Fixing Baseball: Analytics

What's Moneyball done to the game?

When former Chicago Cubs front-office executive Theo Epstein stepped down from that position to take a job with Major League Baseball, here’s part of what he had to say about analytics and their effect on baseball:

“It is the greatest game in the world, but there are some threats to it because of the way the game is evolving. And I take some responsibility for that, right, because the executives like me who have spent a lot of time using analytics and other measures to try to optimize individual and team performance have unwittingly had a negative impact on the aesthetic value of the game and the entertainment value of the game in some respects.”

OK, so what’s that mean in plain English?

The people in charge of baseball teams — people like Theo Epstein — bought into analytics and a Moneyball-style of play which means (and this is the really, really short version) emphasizing walks and home runs on offense and trying to strike everybody out on defense. Base stealing, sacrifice bunts and hit and runs are discouraged because they risk outs and defense doesn’t matter all that much.

So when did all this start?

Analyzing baseball through the study of statistics is not a new idea; the Brooklyn Dodgers hired someone to do it back when they were still in Brooklyn, the Society for American Baseball Research (SABR) was formed in 1971, but according to MLB.com, analytics took a big leap forward after the publication of Moneyball in 2003.

After Moneyball became a success the people who analyzed baseball through statistical analysis were no longer seen as tin-foil-hat-wearing kooks; the book turned the tide of public opinion and teams started taking analytics seriously.

But because walks take at least four pitches and strikeouts take at least three pitches and home runs can only be hit on certain pitches in certain locations (which means letting non-home-run producing pitches go by) games slowed down.

Analytics also took movement (those stolen bases, bunts and hit and runs) out of the game.

To understand why Theo Epstein and the people who run MLB are concerned about the current state of baseball, it’s helpful to compare the numbers from the 2002 season (the year before Moneyball was published and analytics became a big deal) to the 2019 season, which at this point, is the last full season played. 

The following numbers come from Baseball Reference and go right ahead and think of them as pre-Moneyball numbers and post-Moneyball numbers and here’s what you’ll find: 

Fewer hits

In 2002 there were 43,272 hits and in 2019 there were 42,039 hits, which works out to 1,233 fewer hits and the emphasis on home runs is part of the reason.

In 2019 the Minnesota Twins led the majors in home runs with 307 and 274 of them came on balls hit with what Baseball Reference calls “fly ball trajectory.” Pretty simple: if you want to hit home runs, hit the ball in the air.

But as Royals coach Rusty Kuntz once said: the single biggest factor in whether a ball in play becomes a hit is the amount of time it spends in the air. If a ball is up there long enough, some big league outfielder will get under it.

When the Twins hit a fly ball and didn’t hit a home run, their batting average was .089. Everybody trying to hit more home runs led to fewer hits overall.

More strikeouts

Strikeouts went from 31,394 in 2002, to 42,823 in 2019.

To hit home runs most batters have to pull the ball and hit it into the short part of the park – the corners – and that means hitting the ball out in front of the plate, which also means starting swings sooner, which then results in getting fooled more often and an extra 11,429 hitters failing to do the most fundamental thing a hitter can do: get the ball in play.

And these days hitters are less likely to change their approach with two strikes; they keep swinging for the fences because that’s what teams want. As former Royals manager Ned Yost once said: “It’s almost like singles don’t matter anymore.”

If the ball does not get put in play, defenders cannot make great plays and there’s not much for fans to watch.

Fewer stolen bases

Pretty much everything you do on offense risks an out (including standing there watching hittable pitches go by), but since analytics advocates don’t like anything that obviously risks an out, stolen base attempts went from 4,032 in 2002 to 3,112 attempts in 2019.

Whether a runner is out or safe a stolen base attempt is an exciting play, but analytics managed to eliminate 920 of those exciting plays from baseball. 

Fewer sacrifice bunts

And if you don’t like obviously risking an out, the sacrifice bunt is even worse than the stolen base because batters are actually trying to make an out, which ignores the fact that most guys asked to bunt were probably in a situation where they were very likely to make an out anyway, except their team would not have the benefit of moving a runner 90 feet while the highly-likely out was made.

Sacrifice bunts went from 1,633 in 2002 to 776 in 2019. Once again, analytics managed to remove 857 plays where something happened.

Fewer walks

In 2002 there were 16,246 walks and despite analytics emphasizing the importance of the base on balls, walks actually went down in 15 of the following 17 seasons. In 2019 there were 15,895 walks which meant 351 fewer base runners compared to 2002.

Tell your hitters to be patient, take pitches and try to work walks and word will get around. Pitchers will then be aggressive about throwing strikes because they know your hitters aren’t going to be aggressive about swinging the bat.

Apparently, pitchers read Moneyball, too.

Fewer runs

All this boring baseball might be worth it if you scored more runs and that’s what analytics promised: emphasizing on-base percentage combined with slugging percentage was the key to scoring runs.

But in 2002 all big league teams combined scored 22,408 runs and, while the numbers went up and down each year, big league teams combined to score an average of 21,989 runs in the 17 subsequent seasons.

Longer games

The average time of a nine-inning game in 2002 was 2 hours and 52 minutes and in 2019 it was 3 hours and five minutes. And it got worse in the postseason, which is supposed to be baseball’s showcase; in 2019 the average World Series game was a little over 3 hours and 44 minutes long.

If someone was not a baseball fan and tuned into the World Series to see what all the fuss was about, a four-hour game without much action might not be the best advertisement for the sport.

More home runs, but fewer fans

The one category where analytics has seemingly improved the game is home runs; in 2002 there 5,059 of them and in the 17 subsequent seasons the average number was a little over 5,168.

So ask yourself if an extra 109 home runs (about 3.6 more home runs per team) are worth an extra 11,429 strikeouts, 1,233 fewer hits, 920 fewer stolen base attempts, 351 fewer walks and – on average – 419 fewer runs.

For millions of former baseball fans, the answer is no.

Every year since 2012 paid attendance has gone down, World Series TV viewership is in decline and the people who run baseball – the same people who thought using analytics was a great idea – know they’re in trouble. 

That’s what Theo Epstein was talking about and why he’s has been hired by Major League Baseball as a consultant to “on-field matters” and is supposed to work with analytics experts to come up with rule changes to create more action in the game – action the analytics experts took out of the game in the first place.

They say the first step in fixing a problem is admitting you have one, and Major League Baseball just did.  

(The next “Fixing Baseball” will examine some of the proposed rule changes designed to put excitement and action back in the game.)