
Nathan Ray Seebeck-Imagn Images
Earlier this week on the CBS Fantasy Baseball Today podcast, Chris Towers noted that Shane McClanahan was struggling more than he would have expected, even with his STUPH model grades down. Chris wondered if a pitcher’s STUPH drops to a new level, does the pitcher perform worse than those who were already at that level? Pitchers need time to adjust to their new talent level. Well, I decided to give the question a quick look.
When McClanahan last threw back in 2023, he had a Pitching+ of 115, and now has a Pitching+ of 90. While he has the same number of strikeouts (9 K/9), the rest of his results have been horrible (5.00 ERA, 1.33 WHIP) as his walks have ballooned (5.5 BB/9). The biggest issue with analyzing this change is that it is extreme, with a two-year gap between throwing. I needed to simplify the search.
Finding a large enough sample set was a pain. The set needs to include starting pitchers (I stayed away from combining starters and relievers) with a STUPH decline into a range with a large enough sample of other pitchers already pitching in that range. And all this occurs in the few seasons that STUPH models have existed. In the end, I went with Pitching+ as my metric because I wanted the pitcher’s overall talent. Also, I wanted a quick snapshot, not a full-blown study. Future studies could use several factors.
For the sample, I took all starters with at least 40 IP in back-to-back seasons. I subdivided that group into pitchers who saw their Pitching+ drop between five and 10 points into the 95 to 105 Pitching+ range (n=26). Then I created another group of pitchers with a 97.5 to 102.5 Pitching+ range in subsequent seasons (n=37).
Here are the average values from the two samples.
Results when Pitching+ Drops and Remains Constant.
| STUPH Model |
Previous Pitching + |
Pitching+ |
Stuff+ |
Location+ |
IP |
ERA |
FIP |
xFIP |
SIERA |
K/9 |
BB/9 |
WHIP |
| Pitching+ Decline to Range |
106.8 |
100.2 |
96.7 |
102.8 |
108 |
4.50 |
4.51 |
4.26 |
4.34 |
8.70 |
3.39 |
1.35 |
| Pitching + Stable in Range |
100.0 |
100.1 |
97.5 |
102.1 |
125 |
3.86 |
3.97 |
4.07 |
4.15 |
8.45 |
2.84 |
1.24 |
Even though Pitching+ was the factor controlled for, both sets’ Stuff+ and Location+ values ended up similar. And that’s about it for any similarities. The declining pitchers threw fewer innings (possible injury?) with an ERA that’s 0.64 higher and a WHIP that’s 0.11 higher. These pitchers seemed to struggle at the new talent level compared to pitchers who had time to adjust to it.
The effects seen by this one small subset mean someone should dive in to verify the results. The different ways to cut up the data could be endless. Different change sizes? Include TheBot Values? Do changes in Stuff values lead to changes in Location values? After a season with worse STUPH, do the pitchers’ results improve? As long as the sample sizes remain reasonable, the combinations are endless. I’m afraid Chris opened a can of worms.