Is A.J. Burnett Broken?

In two starts following the All-Star Break, A.J. Burnett has allowed 27 baserunners (16 singles, four doubles, two homers, four hit batters, one walk) in 11.2 innings, giving up 11 runs, all earned, for an 8.49 ERA. Granted, that's only two games, but this appears to be more than disappointment over being named to his first All-Star Game but not appearing; his ERAs have risen every month this season:

Month	ERA	GS	IP	H	R	ER	HR	BB	SO	HBP	BK	WP	BF	WHIP	SO9	SO/W
April	1.80	4	25.0	24	5	5	1	7	20	1	0	0	101	1.240	7.2	2.86
May	1.82	6	39.2	31	10	8	1	13	39	1	0	1	156	1.109	8.8	3.00
June	2.41	5	33.2	40	11	9	1	7	27	0	0	2	143	1.396	7.2	3.86
July	4.68	5	32.2	40	17	17	3	7	24	6	1	2	147	1.439	6.6	3.43

Provided by Baseball-Reference.com: View Original Table

Generated 7/26/2015.

I think you know most of those column headings. BF is batters faced, WHIP is walks plus hits over innings pitched, SO9 is strikeouts per nine innings, SO/W is strikeout to walk ratio.

Obviously, the ERA's a concern. He's allowed more earned runs so far in July than he did in any two other months combined. His strikeout rate is down, and while he's not walking more hitters, those six hit batters indicate a lack of control. So what's going on?

I'm going to walk through a few steps that I like to check when a pitcher's effectiveness changes.

tl;dr: Burnett's been throwing pretty much the same pitches, at the same velocities, to the same spots of late. However, he's facing tougher opponents than he did earlier in the year, and when batters make contact, they've been hitting the ball harder than in the past. There is also a fair amount of evidence that Burnett's been just unlucky of late, suggesting that he could rebound from his July performance.

Has his repertoire changed? Sometimes a pitcher will change his mix of pitches, changing his overall outcomes. Burnett throws four pitches: A four-seam (straight) fastball, a two-seam (sinking) fastball, a curve, and a changeup. Here are his mix of pitches, by month, from Brooks Baseball:

Month	Fourseam	Sinker	Curve	Change
April	11%	53%	32%	3%
May	11%	57%	29%	3%
June	8%	52%	33%	7%
July	7%	59%	28%	7%

Do you see any big changes there? Me neither. Yes, he's throwing a few more changeups and fewer four-seam fastballs, but we're not talking a big shift.

Has his velocity changed? Burnett is 38 and in (he swears) his last season. There are only three starting pitchers in the majors who are older: San Francisco's Tim Hudson (40), Toronto's R.A. Dickey (40), and the Mets' Bartolo Colon (42). Might he be slowing down? Here are the velocities of his pitches, per month:

Month	Fourseam	Sinker	Curve	Change
April	91.6	91.4	82.4	85.8
May	92.3	92.2	82.7	86.3
June	92.4	91.7	82.7	85.9
July	92.2	91.5	82.4	85.6

There's almost no change there. Decreased velocity is not an issue.

Has his command changed? Velocity's nice, but it's not the whole story. Among National League pitchers, the Pirates' Arquimedes Caminero has the third highest average fastball velocity, 97.9 mph. But among Pirates pitchers with at least 40 innings pitched, he has the highest home rate and second-highest ERA, because major league hitters can hit just amount any pitch if it misses its intended spot. 

Here are his percentage of Burnett's pitches in the strike zone, starting in April: 48.0%, 49.5%, 47.8%, 48.5%. He's been right around the National League average of 47.9% all year. 

Of course, just throwing strikes isn't the story, either. Brooks Baseball tracks what it calls "grooved pitches." Think of the strike zone as a 3x3 grid, with nine identically-sized squares. The one right in the middle of the grid--the "middle middle" pitch, in the middle of the zone horizontally and vertically--routinely gets crushed. Andrew McCutchen is batting .393 with a .699 slugging percentage on middle-middle pitches. Starling Marte is .347/.551. Neil Walker, .313/.567. On the other side of the ball, Burnett has given up a .270 batting average and .358 slugging percentage overall this year, but .347/.569 middle/middle. So has he been grooving his pitches more? Nope. Here are his percentage of grooved pitches against right handed batters: April 6.9%, May 5.8%, June 6.3%, July 5.9%. Against lefties: April 5.0%, May 5.5%, June 4.5%, July 6.1%. It's all pretty consistent.

OK, so he's not missing the strike zone, and when he's in the strike zone, he's not missing his spots. How about when he's outside the strike zone? How is he at getting batters to swing at pitches outside the strike zone, where they're more likely to make weak contact? It turns out he's gotten better as the year's progressed: swings on 28% of pitches outside the strike zone in April, 24% in May, 33% in June, 30% in July. 

Are his pitches getting hit harder? His ground ball rate is down, and Pirates pitchers live on ground balls: 56% in April, 55% in May, 52% in June, 50% in July. Do hitters know what pitch is coming and therefore hit it harder? Well, they're making a little more contact, whiffing on 19% of Burnett's pitches in April, 21% in May, 18% in June, and 17% in July. The National League average is 21%. However, as he's allowed more contact, he's induced more soft contact: 15% of batted balls in April, 16% in May, 17% in June, 21% in July. Yet at the same time, he's allowing more hard contact as well: 27% in April, 33% in May, 27% in June, but (gulp) 43% in July. That's the second highest hard contact rate in the league in July. However, he's middle of the pack in soft contact this month. So batters are making more and harder contact off Burnett, though I'd really like to see another month of data to see whether this is a trend or a blip.

Is he facing tougher opponents? This gets overlooked a lot, but it's important in an era in which starting pitchers get, at most, 35 starts per year. If a starting pitcher goes against the Mets and Phillies (each scoring 3.5 runs per game) a lot more than the Rockies and Diamondbacks (4.4 each), he's likely to have better numbers than a starter who does the opposite. Let's divide each 15-team league into thirds to define good-hitting, average, and weak-hitting teams. During April, Burnett faced two good-hitting teams (Diamondbacks, Tigers), one average (Reds), and one weak (Cubs). In May, he faced three average teams (Cardinals, Reds, Padres) and two weak (Mets, Phillies). In June he faced two good (Giants, Nationals), two average (Brewers, Reds), and one weak (Phillies). In July he faced two good teams (Nationals, Tigers), and two average ones (Cardinals, Royals). He hasn't caught a breather, opponent-wise, since his June 14 start against the Phillies. So some of the deterioration in his performance is probably just a reflection of pitching against better-hitting teams.

Has he been unlucky? This is the trickiest question, because it depends on how one defines luck. Baseball analysts look at three statistics: batting average on balls in play (BABIP), percentage of baserunners left on base (strand rate), and percentage of fly balls going over the fence (HR/FB).

This doesn't mean that these figures are all a product of luck. Let's take two pitchers: Phillies ace/trade bait Cole Hamels, whose career ERA is 3.30, and his teammate Jerome Williams, currently sporting a 6.28 ERA and a career ERA of 4.55. Hamels' career BABIP is .285, Williams' is .290. Hamels' career strand rate is 76.9%, Williams' is 70.5%. Hamels' career HR/FB rate is 11.0%, Williams' is 12.0%. Does that mean Hamels' superior ERA is a product of BABIP, strand rate, and HR/FB luck? Certainly not; he's legitimately a better pitcher, and his numbers reflect it, across the board.

But variations in those figures can reflect luck. BABIP can be helped by scorching line drives hit right at a fielder and hurt by a pop-up that falls just beyond the reach of a retreating infielder. Strand rate is affected by clustering of hits; if a pitcher allows two singles, and no other baserunners, in each of three innings, he'll give up six hits and no runs. If a pitcher gives up no hits in two innings and six singles in his third, he'll also give up six hits, but will probably give up four runs. As for HR/RB, a 400 foot drive to straightaway center is a home run at Angel Stadium of Anaheim (the fence is 396 feet from home) but doesn't even reach the warning track drive at the Marlins Park in Miami (422 feet). So each of those factors can be influenced by circumstances other than the pitcher's skill.

Consider, for example, Hamels' disappointing 2009 season. After going 15-5 with a 3.39 ERA in 2007 and 14-10 with a 3.09 ERA in 2008, he slumped to 10-11, 4.32 in 2009. He was supposed to be the ace of my fantasy team that year, and toward the end of the season, I changed my team's name to Cole Hamels You Suck. But did he really suck? Let's look at those three luck indicators, from 2008 (when he was good), 2009 (the year in question), and 2010 (good again; 12-11, 3.06 ERA):

• BABIP: .259,.317, .289
• Strand rate: 76%, 72%, 83%
• HR/FB: 11%, 11%, 12%

OK, the last one didn't change much. But his 2009 BABIP and strand rate were career worsts for him. More hits dropped in that year (and no, it wasn't because batters were hitting the ball with authority; his hard contact rate of 26% in 2009 was the second best of his career), and they came more often with batters on base. His "peripheral stats"--strikeout, walk, and home run rates--were more suggestive of a pitcher with an ERA in the mid-threes, rather than his career-high 4.32. And, sure enough, he rebounded to a 3.06 ERA the following year.

So let's look at Burnett's BABIP, strand rate, and HR/FB, April-July:

Month	BABIP	Strand	HR/FB
April	.319	88%	5.3%
May	.294	80%	4.3%
June	.361	79%	4.2%
July	.352	74%	10.0%

Every single luck indicator has moved the wrong way over the past couple months, during which his ERA's gone from 1.81 in April-May to 2.41 in June and 4.68 in July. This isn't to say that he's been equally effective; as I've shown, he's allowed more hard contact in July, and he's also facing tougher opponents. (The latter's not going to let up; heading into today's games, the Pirates have the fifth toughest schedule in the league over the rest of the season, with an average opponents' winning percentage of .506.) But there's been enough going against him in those luck indicators to make me think that we haven't seen the last of the good A.J. Burnett.