The numbers game

Why the most famous idea in sports management stopped working

A The insight that produced the most discussed management story in modern sport was not about athletes. It was about prices. A baseball club with a fraction of its rivals' resources could not compete for the players everyone agreed were good, so its management asked a different question: which measurable qualities actually produce wins, and which of those are the market not paying for? The answer, in the late 1990s, was that the ability to avoid making an out — including by drawing a walk, which requires no athleticism, produces no highlight and had never been valued — contributed more to scoring than several of the qualities scouts were trained to admire, and could be bought at a discount because nobody else wanted it.

B The intellectual groundwork had been laid two decades earlier by an amateur. A night watchman at a food-canning plant began publishing annual pamphlets analysing baseball statistics, arguing that most of the numbers the sport collected measured the wrong things and that better ones could be constructed from the records already kept. He wrote with unusual clarity and no institutional position whatever, and was ignored by the sport for years while acquiring a substantial readership outside it. Almost every idea that later became famous was in those pamphlets first.

C His method was not statistical sophistication. It was the insistence that every conventional measure be asked what question it answered, and the discovery that most of them answered a question nobody had meant to ask. A statistic recording how often a fielder reaches a ball he is close to rewards the fielder who is out of position often enough to have distance to make up. A measure of how frequently a batter hits safely ignores the occasions on which he reaches base without hitting at all. Once stated, the objections are obvious, which is a property of most durable ideas.

D What made baseball unusually suitable for this treatment is a feature of the game rather than of the analysis. Baseball is a sequence of discrete confrontations between one pitcher and one batter, with the other players mostly stationary, and the outcome of each is recorded. Credit and blame can therefore be assigned to individuals with a confidence impossible in sports where ten players move continuously and the contribution of any one of them is entangled with the rest. The statistical revolution began in baseball because baseball had already been recording the necessary data for a century without knowing what to do with it.

E The famous season is generally misremembered. The club won a great many consecutive games and reached the play-offs on a payroll near the bottom of the league, and the account of it published afterwards emphasised the acquisition of undervalued hitters. It gave rather less prominence to the three exceptional starting pitchers the club had drafted conventionally and paid very little, whose contribution was arguably larger. This is a common shape for such stories: a genuine insight, correctly identified, is credited with more of the outcome than it produced. The club's own management has been consistent about this in interviews since, and consistently disbelieved, because the simpler version is a better story and the people who tell it have no stake in the correction.

F The deeper problem with the approach is that it destroys the conditions for its own success. An inefficiency in a market persists only while it is unrecognised, and the publication of a best-selling book explaining exactly which quality was underpriced ensured that within a few seasons every club employed analysts and the discount had gone. The wealthy clubs, having more money and now the same information, were if anything strengthened. What began as a method by which a poor club could compete became a standard cost of doing business that the rich could better afford.

G The migration of these methods into other sports has been slower and more revealing. Football presented the harder case: goals are rare, possession is continuous, and a defender who prevents a chance from developing leaves no trace in any conventional record. Progress came only when clubs began paying for data on what happens away from the ball — every player's position several times a second — which permitted the estimation of how much any given action changed the probability of scoring. Clubs that adopted this early achieved results disproportionate to their spending, and the sequence has repeated: an advantage that lasted while it was unusual and diminished as it spread. There is a further complication peculiar to the sport. In baseball the analysis identifies which players to buy; in football it more often identifies how a team should play, and a manager cannot implement a plan the players cannot execute. The gap between knowing what would be optimal and getting eleven people to do it is where much of the theoretical advantage is lost.

H The reasonable conclusion is not that the analysis was overrated but that its benefits were misdescribed from the beginning. Better measurement does not confer a permanent advantage on anyone; it raises the standard of decision-making across an industry and redistributes the advantage back towards whoever has the most money, which is where it started. The clubs that gained were those that were early, and being early is not a strategy that can be repeated. What remains is a sport that understands itself better than it did, which is worth something, and a lesson about competitive advantage that applies well beyond games.