Baseball analysis is weird.
Wait, let me back up. One of the great aspects of baseball, from an analytical perspective, is that we can truly zero in on the contributions of individual players. In most other major sports, there are varying degrees of "team" factors that can cloud the picture of how much each individual contributes; those factors are quite scarce in baseball.
And yet, more than any other sport, baseball’s data can be rendered irrelevant rather quickly. A hitter can change his stance, or a pitcher can add a new pitch, and suddenly, that player’s true talent level can become dramatically different. Witness Jose Bautista as one of the more dramatic recent examples of this phenomenon. I can’t think of an example in another sport where mere minor mechanical modifications would stimulate such strong shifts in performance.
In essence, therefore, we can comb through tons of data to find the holy grail of a player’s true worth, only to see that true worth suddenly undergo a profound transformation. Every year, there are a few players who have huge breakouts or collapses, and given that we’re nearing the halfway point of this campaign, it’s time to investigate a few of these breakouts and see a) how it’s happening and b) how legitimate it is.
So, today I’m going to take a look at four pitchers who have had substantial improvement in 2012: Jason Hammel of the Orioles, James McDonald of the Pirates, Jeff Samardzija of the Cubs, and R.A. Dickey of the Mets. What can we learn from comparing their PITCHf/x from 2011 to 2012? We'll take some detours along the way, but as it turns out, we can learn quite a few things.
When it was announced in the offseason that the Orioles had acquired Jason Hammel and Matt Lindstrom for Jeremy Guthrie, I thought "Oh, easy win for Baltimore." After all, Hammel had posted two 3.9 fWAR years in Colorado, while Guthrie had never topped 3.0 WAR in his career. Throw in the useful Lindstrom moving to the Orioles as well, and it seemed like a no-brainer. I was thus surprised when the general reaction was outrage that the Orioles were giving up the "consistent" Guthrie for…well, honestly, I’m not sure what the impressions of Hammel were. Nobody seemed to really think him through. Then again, there wasn't much time to think him through when you have to react to the move in 140 characters.
Of course, the piece of information I left out in the previous paragraph is that Hammel collapsed to 1.0 fWAR in 2011 after the two 3.9 fWAR seasons. His strikeout rate collapsed to 12.7%, his walk rate spiked to 9.2%, and even his groundball rate fell slightly. At 29, he’s not exactly supposed to be getting better, so perhaps the idea of the deal’s naysayers was that he was already on his downturn. Perhaps we should thus examine what went wrong from 2010-11 before we think about 2011-12, then.
With the Rockies, Hammel worked off a standard four-pitch mix: a fairly straight 91-95 mph four-seam fastball that featured occasional sink and cut, a big-breaking 82-86 mph slider, a 75-80 mph curveball with solid movement, and a hard sinking splitter-esque changeup in the mid-80s.
Very little changed about those four pitches from 2010 to 2011. His changeup came in 1 mph harder in the latter year, but all the other pitches were within .4 of their 2010 velocities. Hammel used them in a very similar distribution, right down to throwing exactly 60.9% fastballs both seasons. He used the changeup a bit more at the expense of his breaking stuff, but it was a minor change, and actually, his changeups swinging strike rate increased from 10.2% to 13.0%, so that wasn’t the problem.
Really, there wasn’t even that big a difference in Hammel’s overall results on a PITCHf/x level. His SwStr% went from 7.2% in 2010 to 6.5% in 2011. Then again, that doesn’t suggests bad luck with his mediocre 2011 K% as much as it suggests good fortune with his solid 2010 numbers.
The one notable issue with Hammel in 2011 on a pitch level was that he struggled with his curveball. It had essentially the same speed and movement, but his strike rate with the pitch went from a solid 63.9% to a troubling 55%. Sometimes, that’s an acceptable tradeoff if it comes with a swinging strike increase, but that number also went down, from 12.3% to 7.8%. Looking at his curveball location charts from the seasons, one can see he was able to spot his curveball more consistently down and/or to the arm side in 2010 than 2011:
Note also where the balls below the zone are. In 2010, he was throwing a lot of curves right below the zone, where they could still get called strikes and were certainly tough to lay off. In 2011, he tended to miss the bottom of the zone by at least four inches when he missed low.
Combine the disappearance of the curve’s effectiveness with slightly lowered strike rates on the fastball and slider, add in that his 2010 performance wasn’t as good as it looked, maybe a pinch of bad luck, and presto! A 2.9 WAR regression and a FIP and xFIP increase of about a run.
What changed for Hammel in 2012? Certainly not his curveball. It’s only gone for strikes 46.9% of the time, its whiff rate is just 8.5%, and he’s deemphasized it to just 9.9% usage. But a couple of things have happened that have pushed his performance forward. After his one-hit shutout of the Braves last week, Hammel sports a 2.87 ERA and 3.39 FIP despite moving to the tough AL East. His K% has nearly doubled, to 23.1%, and his groundball rate is a career high 52.9%.
For one thing, he’s picked up a bit of velocity. His fastball is sitting at 93.7 mph this season, and his slider at 85.1; both are career highs. His changeup is now into the upper 80s, but he rarely uses it.
Instead, Hammel has turned to sinker/slider pitching. He’s throwing primarily two-seam fastballs this year; his overhand arm slot doesn’t add a ton of vertical drop to the ball, but he’s finally been able to come up with a fastball that can move in on righthanded batters. His arm slot and height create some extra downward plane for the pitch. In turn, this has made his four-seam fastball more effective on the occasions he does turn to it. In 2011, his fastball’s swinging strike rate was 4.1%. This year, the two-seamer is at 6.7% and the four-seamer at 9.6%. Between the two, Hammel is generating swinging strikes about twice as often with fastballs as he was before—even in 2010.
When you consider that the guy’s throwing 91-96 with significant running action and good plane, Hammel suddenly becomes fairly imposing. With a fastball that hitters really have to worry about, the big righty has gotten tremendous results from his slider, which is going for strikes 68.6% of the time and getting an excellent 18.1% swinging strike rate. He’s turned to it more this year at the expense of the ineffective curve and changeup.
The gains seem to be tied to the development of the two-seamer, which explains the increase in groundball rate (two-seamers get grounders) and swinging strike rate (the two-seamer generates more swings and misses than his four-seamer did on its own, and the four-seamer and slider play better off of it).
With a career-high 10.0% swinging strike rate, the increase in grounders, and a good reason for both, Hammel’s breakout seems to be legitimate. Maybe not 2.87 ERA legitimate, but 3.50 seems realistic—not bad for a guy in the AL East with a career 4.78 ERA.
James McDonald was supposed to be a breakout pitcher in 2011. He had pitched to a 3.12 FIP in a half-season the year before, and his fastball/curve combination was considered excellent.
While he didn’t totally collapse (18.8% K), McDonald was plagued by bouts of extended ineffectiveness and wildness, with a 10.3% walk rate and a 4.68 FIP that left him at just 0.4 fWAR.
This year, he’s already easily quintupled that final mark, posting 2.3 fWAR in a meager 81 1/3 innings. His control issues are mostly gone (7.6%) and his strikeout rate has shot up (24.8%). He’s gotten some extraordinary luck from BABIP (.251) and HR/FB (4.6%), but even his xFIP is nearly a run improved over last season.
To understand his 2012, though, you have to understand McDonald’s 2011. And let’s split his 2011 into roughly two halves.
First 15 starts (Opening Day through 6/21): 79 2/3 IP, 42 BB (11.7%), 62 K (17.3%), 11 HR (1.24 HR/9), 4.85 ERA, 4.97 FIP
Final 16 starts (6/26 to end of season): 91 2/3 IP, 36 BB (9.1%), 80 K (20.2%), 13 HR (1.27 HR/9), 3.63 ERA, 4.43 FIP
The homer rate stayed problematic in both segments, but McDonald’s K/BB ratio went from problematic to promising during the course of the 2011 season.
Now, it is something of a cardinal sin of sabermetrics to use half-season splits like that. FanGraphs doesn’t offer "first half" and "second half" splits for precisely that reason, after all.
As it turns out, though, June 26 was a pretty important turning point for James McDonald, such that it makes sense to look at his 2011 as "before June 26" and "after June 26." Here’s his PITCHf/x graph from the first 15 starts of 2011:
And here is his June 26 start:
See those four pitches right along the 80 mph line? Prior to June 26, he’d thrown one all season—on June 16. The rest of the way, his PITCHf/x graph looked like this:
PITCHf/x had no idea what to do with those pitches, since McDonald had never thrown them before. But those pitches are sliders (and have been correctly identified as such in 2012). So his first half and second half of 2011 are not arbitrary endpoints, they are "fastball/curve/change McDonald" and "fastball/slider/curve/change McDonald."
Brooks Baseball correctly classified the pitches as sliders, and listed the pitch as drawing swinging strikes at a whopping 24.2% clip last season—a huge difference-maker for the righthander. Part of its success could be chalked up to the element of surprise—McDonald was always a three-pitch guy, but suddenly here came a plus slider out of the blue, and it took the league awhile to adjust. Since Brooks and Baseball Info Solutions both have the pitch at 5% usage in 2011, and he only used it for half the season, its usage once he added it is probably close to 10%, which means he certainly wasn’t overexposing the offering. Hitters still had to go up looking for his explosive low-90s fastball and huge upper-70s curveball.
McDonald has roughly doubled the slider’s usage in 2012, and its swinging strike rate has come down a tad…to 19.7%. The pitch has allowed him to deemphasize his other three offerings, allowing his fastball to play up slightly (6.0% to 6.6% swinging strikes). It also gives him a second pitch he can throw for strikes; his fastball has been at 64.4% strikes in both 2011 and 2012, but his curveball has lagged in the mid-50s and his changeup is not a major part of his arsenal. The slider is at 61.8% strikes this year.
The slider is a great weapon for McDonald, though it’s probably not enough to get him to maintain a 2.32 ERA going forward. As with Hammel, his 3.60 xFIP is a more likely indicator of where his performance will end up. Still, that’s a quantum leap above his 2011 performance, and significantly better than even his 2010 xFIP (3.98). I would expect McDonald’s walk rate to tick back up toward 8.5-9.5%, as he still doesn’t possess great control, and his homer rate and BABIP are sure to regress toward their 2011 levels (perhaps not all the way back, but at least halfway), but the strikeout ability seems to be legitimate, and that’s enough to make McDonald a quality mid-rotation arm in the "occasionally electric" Brandon Morrow mold.
If I had asked you in, say, February, if Jeff Samardzija’s career was a major disappointment given his huge signing bonus, you probably would say yes. If you asked me that question, I certainly would have said yes.
Samardzija was worth all of 0.3 fWAR in his first four seasons, mostly pitching out of the bullpen. This year, he’s a starter, and he’s suddenly been worth 1.8 WAR. Unlike Hammel, who had two great years under his belt, and McDonald, who had been very solid in the second halves of both 2010 and 2011, Samardzija’s success seems to have sort of come out of nowhere, as it seemed so impossible for him to do this as little as four months ago.
But in calling him a bust, we were missing a trend. Here’s Samardzija’s career swinging strike rates:
This is reflected in his strikeout rates:
Heck, even his fastball velocity tags along:
2008: 94.7 mph
2009: 93.8 mph
2010: 93.3 mph
2011: 95.1 mph
After years of going nowhere in the majors, Jeff Samardzija suddenly acquired the ability to miss bats in 2011. He gained two mph on his fastball, three on his slider, and one on his splitter, and left alone in a bullpen role for a whole season instead of being jerked around between roles and teams, he started getting results.
Samardzija throws four pitches: a four-seam fastball, two varieties of slider (the harder of which I’ll call a cutter for this), and a changeup. He occasionally works in a two-seamer, but for my purposes here, I’ll just roll it in with the fastball.
Just eyeballing his 2011 PITCHf/x numbers, one could tell Samardzija was doing something right. Namely, he was deploying his offspeed offerings with excellent effectiveness:
Fastball: 6.5% SwStr
If we’ve learned anything from Hammel and McDonald thus far, it’s that even players who totally catch us off-guard with improvement haven’t necessarily revamped everything. Hammel added a two-seamer, McDonald a slider, and both kept everything else about their arsenals and approaches almost exactly the same. That was all they needed to go from #5-quality starters to #2/#3-quality arms.
Samardzija’s 2011 performance with offspeed stuff shows he wasn’t that far off from some pretty dominant numbers of his own. What separated him from them at the time? A 58.3% strike rate with his fastball, which led to a 13.2% walk rate. When he got strike one, batters got on base at a .233 clip. After ball one, it was .399. Of course, one would expect significant splits between those two situations, but his were even larger; the league as a whole OBP’d .383 after first-pitch balls and .265 after first-pitch strikes.
Samardzija’s 55.8% first-pitch strike rate last season was well below the league average of 59.4%. This meant a few things:
1.) He was putting himself into bad situations more often than the average MLB pitcher.
2.) In those situations, he had to lean on his (statistically) worst pitch, the fastball.
3.) In those situations, he had to eschew his most effective swing-and-miss pitch, the splitter, which relied almost entirely on out-of-the-zone chases (just 3% called strikes, 52% overall strikes)
Thus, it’s not hard to see why Samardzija’s splits after first-pitch strikes and balls were so dramatic. Up 0-1, he could start throwing his dynamo offspeed pitches in situations where hitters would have a tough time laying off; behind 1-0, he had to rely heavily on an inconsistent fastball hitters could sit on, often missing with the pitch and digging himself in deeper.
Viewed through this lens, the decision to convert Samardzija back to starting starts to look inspired. Yes, starting is harder than relieving, and Samardzija wasn’t a good reliever. But things aren’t that simple, and for a vivid illustration of that, let’s look at John Hellweg’s 2011 season.
If you’re not a prospect hound or an Angels fan, you probably don’t know who John Hellweg is. He’s a 6’9" righthander who’s every bit as imposing as you’d expect a 6’9" righthander to be, sitting in the mid-90s with an effortless motion that gives him extreme downward plane.
Like many tall pitchers, though, Hellweg has all sorts of trouble throwing strikes, which naturally led the Angels to put him in a bullpen role in the minor leagues. Even there, though, he walked well over a batter per inning in fourteen relief appearances in High-A to open the 2011 season. That’s the sort of wildness that can cause a team to release a guy or send him back to extended spring training, but the Angels opted for a different course—they moved him to the rotation.
And the rest of the way, John Hellweg was a totally different pitcher. His ERA as a reliever was 7.71; as a starter, it was 2.12. His strikeout rate was just over 11 K/9 in both roles, but his walk rate dropped from 10.87 BB/9 as a reliever to 3.96 as a starter. In May 2011, he was on the verge of total breakdown; in September 2011, he was often cited as the top pitching prospect in the Angels system, or certainly in the top three.
Relieving allows hard, wild throwers to "air it out," and it’s possible to be an effective reliever walking five batters per nine innings, while that feat is much more difficult as a starter. But that doesn’t mean wild pitchers like Samardzija and Hellweg are necessarily better off in the bullpen. Yes, starting seems to be at odds with the profile of a pitcher like this, but that’s the point. The role can force these guys to calm down and pace themselves, because they’re not going to be able to reach back for their peak velocity with every pitch.
With Samardzija, that certainly seems to be the case. His fastball went for strikes just 58.4% of the time last year; this year, it’s at 66%. He’s also pulled his first-pitch strike rate up to a near-league-average 59.4%. The fastball’s swinging strike rate is even up to 7.7%, so overall, it’s been far more effective.
And that’s really all that was separating Samardzija from stardom in 2011—a consistently located fastball. Its success has had a couple of side benefits as well: first, he doesn’t have to use it as much, since he is more consistently ahead in counts. After using the heater 58.8% of the time last year, he’s down to 54% this year, freeing up more room for the trio of outstanding offspeed offerings. Second, those offerings play even better off the fastball—in particular, the splitter has emerged as a lethal weapon, drawing a 25.8% swinging strike rate.
It wouldn’t surprise me to see Samardzija post low-3’s ERA-quality pitching the rest of the way. With his fastball command problems seemingly in the rearview mirror, he’s a guy with mid-90s heat, a deadly splitter, and two solid breaking balls who gets strikes with all four offerings and also gets a nice percentage of groundballs. That is one hell of a profile, and it’s one hell of a turnaround for the righthander.
I’ve relied mostly on PITCHf/x data for this article, but now I’m talking about R.A. Dickey, so everything basically gets thrown out the window.
I don’t know if you realized this—I sure didn’t, until recently—but R.A. Dickey has a 12.6% swinging strike rate this year. The swinging strike rate leaderboard looks like this:
1. Cole Hamels, 12.6%
1a. R.A. Dickey, 12.6%
3. Stephen Strasburg, 12.3%
That…that really exists. On a measurement of ability to miss bats, a 37-year-old knuckleballer comes out directly ahead of the most electric pitching talent of this generation, and way ahead of other stuff stalwarts such as Justin Verlander (seventh), CC Sabathia (eighth), Matt Moore (ninth), Samardzija (tenth), and Clayton Kershaw (20th).
Unlike the other three pitchers I’ve discussed, R.A. Dickey was quite good in 2011—he was worth 2.5 WAR, posting a 3.28 ERA and 3.77 FIP in over 200 innings. But also unlike those three, Dickey seemed to have little room to improve—he’s far, far past the normal age for any sort of improvement to take place, and HE’S A FREAKING KNUCKLEBALLER. It’s one pitch. He lives and dies by that one pitch. He’s lived and died by that pitch for seven years now. One would think that one pitch was as good as it was going to get.
There are some weird trends with R.A. Dickey, though. Check out his career fastball velocities (not including his rookie 2001 season, for which there’s no data):
*Comes from only one start
With the exception of the one game in 2006 and an uptick from 2010 to 2011, it’s a downward trend.
Now, let’s look at his knuckleball velocity:
When I was growing up, my biggest exposure to knuckleballing was, of course, Tim Wakefield. Wakefield, of course, threw a mid-60s fluttering knuckler. The only other times I saw the pitch (excluding the odd inning of Steve Sparks or Dennis Springer that I'd happen to stumble upon) were in video game form. With the cartoonish depictions of the pitch in ‘90s video games, where the thing would float in at 53 mph fluttering all the way across the TV screen, I always had this image of the knuckleball as almost an eephus, just floating in over the plate.
That’s why, every time I watch R.A. Dickey pitch, I’m always kind of startled at first. His knuckleball doesn’t take three hours to get there—it comes in faster than a lot of curveballs and even some changeups. It comes in about as hard as Jamie Moyer’s fastball!
When he first started throwing the pitch in 2005, it was nearly 20 mph slower than his fastball. Nowadays, it’s about six mph slower. R.A. Dickey has added ten mph to his knuckler in the past seven years. If he threw a changeup, it would probably be slower than the knuckler. In my head, I'm putting exclamation points after all these sentences, but I'm not actually putting them in, because that would be weird, right?
What do these weird trends mean about his 2012 success? I have no idea. He’s only throwing the flutterball 0.7 mph harder than he did in 2011, so while the velocity trend was interesting enough for me to just go on that long tangent about it, it’s probably not all that relevant to why he’s succeeding.
Why is he succeeding? Well, it doesn’t take a genius to guess "The knuckleball’s being more effective." Indeed it is—the pitch drew 9.7% whiffs last year, but it’s at 14.1% this year. After turning to his fastball nearly a quarter of the time last season, he’s cut its usage almost in half in 2012, so his knuckler usage is up from 75% to 85%. Given that it misses far more bats than the faceless fastball, that helps his overall swinging strike rate as well.
As for why the knuckler is succeeding more? I have no idea. That’s the eternal mystery of the knuckler—because it’s basically a spinless ball, it renders much of PITCHf/x useless. Most knucklers are going to show up as moving very little on PITCHf/x (as they should, or else they weren’t thrown well), and any chart of their locations is mostly going to look like a randomly scattered array of dots.
Given that Dickey is throwing the pitch harder than ever before and more than ever before, and he continues to learn in his seventh year of knuckleballing, perhaps it makes sense that he’s taken a big step forward. Given that it’s just half a season, I would remain cautious and expect his strikeouts to snap back somewhat. Certainly, though, it’s hard not to marvel at Dickey’s accomplishments.