentives good or are they bad? How must one word incentives in order to minimize the crowding out of prosocial preferences? Perhaps one must focus on what messages they deliver. But what if different messages are interpreted by different readers? Will a rephrased incentive be read as an appeal to one's care/harm moral foundation or as a dishonest, underhanded tactic, thereby triggering selfish behavior? In short: it's a mess.

Sport is designed to be a complex system. It's why it entertains grown men who dare not cry. Sport is the drama of life taken and magnified a thousand times over so that the tales of rise and fall and redemption which usually flow through the history of a person's lifetime can be told within the span of a few hours. It's why even when all seems lost on the cricket field, people hope; it's why even after soul-crushing, mind-numbing defeats, players and fans return. Anything can happen on the field. Cricket, that funny old game.

Is the science of cricket too complex to be discerned entirely through brute systemization? I don't know. But I am cognizant that it might be, even though I hope that it isn't. For this can create at least two troubles. Firstly, it renders simplification useless. Simplified models are useful because they take a real-world phenomenon and take the fluff out of it so we can study it for its most important constituents. But there is a difference between a system whose ten most important explanatory variables explain 50% of the variation in its outcome and a system whose ten most important variables explain 90% of the outcome. Simplification can claim to not have eliminated crucial detail only in the latter.

Now suppose you throw in all the variables you can find irrespective of rank of importance. This introduces the possibility of overfitting. Overfitting occurs when a model learns the noise and random fluctuations in the training data rather than actual underlying patterns. Your model may fit the training data well but may throw up increasingly silly results as you test it on unseen data. This is the curse of complex systems: you cannot oversimplify them because even their least important constituents are significant, and you cannot overspecify them because overspecified models tend to try to explain even random variations. Some systems are unfortunately too complex to systematically understand.

So the question becomes: is our beautiful game really that complicated? I believe that it probably isn't. But then again, just as my sociologist friends berate me for my dogmatic support of the economics methodology, I am obliged to say this because I want to believe that a systematic analysis of sport still has value. It is not that I am convinced fully of the premises underlying this claim, but that my brain is wired in the kind of way that makes me like systemization. I wake up and I systemize - it's how I get around day-to-day tasks. Others don't. Perhaps they don't share the same belief and conviction as I do in the power of rigorous analysis, and that's understandable to me.

Part of the motivation to write this article stems from the observation I made in my last article, that modern analysis often lacks humility, which causes it to receive ridicule and antagonism from cricketers. But it is also much more. I think it is beautiful that different people can view sports in different ways, guided ultimately by their internal likes and biases, and that each of those perspectives may have value. I think it is beautiful that no matter where we look viewpoint diversity is at the heart of learning.

One of the best things I have learned in life is to say, "I don't know", because the need for certainty spawns delusion. I don't know whether my deeply held belief that systemization is the way to understand cricket - and indeed the world - is true or false. But I shall believe it and argue fervently in its favor because it is the path I have chosen to walk. At the same time, I shall remain aware: aware of my limitations as a walker of this path, aware of the road not taken that was perhaps grassy and wanted wear. For what is the point of trying to understand the human condition if this does not cultivate empathy?