For a better browsing experience, please upgrade your browser.

FiveThirtyEight

Sports

Criticizing baseball managers is a pastime as American as baseball itself. Players make their share of gaffes, but few mistakes are dissected with as much scrutiny as those made by the men in the dugout, especially during the postseason. Whether it’s his faulty in-game tactical choices or just the vague sense that he’s lost control of the clubhouse, the manager is an easy target. (As the old saying goes, you can’t fire the players.)

Yet sabermetrics tells us that most dugout decisions barely have any effect on the outcome of the game.1 Furthermore, if we look at effects on player performance, it’s evident that hardly any manager can distinguish himself from his counterparts. Based on my analysis, 95 percent of all managers are worth somewhere between -2 and +2 wins per 162 games. Last year alone, 21 batters and seven pitchers were worth more to their teams than nearly every manager of the last 112 years.2

To prove the relentless mediocrity of most MLB managers, I did what Phil Birnbaum did with his old “Were the ‘94 Expos Just Lucky?” study.3 I examined how players played in a given year (and under a given manager) compared to how we expected them to play based on their past and future performance.4

Take the Atlanta Braves’ 1990s first baseman, Fred McGriff. He had a 3.1 win above average mark in the strike-shortened 1994 season, which means he produced 3.1 more wins than an average player would have, if that average player were given the same amount of playing time. Based on his performance in the seasons before and after ’94, we would have expected McGriff to produce only .4 wins above average in ‘94 — 2.7 fewer than he actually did.

To whom should we credit those extra 2.7 wins? For the sake of this exercise, let’s give it to ol’ Bobby Cox, the manager of the Braves that year. This isn’t an outlier season a Braves player produced during Cox’s 25 years with the team. It turns out that Cox is one of the few managers of all time who could lead his players to unexpected performances year after year. Over the course of his career, Cox’s teams outperformed expectations by 3.1 wins per 162 games on average, sometimes exceeding their projected talent level by as much as 10 wins.

Nearly every other manager of the last 30 years — 172 overall — was, statistically speaking, indistinguishable from average. They either didn’t manage for long enough or didn’t separate themselves from the pack while they were still filling out lineup cards. Cox is one of only six managers since 1986 — Russ Nixon, Tony LaRussa, Davey Johnson, Billy Martin and Earl Weaver — who we can say with confidence actually affected the performance of the players he was managing more than the average manager.

One consequence of this apparent mediocrity is that general managers and owners are quick to pull the plug on their skips. The problem is that they often do so before they have all the data necessary to make an informed decision. If we expand our search further and take the data back to 1901, only 36 of 490 managers supplied enough evidence for us to know whether they were anything but average. This isn’t to say that all managers are average, only that we can’t prove whether they are until they manage more than 1,000 games.5

Turns out that most don’t ever get to that threshold. Out of those 490 managers over the last 113 years, only 119 (24.3 percent) made it to 1,000 games. The ones who didn’t make it were probably fired too soon, before they actually had a chance to prove whether they were any good. Barring Bobby Valentine-level catastrophes, no team should fire a manager before he works those 1,000 games, because there’s no way to prove he’s any worse — or better — than his replacement.

In many ways, managing a baseball team is a case study in survival analysis. In order to stay employed, managers have to outperform expectations repeatedly — something they show very little skill at doing from season to season. Aside from true outliers like La Russa and Cox, most of the differences we see among managers in squeezing wins out of their talent is in fact due to luck (or at least factors beyond the manager’s control).

For the typical manager, only 12 percent of his players’ unexpected wins in a given season are attributable to his own talent.6 On the flip side, in a given season, even one as bad as, say, the one Eric Wedge and the Seattle Mariners had last year, 88 percent of the unexpected losses had nothing to do with the manager. Yet it’s that 88 percent that often makes or breaks managers’ careers, if not in one season than in the seasons to come.7

As is always the case with calculations like these, it’s too simplistic to look at the small range of talent and conclude that Major League managers don’t matter. These findings do, however, help to underscore the brutal inefficiency of the manager market.

The hirings and firings, the manager merry-go-round every offseason, the relentless second-guessing — all are relatively pointless. You can’t prove that your team’s manager is any better or worse than average. Come to peace with that, and go back to worrying about the players.

Footnotes

  1. Perhaps realizing this, modern managers appear to have begun limiting their in-game calls. Steal attempts, a longtime bane of stat-heads, are lower now than they’ve been since the early 1970s, and the bunt, a similar bugaboo, has never been used less than it is today. ^
  2. This worth is measured not by wins above replacement (WAR) but by projected wins above average, using data across multiple years. That’s the closest approximation to a player or manager’s talent that we have. ^
  3. My version upgrades it by taking into account all of the modern features of WAR. Birnbaum’s just used basic Runs Created for batters and Component ERA for pitchers. ^
  4. I gathered WAR data from Baseball-Reference.com going back to 1901, focusing on a player’s wins above average (WAA) per plate appearance (or for pitchers, WAA per inning) in the six years surrounding — but not including — the season in question, which we’ll call Year N. Weighing these by their proximity to Year N and regressing to the mean, I established talent estimates for every player in a given season, informed by his performance in nearby years. ^
  5. The actual number is 1,075.5. Once he’s managed that many games, we can say that 50 percent of a manager’s unexpected wins were because of his own managerial skill. ^
  6. I looked at the correlation between unexpected wins in one season and unexpected wins in the following season, and the correlation between the two was only .12. ^
  7. I prepared a chart that shows the likelihood a manager will still be working the next year, the year after that, and so on. The longer somebody manages, the more chances he has to appear to screw up and get axed.

    538_chart.jpg

    ^

Filed under , , , ,

Comments Add Comment

Never miss the best of FiveThirtyEight.

Subscribe to the FiveThirtyEight Newsletter
×

Sign up for our newsletters to keep up with our favorite articles, charts and regressions. We have two on offer: The first is a curated digest of the best of what’s run on FiveThirtyEight that week. The other is Ctrl + , our weekly look at the best data journalism from around the web. Enter your email below, and we’ll be in touch.




By clicking subscribe, you agree to the FanBridge Privacy Policy

Powered by WordPress.com VIP