For over a month, from 18 June to 19 July, Zack Greinke of the Los Angeles Dodgers didn’t allow a single run. He started six games in that span, finished 43 2/3 innings, struck out 42 batters and allowed just 24 baserunners. Opposing hitters managed just a .129 batting average and a .158 on-base percentage. He lasted two more scoreless innings into his next start on 26 July before allowing a run to close the streak at 45 2/3 innings, the fourth-longest such streak since 1968. Greinke wasn’t perfect, wasn’t unhittable – his team even lost one of those games, 1-0 to the Cubs on 23 June – but he pitched about as well as humanly possible.
How did he do it? How has he been so good all year, as the owner of a 1.71 ERA in over 150 innings, threatening to become the first player to post an ERA under 2.00 in a full season since Pedro Martinez did it in 2000? Streaks Like Zack Greinke’s Require A Ton Of Luck, reads a headline on FiveThirtyEight.com. Truth #1: Of course Zack Greinke has been a little lucky, reads a header in a piece by Grant Brisbee at SBNation. “Greinke is clearly having a good year, but a fair amount of luck has contributed to his historic start to 2015,” Joshua Sadlock of BaseballEssential.com wrote in July, as Greinke was in the midst of his historic streak. Jeff Sullivan at FanGraphs wrote, “Zack Greinke has had some luck. Sorry, but, duh. It’s always there. Zack Greinke’s true talent isn’t a 0.00 ERA.
Such claims have become requirements in certain corners of the baseball writing world, a box that must be checked before we can be assured the author knows what they’re talking about. Those claims certainly appear in a copious amount of my posts at FanGraphs, where I wrote regularly from 2010 through 2013. The concept of luck has become so overused as explanation by those who would read the statistical tea leaves of sport as to become a point of ideology. Can we explain it? No? Then it’s luck.
This view of the world might make sense for someone with a singular focus on predicting the future, but when it becomes a universal explanation for why predictions fail, it has a stifling effect on learning, understanding and growing our knowledge. Chance and random variation will always have their place in games and sports. I wouldn’t dare suggest Zack Greinke willed this ball into Alberto Callaspo’s glove, for instance, or that this fan interference call involving a baby and a tarp is a predictable baseball event. But I think reducing the concept of luck to simple chance – or, more accurately, what we perceive as simple chance – is a severe misunderstanding of how charged and loaded the word “luck” is, particularly in the world of sports.
“The belief in luck,” economist Thorstein Veblen wrote in his 1899 work Theory of the Leisure Class, “is a sense of fortuitous necessity in the sequence of phenomena.” In other words, to believe in luck is to deeply believe that things are supposed to happen a certain order, that there is a way to the world. Luck, if you’re a believer, is the force that creates the inevitable deviations between the real world and what’s supposed to happen. Luck resolves the conflict when you don’t know why things didn’t go your way.
Evidence of this kind of belief in luck is apparent throughout the sports world. “There are few sporting men,” Veblen writes, “who are not in the habit of wearing charms or talismans to which more or less efficacy is felt to belong.” It’s still true 116 years later. Whether it’s the lucky underwear (or maybe the garter belt under the baseball pants) or the lucky “aqua-metals technology,” athletes to this day continue to attempt to court the forces of luck however they can.
Of course, this kind of behavior is far from limited to athletes. It is at least as common, and probably more prevalent, among gamblers. The Lucky Mojo Curio Co in California, just as one example, offers an incredible variety of charms, objects, books and spells meant to enhance the gambling spirit. “Gamblers – whether they play at cards, bingo, horse racing, policy, the lottery, or other ‘numbers’ games – tend to want a winning edge. For this reason they often enhance their personal power through the use of amulets, charms, and ritual spells to bring favour to their enterprises,” the website claims. The available “lucky charms and spells” range from Alkanet Root Bark (“To prevent losers from jinxing you or your winnings) to Gamblers’ Gold Lucky 7 Hand Wash (“An herbal wash to cleanse the hands for play.”) to – seriously – Raccoon Penis Bone (“Wrapped in a $20 bill to keep on winning”).
Anything to bring the forces of luck to our side. Even the act of betting can be seen as an attempt to put our own weight behind the outcome we desire most. “The wager is commonly laid also with a view,” Veblen writes, “to enhancing the chances of success for the contestant on which it is laid.” This holds true even in the coldly logical sabermetric community. In 2014, the Baseball Prospectus podcast Effectively Wild held a season preview series in which one guest was brought on for each team, in most cases a blogger or writer with some connection to the team (including myself; I created the Milwaukee Brewers blog Disciples of Uecker and previewed that club for the series). At the end of each show, guests were asked to predict the win total for the club they previewed.
The results, as compiled by Jeffrey A Friedman, confirm Veblen’s observations. The average win total for these predictions was 84, against a necessary average of 81 in Major League Baseball’s 162-game season. The average predictor overrated his or her team by 3.2 games, and only four forecasters had a worse prediction for their club than Baseball Prospectus’s house projection system, Pecota. I was not immune, as I picked the Brewers to win 82 games against a projection of 80. It’s particularly striking to me that this clear bias would appear even within the sabermetric community, because in many ways, our predictions are bets on ourselves – bets on our own knowledge, on our ability to read the statistics and divine the future. Even people like us, supposedly drawing our authority from our ability to predict the future, are so subject to such biases that we make predictions that cannot possibly come true.
When the baseball community lines up in solidarity to say Zack Greinke was lucky, what they’re saying is what happened in reality wasn’t what was supposed to have happened. It’s throwing our collective weight instead behind Greinke’s 3.41 career ERA, or his 2.94 projected ERA by the ZiPS projection system, or his 2.97 projected ERA by Pecota as Greinke’s correct marks, as his “true talent” level. But to believe each player has a single number that serves as their true talent level creates inevitable conflict whenever an outlier streak like Greinke’s occurs. Even one great start or game can make our previous conception of a player’s true talent look ridiculous. And that’s when we look to luck to fill the gap.
I find this response substantially lacking. It might make sense for the fantasy baseball player, the gambler, or the general manager, people who have to make decisions based on limited information and have to rely on the performance of others – and, of course, some random chance – in order to deliver their results. Just because correctly predicting streaks like Greinke’s requires a substantial amount of luck – because we don’t have a statistical method capable of doing so – doesn’t mean those streaks are the products of luck. It only means we don’t know how to predict them.
I don’t know exactly how to explain Greinke’s streak. But I think describing it as luck reveals a perspective that sees the general manager – a baseball organization’s predictor and decider – and not the player as the prime mover of the game. I think there was something different about how Greinke pitched during this stretch. It may not be repeatable or sustainable or any of the words that matter to somebody whose job it is to predict the future. But this is not the only application of statistics and sabermetrics.
Sabermetric thought, at its best, is engaged in a constant deconstruction and reconstruction of how we see the world. This is how we come to the realization that on-base percentage is a better indicator of hitting prowess than batting average, how we begin to understand that the mid-range basketball shot is inefficient and the three-pointer is the future, how we finally see that the NFL’s conventional wisdom on fourth down strategy is anything but wise. Relying on luck to fill our gaps in understanding instead leads to stagnation, in which ideas harden into ideology and dogma. Removing luck as a crutch may mean we have to admit our priors are wrong or incomplete, that our predictions will always be flawed, and that nothing in sports is supposed to happen, no matter how hard we believe. It means opening the mind to other explanations, other conjectures to build off and eventually tear down and start the whole cycle over again.
Being wrong is hard. But it is also inevitable, particularly in the sharply human world of sports. Every single leap forward in our understanding of sports has first come with the acknowledgment that our priors were wrong – that we don’t know everything about the game, that the world is not, in fact, flat. And with every game, every season, and every hot streak comes a new set of priors that must be assimilated into the worldview, a new set of things we were wrong about.
Luck is much, much easier than acknowledging all of that. But if we want our predictions to be less wrong – or if we just want to know more about the games we love to play and watch – we need to stop artificially filling the gaps between those predictions and the realities with which they come into conflict, no matter how loud our sense of a fortuitous necessity in the sequence of sporting phenomena is screaming otherwise.
