In 2003 a book was published which completely changed analysis. That book is Moneyball, or, Moneyball: The Art of Winning and Unfair Game to give it its full title. Moneyball has become such a phenomenon that Mike Hughes, former England and British and Irish Lions analyst, and I, when recording the Running the Numbers podcast, frequently joke about a drinking game for each time someone mentions Moneyball. It averages out at once an episode. The Simpsons have even parodied it in their episode MoneyBART. But is Moneyball all it’s cracked up to be? This long read will investigate.
Type Moneyball and Management in Google and you get 703,000 results. Moneyball, as a concept, has got a bit silly. If any football manager talks about their analysis department then some newspaper or website will mention how they are playing Moneyball or Moneyballing the system. Thanks to the film it has passed into popular lexicon. Non-analytically inclined fans will talk about how rugby is being Moneyballed because of all the modern statistics that we have. This does a disservice to both Michael Lewis’ book and the general concept. Moneyball isn’t about using statistics in sport. At least, it’s not as simple as that.
Statistics were used in baseball from the year it began. Every Major League game since its inception, and many of the unofficial leagues prior to that, have publicly available statistics for every single play. If we want to characterize Moneyball as anything, it is using data to identify undervalued statistics that correlate with success. Imagine losing a £50million striker who scored 10 goals more than average and replacing him with an average striker at £20million and a £15million goalkeeper who saves 10 goals more than the average.
The book centres around the Oakland Athletics. In 2002, the season the book covers, the Athletics had the third lowest payroll of any team. Less than a third as large as the New York Yankees. That really mattered. From 1992 to 2000 the World Series had been won by a team with a top five payroll. In five of those years the winning team had the largest payroll. Without an option to just increase payroll the Athletics had to think differently. It’s here that we have to explain some basic concepts of baseball.
In baseball each team has nine innings to bat, and nine innings to field. An inning stops when the batting side have had three batters out. The most common types are strikeouts, groundouts, and flyouts. A groundout is essentially a runout, so the ball is thrown to someone on the base before the batter gets there. A flyout is simply when the ball is hit in the air and caught. A productive at bat is when a batter either hits the ball or earns a walk. A walk happens when the pitcher misses the strike zone four times.
For as long as baseball has been around, and even today, the most popular measure of a batter’s ability was batting average. This is simply the percentage of times that a batter hit the ball fairly when at bat. An elite batter will have an average of around 33%, styled as 0.333. The Mendoza Line is said to be 0.200. This is the line where a batter is so bad that even elite defence can’t make them a profitable player to keep in your lineup.
Some of you will already be familiar with why batting average is a poor measure of success. Let’s take two imaginary batters; Bill and Ted. Both have 100 plate appearances, that is they come up to bat 100 times. Both hit 33 of those 100 balls given them an elite 0.333 batting average (BA). But Bill hits 33 home runs and Ted hits 33 singles, a hit just enough to get to first base. Clearly the two batters are not equally valuable.
Bill James had kickstarted the Sabermetric revolution all the way back in 1977 while working as a security guard at a meat packing plant. His first baseball abstract was read by fewer than 100 people, but his star has since grown. He won four World Series with the Boston Red Sox between 2004 and 2018 and was heavily involved in setting up Stats, Inc, which has become Stats Perform. James’ work was centred around identifying the stats which really mattered in baseball.
One of these stats was OBP (On Base Percentage) this is the percentage of time a batter gets on base and this crucially includes walks. Walks are usually credited to batters because there is a skill associated with not swinging at balls that are outside of the strike zone. It was a skill but it was a skill that wasn’t going to be costing the Athletics.
In Moneyball the Athletics looked for the players with high OBP and eschewed either the expensive players with high batting averages but low walk numbers or the home run hitters with miserable OBP, because they struck out so often. It is a compelling idea, look for the meaningful statistics which get hidden by the gaudier numbers around them, such as home runs. One of the two main stories of Moneyball was that baseball teams could be successful on small budgets simply by acquiring talent that was good statistically but not in the stats which earnt the money. In the case of the Athletics their batting performance was attributed to high on base percentages by their cheap band of rogues.
Oakland’s payroll put them 28th in the league but their batting in 2002 way outperformed that. They were the seventh best batting team by OBP, the holy grail of Moneyball statistics. But they were also fourth best by home runs with 205, the really expensive stat they were trying to navigate around. It’s here that Lewis starts to come undone with his telling of the story.
Wins Above Replacement (WAR) is probably the best sports metric we have. It combines all of a player’s contributions and then gives them a score compared to a replacement player. Anything over five is exceptional in a single season. The two best Oakland players in 2002 were pitchers, more on that later, and the third best was Miguel Tejada with a WAR of 5.7. Tejada is almost entirely absent from Moneyball because he doesn’t fit the narrative of the story. Tejada wasn’t someone who walked on base, he was someone who hit home runs, 34 of them in 2002. His OBP was worse than three other starters.
For anyone who has read the book or seen the film you will remember Scott Hatteberg. He was the catcher who could no longer play catcher because of a nerve issue and so was redeployed to first base. Hatteberg was the darling of the book because he was someone who walked a lot and was undervalued. He was good in 2002, but he was only the 7th best, behind Tejada and also Eric Chavez, both of whom were successful because they swung for the fences and got home runs and not because they walked.
The Athletics were a good batting team, certainly they outperformed what was expected of their budget. However, for all the talk of batting and walking in Moneyball, there is very little discussion of their real strength; pitching. Oakland’s starting pitchers were the second best in the league. They were led by the two best Oakland players; Barry Zito and Tim Hudson. They had WARs of 7.1 and 7 respectively. Third pitcher Mark Mulder was just slightly off the pace as the fourth best pitcher with a WAR of 4.6. None of those players earn more than a brief mention in Moneyball though. Largely because they weren’t ‘Moneyball’ players.
Zito was drafted in the first round of the 1999 MLB Draft with Oakland’s first pick. Not an under the radar sneaky pick but simply the best player on the board at that time. Mulder was drafted in the first round in 1998. Again, not a sneaky pick but the best player available at the time to the Athletics. That’s not to say they weren’t good picks, clearly both were, but they weren’t due to advanced analytical methods and so they were left out. Hudson was drafted slightly later and was more ‘Moneybally’ but still isn’t given much of a mention.
Lewis focuses his pitching energies on the relief pitchers. Usually baseball teams will start a pitcher who will throw between five and eight innings, depending on how successful he is. He will then give way to a gaggle of relief pitchers who will come in for shorter stints, sometimes just for one batsman. Finally, the closer will usually come in for tight games in the final inning to win it. Relief pitchers are a battleground for Sabermetrics. They are generally cheaper but still have a significant impact on the game so a small payroll team could punch above their weight by putting together a really good bullpen (the name for the collective relief pitchers) for a low cost.
That is what Moneyball suggests the Athletics did. Lewis devotes one chapter to relief pitcher Chad Bradford. Bradford was a quirky pitcher; he threw basically underarm with his hand grazing the turf. It was unlike anything batters had seen before and he was successful and largely ignored by other teams because of his quirky action. Bradford was the perfect Moneyball character. He was someone who was talented but was overlooked, and therefore cheap, because he was unusual.
Other critiques of Moneyball have gone in hard on Bradford. Allen Barra in The Atlantic, while quoting authors Sheldon and Alan Hirsch explains that Bradford’s role was overblown and even go so far as to say he’s not a very good pitcher. That is over the top, Bradford was very good, he was the second-best relief pitcher after Ricardo Rincon. Bradford though was eight times cheaper than Rincon.
Despite the exceptional pick-up of Bradford, the Athletics actually weren’t very good at building their bullpen. Their relief pitching was slightly below average, right in the middle of the pack. For all the coverage it got in the book, it wasn’t all that great and certainly not as good as the elite pitching of Zito, Mulder, and Hudson.
What Moneyball got right
Some modern analysis of Moneyball has suggested that it is irredeemably flawed. I don’t think that is true but there certainly are things wrong with it. The issues above, but also simple factual inaccuracies such as players being discussed when they had already left the team. Moneyball is more than just a fictional legend that happened to unearth useful truths. There is some really key data in it.
One of the Moneyball motifs was that drafting college players is better than drafting high school players. In the MLB draft players will enter from both the college level and the high school level. College players are usually characterized as having lower potential to improve but also less chance that they will end up being flops. High school players on the other hand have a much broader spectrum of potential outcomes.
Billy Beane, the General Manager of the Athletics and the character played by Brad Pitt in the movie, was of the belief that high school players couldn’t handle the pressure of playing as a professional. Incidentally, one of the tensions in the book is between the traditional scouts and the data analysts. The traditional scouts are portrayed as backwards old experts who score players down if they believe they have an ugly girlfriend as they think it means the player lacks confidence. Given that Beane is on the side of the analysts his opinion on high school players is pseudo-science in the same vein as that used by the scouts. In this one Beane, it turns out, was correct. College players are more successful than high school ones. They are more likely to play in Major Leagues, they play better when they are there, and you are more likely to find a superstar in college than in high school.
The Moneyball Impact
Batsman who walked were valued highly by the Athletics because they were valued poorly by the other teams. The other teams preferred to have the swing for the fences home run hitters. The influence of Moneyball has had a historic impact on baseball, but absolutely not in the way that you might imagine.
The problem with being successful with new ideas is that it doesn’t take a genius to just copy them a season or two after you have debuted them. If teams can tell that you are successful because you have built a batting line-up that walks a lot then the price of walking batters increase and the big payroll teams can just pick the best ones. Equally, if you think pitchers with quirky actions are the way forward then you would just find the quirkiest pitcher and pay them more than a small market team could afford. We would therefore expect that in the years following the 2002 Athletics’ season and the 2003 publication of the book there would be telltale impacts on the statistics. There was.
If we take 2003 as our year zero the walks per team per game in that year was 3.27 and the OBP per team was .333. In 2020 the numbers were 3.29 and .322. Essentially no change in the number of walks per team per game and a drop in OBP. What is even more surprising is that in the years following Moneyball both numbers sunk even lower. In 2014 the numbers were 2.88 and 3.14, the lowest since 1968 and 1972 respectively. Moneyball was a gospel about the virtues of getting on base by walking which saw historic drops in the rate of walks soon after its release.
What we have seen instead is the rise of the three true outcomes game. The three true outcomes are strikeouts, walks, and home runs. These are the skills that are deemed to be executed purely between batter and pitcher and without the need for the other eight fielders. From 2008 to 2018 each year saw an increase in strikeout rate. 2019 and 2020 are the two highest complete seasons by strikeout numbers and 2021 is on track to overtake that as well. Back in 2003 there were 6.34 strikeouts per team per game, last year there were 8.68 and this year there have been 9.38. In 2003 there were 1.07 home runs per game, in 2020 there were 1.28 and 2019 leads the way with 1.39. Baseball has actually moved towards those gaudy stats that Moneyball tried to work around.
The good news is that money is less important now. As we saw above, the era before Moneyball was written was dominated by the big payroll team. Oddly, that dominance really stopped in 2002, the season about which Moneyball is written. The 15th ranked Anaheim Angels won that year with a budget half that of the Yankees, just over $20million more than Oakland. Then in 2003 the Florida Marlins won it all with their 25th biggest payroll, a third the size of the Yankees and $1million less than the Athletics. Those weren’t one offs either, the Chicago White Sox won in 2005 with the 13th largest payroll, about 40% as large as the Yankees. St Louis won in 2006 with the 11th highest payroll, over half the size of the Yankees. Philadelphia in 2008, the 13th largest payroll. San Francisco in 2010, the 10th largest. St Louis again in 2011, the 11th largest. Kansas City in 2015 with the 16th largest. Finally, the Houston Astros in 2017 with the 18th largest. Sure, in 2019 the payroll of the biggest payers, the Boston Red Sox, and the bargain basement Miami Marlins, was three and a half times but there was hope for those smaller clubs.
I think that is one of the key outcomes from Moneyball. The actual solutions the Athletics found, acquire more batsman who can walk and get pitchers with odd deliveries into your bullpen haven’t really stood the test of time. Both can work but they’re not golden bullets. What has remained is the idea that financially weaker teams can thrive if they zig when the richer opposition is zagging.
Football and rugby teams have had great success rehabilitating older players beyond the peak of their careers. Rugby in particular has a rich history of older players moving on for one final contract only to be transformed at their new home. That is a tactic that is unlikely to be recreated by the rich clubs because they don’t need to take the risk. They could just acquire a player in the prime instead.
The problem for most smaller teams is that you can be used as a case study for the bigger teams. Once your zig has been proven to work the rich teams can just steal it. This is something that happened in cycling. Smaller teams would look to untapped markets such as the UK, Australia, or South America. The small teams would pick up talent and some of those riders would find a place and others would fade away never to be heard from again. Once the seam of talent is producing more hits than misses the big teams can swoop in and grab the best riders. Of course, with their budgets, they can afford to absorb a few misses in exchange for finding the really big hit.
Moneyball certainly wasn’t the first example of these kind of ideas but its huge critical and commercial success made us notice it. Sure, some of the viewpoints are outdated and there are perplexing inaccuracies in the storytelling but that really doesn’t matter. It is legendary, in all senses of the word. Yes it may be frustrating that when talking with most people about analysis the first thing they reference is the book or the film but the landscape that we all inhabit looks very very different without Michael Lewis and that book 18 years ago.