When Vernon Wells was traded to the Yankees from the Angels, the move was rightfully panned by most sabermetrically inclined writers as he had inconsistent stretches in Toronto, and the trade to the Angels was one of the worst trades in recent memory, as Wells became one of the worst players in baseball. Supporters of the trade pointed to his strong spring training, and seemed to be supported when he started off hot. However, since then, he has been bad, and on a whole, has perhaps been even worse than he was as an Angel.
I never wrote about Wells because I just assumed that his start wasn’t real, but he can provide a test case to show us what to look for if a player really changed. This is very important for determining when hitters do change, and also important because I look at rookie hitters using Pitch F/X data. If it isn’t predictive, than it is purely descriptive, which isn’t all that helpful from an analysis standpoint. Wells had a very good April with an OPS of just over .900 and has been terrible since. The BABIP was pretty normal in April, though higher than it has been for him for the past few years. It also wasn’t the park, as he has actually been better on the road so far this year. Using the April cut-off, I wanted to separate the data into three parts, the Angel Wells, the April Wells, and the Wells since then.
Here are how pitchers pitched him on average (entire graphs are the strike zone):
When he was succeeding in April, pitchers were actually throwing more inside on average than they had when he was with the Angels. They also threw a touch lower as well. However, as the season has went along, while he seems to be seeing pitches that are slightly lower, they are also pitching him more outside, even more outside than they pitched him when he was with the Angels. Here are the pitches Vernon Wells has had success with (whether in the “in play, runs” or “in play, no outs” designations):
Obviously the pitches are grouped in the same general area as the all pitches. The April runs are the ones closest to the middle of the plate and inside. Like most hitters, Wells appears to like the ball closer to him than further away. When he was getting more pitches inside, he was having more success and taking advantage of them. Now, his run scoring plays are lower and further away, actually lower than his run scoring plays with the Angels, which were close to his April run scoring plays. His success is having to come on pitches lower and away than ever before, and there hasn’t been much success. This suggests a rather simple adjustment in how he is being pitched, but let’s look at his swinging strikes and balls put in play for outs:
His swinging strikes are coming on pitches that are lower and more away than his swinging strikes with the Angels, suggesting that he is having more problems with breaking pitches and off-speed pitches than he had as an Angel. The outs are much more tougher to judge, as they seem to be around his average pitches seen and hits, and often are driven by BABIP luck.
One other thing, other than just the adjustment of how he was pitched, was that he saw a different kind of pitcher on average, and I don’t mean by stuff. His average release points seen were drastically different*:
April: 5.95 vertically, -.07 horizontally
Non April: 5.87 vertically, -.69 horizontally
This shows that he was facing pitchers that were higher on average and closer to the center of the rubber in April, but since, he is facing lower pitchers on average and more pitchers that are further out right-handed on average, or just purely more right-handers in general. He has had the platoon advantage 35% of the time so far this year, which is slightly better than what he had as an Angel. From my count, 44.6 % of the pitches he saw in April were from left-handers. Since then, only 28% of the pitches he has seen have been against left-handers. I think this helps explain the differences in the pitches he saw, which helps explain his hot start.
I think this shows how looking at average locations and average release points can be useful in relatively small sample sizes, as they help explain anomalies. The answer isn’t quite that everyone should just ignore April statistics, though that is certainly a good guideline. Some hot starts by seemingly mediocre or bad players are actually real (for instance, Jose Bautista and Edwin Encarnacion), but the vast majority are not. It is important to look at the reasons for hot starts, whether randomness on balls in play, or in Wells case, a difference of pitchers faced and how they pitched him for a short period of time, and see if they suggest whether or not the start was real or not. That is, many times we can explain it a little better than just saying “regression to the mean” and seeing where and why the regression is taking place.
*Here is how the average release points broke down for Wells in 2012:
2012 All Pitches: 5.89 vertically, -.62 horizontally
2012 Swinging Strikes: 5.85 vertically, -.59 horizontally
2012 Runs: 5.84 vertically, -.64 horizontally
2012 Outs: 5.93 vertically, -.7 horizontally
2012 Hits: 5.94 vertically, -.98 horizontally