For Major League Baseball teams, spring brings the promise of a better year. For the baseball media, it means putting their expertise to the test and forecasting player and team performances. Most of these forays into the future will be quickly forgotten, and for baseball’s prognosticators, the public’s amnesia is fortunate: they’re prone to swing and miss with great frequency.
In fact, last season featured some of the most surprising final regular season standings in the last 60 years—at least, when compared to how the media experts envisioned things panning out.
A close look at the final 2015 MLB standings finds American League West teams finishing in almost inverse order of some preseason forecasts. The previous season’s bottom dwellers—the Texas Rangers and Houston Astros—captured first and second, respectively.
In the American League Central division, the story was much the same. The favored Detroit Tigers, division winners the four previous years, collapsed, finishing last. The White Sox, seen by many as the “winner of the off-season” player acquisitions, were only a game and half better than the Tigers. The Cleveland Indians, picked by Sports Illustrated to win the World Series, finished third in their division.
And in the National League East, the Washington Nationals were nearly everyone’s pick to claim the division flag, but finished seven games behind the Mets, who surged after their trade deadline acquisition of Yoenis Cespedes from the underachieving Tigers.
Isn’t that what we expect and love about baseball: its unpredictability?
Yes. But the division races in 2015 were unpredictably unpredictable. After breaking down the numbers, we pose a couple of reasons why this may have happened.
From mediocre to abysmal predictions
Last spring, we examined the accuracy of nine media sources that predicted MLB’s division races between 2009 and 2014.
For the nine sources, we found that the average correlation between the predicted finish and the final standings was .49. Perfect predictions would yield a correlation of 1.0, while perfectly awful predictions would produce a -1.0.
A .49 correlation isn’t anything special. A budding MLB Nostradamus could produce exactly the same correlation over the six years by simply taking the previous year’s final standings for each division and using it as the next year’s prediction.
But compared to 2015, the mediocre 2009 to 2014 media forecasts were positively clairvoyant.
This spring, we analyzed the 2015 preseason predictions from Sports Illustrated, Sporting News, and a panel of baseball experts who attended the NINE Spring Training conference in Tempe, Arizona in March 2015. All of them fell well below the .49 average achieved by the media prognosticators in our previous study.
The strongest correlation between preseason rankings and the final standings was Sports Illustrated’s forecast that produced a .23 correlation. The NINE experts were next with at .12, while Sporting News finished in the cellar with a .08 correlation. For perspective, Sporting News’ average correlation for all years from 1955 to 2015 was a solid .60.
Yes, 2015 was a very tough year indeed.
We also wanted to see exactly how these poor predictions stacked up, historically.
So in the second part of our study, we examined the preseason/final standing correlations for three publications over several decades: Sporting News and Sports Illustrated from 1955-2015 and Street and Smith’s Baseball Yearbook from 1962-2007. Combined, the publications produced an average yearly correlation of .58, vastly better than our 2015 sample. When individual years were examined, only the years 1984, 1990 and 1991 produced a poorer correlation than 2015.
Splashy free agent signings build hype
So why was 2015 such a bad year for MLB predictions?
Here we leave the data behind and move to slippery slope of speculation.
The 2015 season saw the blossoming of young teams like the Houston Astros and Chicago Cubs, the rebound of steadily solid teams with injury-riddled 2014 campaigns like the Texas Rangers, and unrealistic expectations for teams active in off-season player shuffling like the Boston Red Sox, San Diego Padres, and Chicago White Sox.
There is also the element of luck. Balls batted into play fall for hits in greater numbers one year than other. High flies drift over the outfield wall with more regularity some seasons. Injuries can destroy one team’s chances and hardly influence another’s. The baseball gods smile and frown at whim.
Deities aside, media prognosticators could have been overreacting to the impact of high-profile free agent signings—a tendency pundits have shown in the past.
When we broke up our data sets, we found that predictions were weakest from 1977 to 1992—the era when free agency emerged and took effect.
The average correlation for this era was .40, compared to a .58 average for all other eras. The shuffling of marquee players from team to team certainly produces big headlines. But it also may cause media observers to overestimate the difference a handful of players can make.
When we examined the 30-team era—from 1998 to 2015—the average yearly correlation moved up to .56. This coincided with the rise of sabermetrics, which allowed journalists and front offices to begin using advanced statistics to precisely evaluate the contributions of individual players. This may have helped media prognosticators do a better job of evaluating the true impact of off-season transactions. For example, the Wins Above Replacement Player (WARP) statistic shows how, most of the time, the addition of a couple of new players is unlikely to significantly move the win needle.
We also found that some teams are more predictable over time than others.
Over our decades-long study, the easiest teams to predict were the Astros (.79), Athletics (.78), Mets (.77), Blue Jays (.76), and Mariners (.71). Dramatically more difficult to forecast were the Diamondbacks (.10), Dodgers (.20), Cardinals (.20), Giants (.30) and Reds (.30).
The pattern is fairly clear: four of the most predicable were expansion teams that went through a long period of predictably poor play before finally finding some success. When we examined what place in the standings was easiest for media sources to predict, it was last place by a clear margin. In other words, very bad teams are the easiest to spot. On the other hand, four of the hardest to predict were National League stalwarts (Dodgers, Cardinals, Giants, Reds) that weren’t likely to finish last very often.
What stands out from our study of division standings predictions over the past 60 years is just how bad a year 2015 was for MLB division diviners. The year 2016 will have to be better—at least, that’s what we’re predicting.
And for the record, our money’s on the Cubs or the Pirates to win it all.