What's the matter with Brandon McCarthy?

Pretty frustrated with being terrible thus far, so tomorrow I'm going to go scream at pigeons in Central Park if anyone cares to join.

— Brandon McCarthy (@BMcCarthy32) April 17, 2013

His sense of humor is intact, but this continued a string of poor appearances by McCarthy, which have seen him allow 28 hits and 15 runs, 13 earned, in 15.2 innings. with nine strikeouts and four walks. The resulting 7.47 ERA is not what we wanted to see. That was the first time I've been able to watch McCarthy pitch in a regular season game, and it wasn't very impressive. He seemed to be missing his spots with disturbing regularity, and even when he put the ball where he wanted it, the batters seemed either to make good contact or foul the pitches off. The home-run allowed to Robinson Cano was the most obvious case where McCarthy missed, and the hitter didn't.

Three games in is certainly too early to panic. But I don't think it's too early to be concerned. Let's take a look at the story so far, and see what the difference is between Brandon in 2013, and the one who has enjoyed great success (when healthy) over the previous two seasons.

The narrative

Game 1:Apr 3 vs. Cardinals. 5 IP, 9 H, 6 R, 6 ER, 0 BB, 1 K
Overshadowing McCarthy's outing, his debut for Arizona, the Diamondbacks eventually won this in 16 innings, in the longest game ever played at Chase Field. It seemed a strange night for McCarthy: "It's still one of the weirdest starts I can remember in a long time. I felt really good and I threw a lot of strikes, but I just couldn't throw good strikes. It was weird, everything just kept missing by two inches. It was an inch up, when it should have been an inch down." By the time the game was finished, reliever Josh Collmenter had thrown as many innings as McCarthy and, as Brandon put it, "Starter two was better than starter one."

Game 2: Apr 9 vs. Pirates. 6.2 IP, 10 H, 6 R, 4 ER, 2 BB, 4 K
After this game, which was largely decided by the five-run inning enjoyed by the Pirates in the fourth, McCarthy came out with what seem initially like a curious comment, "That's as well as I've thrown a baseball in my professional career, without a doubt, and I absolutely felt like I was able to do everything I wanted to." So, what went wrong? "I just think we got into a routine there and we weren't able to change gears fast enough. Too many pitches that with two strikes we weren't changing things up enough..I've got to be smart out there and shake and mix things up more the way we did the fifth through the seventh inning than we did the first few."

And who's fault would that pitch selection be? Step forward, Miguel Montero, who took the unusual step of issuing what was basically a public apology for the decisions made during the game. "I apologized to him, because I feel like I called a really bad game. I think he had really good stuff going, I just didn't make the right calls. He threw everything that I put down, so I felt bad because he really trusted me, and I feel like I let him down... I just didn't have the right fingers yesterday. It's just a bad feeling. I take it personal."

Game 3: Apr 16 vs. Yankees: 4 IP, 9 H, 3 R, 3 ER, 1 BB, 4 K
You could argue that McCarthy was one pitch away from a solid outing: the same change-up which he had used to strike out the previous batter, Brett Gardner, was deposited into the bleachers by Robinson Cano for a three-run homer that proved decisive. But you could still argue against the pitch selection: it's hardly what you'd call McCarthy's best weapon, and even he acknowledged that: "It's still such a hit-or-miss pitch. Sometimes it's there, sometimes it's not. It's very hard for me to get to a place where it can be relied on in a situation."

The decision appeared partly fallout from the previous start, where Montero thought they had relied too much on fastballs. Last night, 19% of McCarthy's pitches were change-ups, compared to 15% in the first two starts. For the Cano HR, Montero said, "I felt he had a good feel for the changeup, and I just called it. I don't second-guess myself. It was just a bad pitch, regardless." One of a few that night, as McCarthy was lucky to post zeroes that far. "I was just completely out of sync and just kind of not able to do a whole lot.... I just couldn't get right from the 'pen all the way through. Even after evaluating, I don't know what it is mechanically that I was looking for. Just everything was so off."

The results

The chart below is based off data from fangraphs.com, and shows what has happened to McCarthy's pitches in the last three seasons. Any row beginning "O-" refers to pitches out of the strike-zone, while "Z-" is for pitches in that zone. So O-Swing% is the percentage of pitches that would have been balls, which provoked a swing, while Z-Contact% is the percentage of times batters made contact with a pitch in the strike zone, when swung at.

The two things that stand out are that batters are both being more aggressive, and making more contact with McCarthy's pitches, than in previous years. The former applies to pitches both in and out of the strike-zone, with batters now swinging at a majority of his pitches: 52%, compared to 44% last year. Contact is particularly high when hitters swing out of the strike zone, jumping from 70% to 86% when taking a cut at McCarthy's balls (er, as it were), but contact has also ticked up in the zone. The net result is that McCarthy's percentage of swinging strikes is down by about a third, from 7.1% last season (and 7.6% career) to only 4.8%.

The pitch selection

Some explanation of "Pitch values" is needed: Think of it as like Win Probability for pitches. Say, it's a 1-1 count with one out and a man on second. There is, historically, an expected number of runs scored in an inning from that situation. Every ball, strike or ball in play changes the expected number - maybe up, maybe down, and fractionally, in most cases. Given that change, and the kind of pitch thrown, you can add up the positives and negative results for a particular type, and thereby find out which were the best and worst weapons in a pitcher's arsenal. The result is usually expressed in runs per 100 pitches thrown.

The table below shows, over the past three years, the percentage of pitches of each type thrown by McCarthy, and their Pitch Value per 100 pitches of that type (PV). You should bear in mind that, especially for this season, we are still talking a relatively small number of pitches, and so their values are subject to large fluctuations, purely by chance. [in 2011, there were also a small amount of sliders thrown by McCarthy (2.1%), but I've excluded them from the chart as they weren't apparently part of his arsenal in 2012 and 2013]

What this seems to show is that McCarthy has been using his change-up more frequently this season, and all told, the results haven't been too bad. Most of the damage seems to have come off Brandon's curveball, which has gone from being a plus pitch in the previous two years - and his most successful last season - to being the one that has cost him the most. Rather more damage than I'd like has also come off McCarthy's regular fastball, and considering it's the pitch he throws most often, it's like the biggest overall contributor.

Bad luck?

This sounds like something Russ Ortiz might have blamed in his heyday with the Diamondbacks, but there's reason to think McCarthy has been a little unlucky. We have a lot of data to suggest that once you take walks, strikeouts and home-runs out of the equations, just leaving those balls in play, pitchers don't have very much control over what happens to them, and BABIP (batting average on balls in play) tends to even out over time. For instance, Ortiz's career BABIP was .290; Randy Johnson's was .295. McCarthy's prior to this season was .284, and the National League average this seas is .296. Around that is the kinda range you'd expect to see.

McCarthy's 2013 BABIP is .429.That's the highest by any qualified NL pitcher this season: last year's leader was more than a hundred points lower (Lance Lynn, at .323). But there's a "but". McCarthy has been giving up hard-hit balls in 2013: his line-drive rate allowed is 30%, compared to a career (and MLB) average of 19%. Line drives are a lot more likely to become hits: last year, their BABIP was over .700, so more of them will inevitably jack up BABIP. A few years ago, Dave Studeman concluded, "As a general formula, BABIP equals the percent of batted balls that are line drives plus .120.:" Do that for McCarthy's line-drive rate and you get... .420.

Going forward

This year, he has thrown a first-pitch strike 70% of the time, compared to league average 59%. Maybe it's something opposing hitters are exploiting, knowing they're likely to get a good first pitch? He also needs to do a better job of putting batters away: with two strikes, batters are hitting .314. But I think the key for McCarthy is getting the line-drive percentage back down to more normal levels: those are becoming hits at the expected high rate, and leading to a lot of the damage. He also needs to get more swinging strikes: right now, none of his pitches seem to be capable of doing that, leading to all of them sitting in negative Pitch Value territory - that's a bad place to be.

Hopefully, Brandon can sort it out, because our rotation will look a great deal better with him slotting smoothly in and performing at the levels he did over the past two seasons. But if anyone has the brain to work it out, I'm confident McCarthy can.