Later today, the Jays will play their last Spring Training game. Notwithstanding the current two game losing streak, you probably already know it's been a pretty good Spring. At 23-7 the Jays have already wrapped up the Grapefruit League title and have the best winning percentage of any MLB team and the best run differential at +74. Indeed, it's even been enough to catch the attention of Leaf fans.
There's just one little problem - the old maxim that Spring Training doesn't matter. Hitters are working on timing, pitchers tinkering with mechanical changes and new or modified pitches, prospects getting a lot of playing time - Spring Training and the Regular Seasons are just simply different beasts. So it shouldn't be surprising that when Beyond the Boxscore recently looked at the correlation between Spring Training winning % and Regular Season winning Percentage for 2007-2011, they found a pretty weak positive correlation. It makes some sense that there's a positive correlation - better teams with better players should probably also have some advantage in the spring, but it's pretty diluted. MInor Leaguer has a similar chart, which I'd love to link to but have not been able to find it (despite it having been posted multiple times).
That seems to the close the book on any hopes of grandeur stemming specifically from the Jays' Spring Training dominance this year...but there's more to this story. When I was looking through John Dewan's Stat of the Week archive, I came across an article from April 2008 titled "Do Spring Training Team Records Matter?". Basically, looking at the time period 1996 through 2007, he found that 66 of the 96 playoff teams had spring training record at or over .500 (69%) whereas only 30 playoff teams were under .500 (31%). I found this interesting, but I also found the binary nature of the .500 designation limiting. Would it follow that the better the Spring Training winning %, the better the chance at making the playoffs? If this were the case, a great spring training record could actually be a pretty positive signal.
I gathered team-by-team data for the Wild Card era, from 1996 to 2011 (1995 is excluded as it was an abbreviated schedule due to the strike) which is 476 team seasons and 128 playoff teams across those 16 seasons. Then, I separated the data into different bins of Spring Training success as measured by the team winning percentage, and looked at the percentage of teams in each bin that ended up making the playoffs. I started with intervals of 100 points of winning percentage:
The numbers in white are the number of of teams that fell into each bin, and the success rate of reaching the playoffs for the teams in that bin is above the bar. There's a pretty clear pattern that as the Spring Training winning percentage increases, the incidence of making the playoffs increases as well. Strikingly, of the 8 teams that finished below .300, none made the playoffs. Not good 2012 news for Cleveland (.241, which would be the worst in the sample) and the Mets (.296). On the other hand, of the 8 teams with records at or above .700, 5 reached the playoffs. Currently, Toronto (.767), Detroit (.741) and Oakland (.700) meet this standard. So that seems like a good sign. Another good sign for the Jays is Tampa's .360 clip, which puts them among teams that make the playoffs at a 14% clip, with a fairly robust sample of 73. Overall, just under 27% of teams make the playoffs (though that will be higher this year), so that's significantly under. Next, let's make those bins a little more granular to see if there's anything else to glean, and go to 50 point ranges:
Once again, the same pattern is clear. There's a bit of a break from .550 upwards, where the chance of reaching the playoffs spikes upwards. This is still generally good news for the 2012 Jays. But Spring Training is relatively short, and teams can get lucky or unlucky or short periods, in terms of winning a bunch of close games. One way to try to measure and neutralize this is to use Pythagorean winning percentage, which calculates an expected winning percentage from runs scored (RS) and allowed (RA), using historical relationships. A team that scores more than it allows will have a Pythagorean winning % above .500 and vice versa, with a .500 Pythagorean team scoring exactly as many as it alllows. I could only find RS and RA data from the Spring for 2003-11, but nonethless, we get the following chart:
I find this to be the most striking chart of all: if a team scores more runs than it allows during Spring Training, its chance of making the playoffs is significantly higher than if if gives up more than it scores. However, there's really no observed benefit from the magnitude of the differential, since the higher the differential, the better the Pyhtagorean record. This is somewhat disappointing for the 2012 Jays and their majestic .728 Pythagorean winning percentage, though it is certainly better than the implication of being outscored.
Having gathered all the data, I thought it would be fun to take another looking into any correlations that exist between winning in Spring Training and the Regular Season. I started by replicating the chart that Beyond the Boxscore and Minor Leaguer produced, for the entire 1996-2011 period:
With more data points, there's a much weaker, albeit still positive correlation. Less than 5% of the variation in Regular Season record than be explained by variation in Spring Training record, which is very weak. Nothing new here, so moving on the looking at the correlation between Pythagorean winning percentages instead (again, 2003-2011):
The correlation is similarly weak, so no help from trying to do this. One other suggestion I have seen is that the end of Spring Training may be more sigificant, since starting position players are getting more time and pitchers are going deeper. For 2004-11, I have the record for the last 10 games of the Spring, which can be compared with the regular season record:
Again, a very weak positive correlation, and once again, there's essentially nothing to the relationship. One final straw at which to grasp: maybe the improvement (or decline) from season to season in Spring Training is a predictor of improvement in the Regular Season. Below are charts of this relationship for both the actual winning % and Pythagorean winning %.
These are the most random charts yet. Simply put, there's just no relationship of which I can plausibly conceive between winning percentages in Spring Training and the Regular Season. On the other hand, it appears that team Spring Training records have some positive relationship with making the playoffs. Of course, correlation is not causation, and so I hesitate to draw any substantiative inferences about the Blue Jays 2012 fate from their great record other than that winning is better than losing. With that in mind, let's hope for a good final game with Alvarez and Drabek on the mound.