For a better browsing experience, please upgrade your browser.

FiveThirtyEight

Sports

In the ninth inning of the New York Yankees’ 2-1 loss to the Tampa Bay Rays on July 1, Brian McCann1 watched Grant Balfour’s fastball sail over the outside corner of the plate for a called third strike. As he walked dejectedly back to the dugout, McCann’s 2014 batting average sank to .220, the lowest seasonal average he’d carried in July at any point during his 10-year major league career.

A few days later, Terry Pendleton, McCann’s old hitting coach with the Atlanta Braves, delivered a theory as to why the catcher, who signed a five-year, $85 million contract with the Yankees last November, was struggling at the plate this season: He can’t handle the burden of expectations that come with playing in New York City.

This is a refrain we hear often whenever a player struggles in a huge (usually northeastern U.S.) market. The theory is that a rabid fan base and an overzealous sports media corps can cripple a player’s confidence and change his play. Based on a LexisNexis search of news reports, this happens almost exclusively in New York (and for the Yankees more than the Mets) and in Boston, with the occasional reference to Chicago and Philadelphia.2 The pressure to win in those cities can be staggering, but the “can’t handle the pressure” argument is laced with self-flattery: By linking a player’s production to the weight of external expectations, fans and the media presume they have significant power over how well he performs.

It’s easy to write this off as little more than a delusion. McCann certainly seems determined to prove that he can handle New York, that the narrative is nothing but manufactured nonsense. He’s produced a .352/.379/.463 triple-slash line since his batting average reached its nadir against the Rays earlier this month.

And yet, when I ran the stats on players like McCann, I was surprised to see that there may be something to the whole idea of big-market pressure affecting play, at least for batters. (Pitchers, not so much.)

To start my data work, I compared performance relative to expectation. Statistically speaking, expectation is best measured by one of the various and sundry projection systems sabermetricians have developed.3 The projection method most adaptable to the task at hand also happens to be the simplest one: Tom Tango’s Marcel system, because it’s open-source and can be customized endlessly. While Marcel is named for a monkey because it barely requires a primate’s intelligence to operate, it performs roughly as well as far more sophisticated setups.

A Marcel-like approach4 can generate predicted runs above average numbers for every player’s seasons going back to 1985; each is a credible estimate of what the player could have been expected to do before the season started. Adjust those numbers for playing time (so as not to punish players for injuries), compare them to the player’s actual runs above average marks, and we have a way to assess whether a player lived up to statistical expectations.5

These differences between actual and expected performance are the kinds of statistical disparities that theories like “he just can’t cut it in New York” are attempting to explain. If we believed that it was truly more difficult to perform to expectation in a city like New York, then we’d expect a very specific subset of player — a new acquisition with a high salary, like McCann — to perform worse in the aforementioned pressure-packed cities than in comparable destinations with more relaxed reputations. I used Los Angeles, Dallas (the Texas Rangers’ media market), San Francisco/Oakland and Atlanta as the control group,6 and then compared how batters and pitchers did in each set of cities.

When I weighted the results toward more highly paid stars,7 the average batter in our subset of newcomers saw his actual performance undershoot his projected performance by 1.9 runs when playing for teams in New York, Chicago, Boston or Philadelphia. Some notable deficits included Albert Belle underperforming by 47.5 runs after joining the Chicago White Sox in 1997, Chuck Knoblauch falling short by 39.3 runs as a new member of the New York Yankees in 1998,8 and J.D. Drew missing by 24.1 runs for the Boston Red Sox in 2007.

paine-mccann-batters-2

For comparison, here are the batters who did best in these high-pressure markets:

paine-mccann-batters

Meanwhile, the cohort playing for teams in Atlanta, Dallas, LA or the Bay Area averaged a -0.4 run difference between expectation and reality, or 1.5 runs better than those in the “high-pressure” cities.9 Score one for the idea that fan and media scrutiny play a role in player performance.

Then again, don’t tell that to the highly paid hurlers who joined teams in cities that are supposedly tougher to play in. Over the same period, that subset of pitchers prevented 4.5 more runs than would have been expected from their projections, while their counterparts in the more “easygoing” environs saved 3.2 more runs on average.10 Here are the top five overachieving pitchers in our four high-pressure cities:

paine-mccann-pitchers

For every John Lackey who struggled upon joining a team like the Red Sox, there was a Pedro Martinez, Curt Schilling or Roy Halladay who thrived in a tough market. And LA-area fans will remember the likes of Aaron Sele, Jason Schmidt, Jon Garland and Bartolo Colon, all of whom struggled upon landing with LA-area teams.

It’s hard to know what to make of all this. The conventional wisdom holds that pitchers would be more affected by added pressure in a new, rabid city, since they’re the players who have to stand alone on the mound and think about the enormity of throwing before, say, a packed Fenway Park. But, statistically, hitters had the more difficult adjustment to life under the microscope. This may be due to factors beyond fan devotion and an intense media climate; for instance, there may be something about the types of players teams like the Red Sox and the Yankees tend to acquire in free agency that is more correlated with those players underperforming projections. So while there’s correlation here, we don’t necessarily know which way the arrow of causation points.

McCann’s roller-coaster month of July might be the clearest argument against the armchair psychology of declaring players unfit for big markets. Baseball remains largely a game of luck and randomness, with sample sizes never quite as big as we’d like. As clear as the “he can’t handle New York” explanation might have seemed when McCann was in his early-season slump, the real culprit was probably just bad luck. In baseball, the influence of chance trumps anything the fans or the media can throw at a player.

Footnotes

  1. In this case, pinch-hitting for right fielder Alfonso Soriano. ^
  2. When I did a historical LexisNexis search of U.S. news sources for phrases such as a player having “what it takes to play in [city],” knowing “how to pitch in [city],” “handling the [city] media,” etc., those major league cities were the only ones repeatedly mentioned. ^
  3. Such as PECOTA, originally the brainchild of FiveThirtyEight editor-in-chief Nate Silver. ^
  4. I tweaked Tango’s weights slightly to maximize predictive accuracy, and split out offense and defense into separate projections for position players (because of limitations in fielding statistics, defense is regressed to the mean more strongly than offense). ^
  5. It’s important to note that these expectations account for past performance, regression to the mean, and the aging pattern of a typical major leaguer. In particular, the regression-to-the-mean component of the Marcel process should capture any effect of an abnormally good performance in a contract year. ^
  6. New York, Chicago, Philadelphia, and Boston represent four of the eight biggest media markets in the United States. Los Angeles, Dallas, San Francisco/Oakland and Atlanta are the other four. ^
  7. Adjusting for salary “inflation” using baseball’s version of the Consumer Price Index — the relative cost of a win in free agency. ^
  8. He was just about the only Yankee who fell short of expectations that year; his teammates on the batting side combined to exceed their projections by 175.2 runs in 1998. ^
  9. A weighted t-test between the two means yielded a p-value of essentially zero, so this difference probably didn’t happen due to chance, either. ^
  10. Again, weighting the average toward the more highly paid of the pitchers. And again, a weighted t-test produced a p-value of zero. ^

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