Cole Hamels has not looked like himself this year. After ERAs of 3.06, 2.79, and 3.05 from 2010 through 2012, Hamels now has a 4.61 ERA nine starts into this season. By itself, that doesn’t mean he’s doing anything wrong. Pitchers are vulnerable to random fluctuations and luck in nine — or even in 30 — starts. For example, take the Hamels from 2009: After the left-hander finished 2008 with a 3.09 ERA and a World Series MVP trophy, expectations were high for the next season. But Hamels fell by the wayside and posted a 4.32 ERA, even though his walk, strikeout, and ground ball numbers were all in the normal range for him.
His DIPS numbers suggested that there wasn’t anything really wrong with Hamels then, and that with a little bit of patience, he would go back to dominating hitters as he had previously. In 2010, that suggestion came to fruition, and Hamels went right back to being one of the game’s premier left-handed starters.
Fast forward to 2013, and Hamels ERA is up once again. However, this time, something does look different. Look at his walk rate, his strikeout rate, his SIERA and his ERA in the past six years:
Unlike in 2009, the spike in ERA is matched by his peripherals— his SIERA is similar to his ERA. The reason is simple — he’s walking a lot more batters and striking out fewer of them. That can be a change in talent level, and it may actually be. But it could also be that he or hitters have changed in some way, and that a game theoretic approach might be better. In other words, it could be that he — or the hitters he faces — have adjusted their strategy and he needs to evolve.
The first thing I always check to determine whether a pitcher’s talent level has changed (or if he’s hurt, etc.) is his velocity. That has been shown time and again to be a somehow underrated aspect of pitching skill. As shown on the table below, that’s not the problem at all — not on any of his pitches.
|Year||Fastball Velo||Changeup Velo||Cutter Velo||Curveball Velo|
The velocities are similar year to year. But what about movement?
|Year||Fastball x-Mov||Changeup x-mov||Cutter x-mov||Curveball x-mov|
|Year||Fastball z-Mov||Changeup z-mov||Cutter z-mov||Curveball z-mov|
That has not changed, either. So it looks like each of Hamels’ individual pitches is about the same. And what about the performances on those pitches? A common approach is to look at the run values on each pitch (normalized on a per-100-pitch basis), and you can start to see where Hamels is begins to see his results differ:
This makes it look like the main drop in performance has come on Hamels’ fastballs. But as I explained in my research on game theory and pitch selection, these numbers are misleading because they don’t consider the actual count in which the pitches were thrown.
Unsurprisingly, Hamels tends to throws fastballs when he’s behind the count so all this is tells us is he’s walking more people — which we knew already. Instead, we need to look at things more deeply. Ideally, I would have had performance per plate appearance by pitch thrown in each count. Still, without that data, I was able to detect a likely issue.
Let’s take an extra step and look at the walks and strikeouts by pitch type:
|Year||FB K%||FB BB%||CH K%||CH BB%||FC K%||FC BB%||CU K%||CU BB%|
This makes it look like he’s walking more batters on his changeup and fastball, but is he actually throwing them for fewer strikes?
|Year||Fastball Strike%||Changeup Strike%||Cutter Strike%||Curveball Strike%|
Look where the biggest dropoff happens: changeup strike percent is way down. If you look at a few other peripherals, it’s clear the issue is he’s throwing fewer pitches in the zone. Although he’s striking out fewer hitters, it doesn’t seem to be associated with a major change in contact percent:
Since we know he’s throwing fewer pitches in the zone this year — and that’s leading to the change in his walks and his strikeouts — we can now break that down further by pitch type:
|Year||Fastball Zone%||Changeup Zone%||Cutter Zone%||Curveball Zone%|
The obvious difference seems to be showing up on change-ups. Is that because of some difference in contact rate or swing rate?
|Year||Changeup Swing%||Changeup Contact%||Changeup O-Swing%||Changeup O-Contact%||Changeup Z-Swing%||Changeup Z-Contact|
Hitters seem to be responding rather similarly to Hamels’ changeups when they are in the strike zone, as in previous years. Hitters are also responding to Hamels’ changeups when they’re out of the strike zone. It’s just that he is throwing it out of the zone more. This puts him further behind in the count, and allows hitters to be more selective and draw walks, all while driving down Hamels’ strikeout rate.
This leaves open the question of whether Hamels can actually change that and throw more changeups for strikes. It may be that his talent has deteriorated such that he can’t locate his changeup as well, but more likely, this is a temporary issue. Hamels knows that falling behind hitters isn’t working, and his track record suggests that he is capable of throwing the pitch for strikes. If he can get back to throwing the change-up in the zone with more frequency, hitters will chase more often when he buries it out of the zone, and he should get back to being the dominating left-hander he’s been for most of his career.