Spring Outliers


Cubs manager Joe Maddon informed the media yesterday that Ian Happ would be starting the season in AAA. Spring-training results are not usually significant. In spring, teams tend to emphasize process over results. However, that doesn’t mean that results have no effect on opening-day roster decisions. Several years ago, FiveThirtyEight’s Neil Paine found that a raise in spring-training wOBA of 17 points compared to projected wOBA should raise a player’s projection by one point. Spring performances do have an effect on how we should expect players to perform–the small magnitude of the correlation just means it takes extreme results to shift expectations. With spring training nearing its end, we can see which players’ projections should change the most based on their spring-training performance.

This is a table of the 15 greatest decreases in projected wOBA based on spring results, showing spring-training wOBA, Steamer projected wOBA, the difference between the two, and Steamer projected wOBA adjusted according to spring-training wOBA. The second column, “Opp,” is Baseball-Reference’s measure of opponent quality, where 10 is MLB-level pitching, 8 is AAA, and 7 is AA.

PlayerOppPASpringProj.Diff.Adj.
Edwin Encarnacion8.3310.1270.342-0.2150.329
Andrelton Simmons8.6370.1160.316-0.2000.304
Marwin Gonzalez8.3280.1470.330-0.1830.319
Matt Olson7.7350.1730.347-0.1740.337
Brett Nicholas7.4250.1530.311-0.1580.302
Kelby Tomlinson7.5270.1320.283-0.1510.274
Ian Happ7.7560.1810.325-0.1440.317
Sean Rodriguez8.1520.1410.282-0.1410.274
Ichiro Suzuki8.2280.1410.275-0.1340.267
Adolis Garcia6.4250.1510.283-0.1320.275
Christian Vazquez7.4360.1630.293-0.1300.285
Clint Frazier7.0530.1950.322-0.1270.315
Socrates Brito7.9550.1720.295-0.1230.288
Jacob Nottingham7.1320.1560.277-0.1210.270
Jason Heyward8.2390.2060.325-0.1190.318

In Happ’s case, his bad spring training took him from a 0.325 projected wOBA to 0.317, while Almora’s strong spring brought him from 0.309 to 0.316, nearly Happ’s equal. Another Cubs outfielder, Jason Heyward, also dropped his projected wOBA to just around Happ’s level, but it seems as if Heyward’s lack of options and superior defense won out. Other outliers of note are Marwin Gonzalez, who is off to a slow start after missing a the beginning of spring training as a free agent, the recently retired Ichiro Suzuki, and the aging Edwin Encarnacion.

This is the same table as above but for the biggest increases in projected wOBA based on spring-training results:

PlayerOppPASpringProj.Diff.Adj.
Domingo Santana8.1280.5700.3200.2500.335
Byron Buxton7.8410.5470.3060.2410.320
Phil Gosselin7.9420.4990.2790.2200.292
Rob Brantly7.3250.4820.2640.2180.277
Lourdes Gurriel Jr.6.7440.5300.3130.2170.326
Justin Turner8.1480.5800.3640.2160.377
Brandon Belt8.1440.5540.3440.2100.356
Jay Bruce8.7330.5230.3140.2090.326
Corey Dickerson8.1340.5410.3320.2090.344
Chance Sisco6.8420.5000.2960.2040.308
Austin Hays6.9400.5120.3090.2030.321
Christian Yelich8.4390.5830.3820.2010.394
Isiah Kiner-Falefa7.5280.5020.3030.1990.315
Cristhian Adames7.8500.4720.2730.1990.285
Yan Gomes8.1370.4990.3030.1960.315

At the top of the list, Domingo Santana appears to be somewhere in between his 2017 and 2018 selves after being traded to the Mariners due to a Milwaukee roster crunch. Paine also recently wrote about Buxton using the same methodology used in his earlier article. Meanwhile, Christian Yelich appears poised to beat the initial projections again and put up numbers closer to his MVP-level 2018 season.

We frequently put too much stock in spring results, but by dismissing all of them we miss out on the outliers that can be indicators of what’s to come. Not all of these outlier performances will prove meaningful, but it’s useful to look at players who have managed to move the needle a bit.

Spring-training stats from Baseball-Reference. Steamer projections from Fangraphs. The R code to produce the above tables can be found in a gist I published.