We talk about velocity quite a bit when evaluating pitching and rightly show concern when a pitcher loses it. Jeff Zimmerman has done a great job tracking and evaluating it both in his articles and constantly updated google docs. But while generally speaking more velocity is better, and losing velocity is costly (Zimmerman estimates a 1 mph drop raises a starter's ERA by .28), not all pitchers respond to velocity changes the same way.
The analogy I want to make is to the concept of price elasticity in economics. Generally, when you raise the price of a good, the demand for it declines, and when you lower it, demand increases. But the demand for some goods is inelastic, i.e., it is not especially sensitive to a change in price. When gas prices go up, for example, demand remains largely the same because people need to drive regardless.
I'd argue some pitchers are especially elastic with respect to velocity - even a relatively small loss or gain greatly affects their performances, and others are relatively inelastic - even with diminished velocity, they remain more or less just as effective.
Let's take a look at some examples:
| Player | *Peak Velo | ERA at PV | Low V, Peak** | LVP ERA | 2015 Velo | 2015 ERA | 2016 Velo | 2016 ERA |
|---|---|---|---|---|---|---|---|---|
| Felix Hernandez | 94.1 | 2.27 | 91.9 | 2.76 | 91.8 | 3.53 | 89.5 | 1.8 |
| C.C. Sabathia | 93.7 | 2.7 | 92.3 | 3.38 | 90.1 | 4.73 | 88.1 | 5.28 |
| Justin Verlander | 95 | 2.4 | 94.3 | 2.64 | 92.8 | 3.38 | 91.9 | 5.79 |
| Tim Lincecum | 94.1 | 2.62 | 92.3 | 2.74 | 87.2 | 4.13 | N/A | N/A |
| Cliff Lee | 91.5 | 2.4 | 90.5 | 2.54 | 89.6*** | 3.65 | N/A | N/A |
| Zack Greinke | 93.7 | 2.16 | 91.8 | 1.66 | 91.8 | 1.66 | 91.2 | 6.16 |
| Jered Weaver | 89.1 | 2.41 | 86.5 | 3.27 | 83.3 | 4.64 | 81.8 | 3.12 |
| Clayton Kershaw | 93.9 | 2.79 | 92.6 | 1.83 | 93.6 | 2.13 | 92.6 | 1.5 |
* peak velocity season when a pitcher was at his best, i.e., I excluded seasons with higher velocity when the pitcher just came into the league but wasn't especially effective.
** pitcher's best year while pitching with somewhat diminished velocity.
*** 2014 stat line as Lee was retired in 2015
From the chart, you can see some pitchers seemed to be more affected than others by velocity loss. Sabathia seems much more elastic than Hernandez, for example, while Greinke seems completely inelastic over his range. Of course, the data is noisy because it doesn't account for park and league, and ERA itself is an imprecise stat. But I used it rather than FIP because I don't believe BABIP is mostly luck. And the small-sample 2016 numbers are included more as a curiosity, something on which to keep an eye.
Even so, it makes intuitive sense a pitcher like Hernandez whose pitches have so much movement might still be able to miss bats as almost as effectively in the low-90s as he did when he was throwing 94. At some point, though velocity loss catches up to everyone - Jered Weaver, who was a good pitcher even at 86.5 (probably because he was effectively throwing much harder), got shelled at 83.3. It remains to be seen if Hernandez is inelastic enough to thrive at 90 mph - assuming he can't get back to his 2014-15 levels.
Bottom line, if we assume pitchers vary in their "velocity elasticity," it would be useful to know what kinds of pitchers are more or less sensitive to changes. Does it have to do more with movement, pitch selection, repertoire, build or batted ball profile? And are there acceptable ranges where a velocity drop makes almost no difference versus danger zones where even a half mph spells doom for a once-effective pitcher? Or does velocity loss operate linearly across all ranges?
Velocity data is a great tool, but we run the risk of overvaluing it - just as we did with BABIP initially - until we answer these questions.