Rubbing Mud: Carlos Martinez, Tunnels, and PECOTA

In one sense, Cardinals right-hander Carlos Martinez is an easy pitcher to understand. He can touch 100 miles per hour with his fastball. He throws both a four-seamer and a sinker, has a slider and a changeup to go with them, and all four pitches could be counted as above average. He’s fiercely competitive and a great athlete. Bob Gibson was a bigger guy than Martinez at a time when everyone else on the field was smaller. Gibson had only two dominant pitches, and rarely even bothered with others. He’s also a Hall of Famer. Still, it’s really hard not to compare Martinez to Gibson.

In another sense, though, there’s a whole lot we don’t know about Martinez. No, that’s not true. We know a ton about Martinez, far more than we would have known 10 years ago. Yet, we would have been much more confident in our assessments of Martinez then than we are now. Sometimes, even valuable new information only makes the essential truth about something feel further beyond our reach.

To wit: PECOTA is not all that sure that Martinez is very good. Despite his 3.02 ERA and 6.7 WARP in 375 innings over his two seasons as a starter, the projection system forecasts a 4.19 ERA and 1.7 WARP in just over 175 innings for Martinez in 2017. Our pitching value metric, Deserved Run Average, had Martinez just over 13 percent better than a league-average starter in 2015, and less than two percent better than average last year. Our estimator of true talent, cFIP, actually put him on the wrong side of average in 2016.

Those numbers took into account that Martinez’s solid run prevention had a lot to do with the framing of Yadier Molina—he saved Martinez 4.2 runs in 2016, more than all but 14 other pitchers profited from catcher framing—because Martinez himself rates poorly when it comes to command (1.3 runs worse than average on balls not in play in 2016 according to our DRA Run Values table), and is in the 10th percentile of all pitchers with at least 100 innings pitched in Called Strikes Above Average.

Indeed, Martinez’s fairly high walk rate and occasional difficulty getting left-handed hitters out are holding him back from reaching the stratosphere populated by other guys with stuff as nasty as his. Still, there might be some things Martinez does well that only more descriptive stats can capture. In fact, I’m pretty sure there are. To get a practical sense of Martinez’s pitch usage, I went to Brooks Baseball and sought out the table in which that usage is broken down by handedness of opposing hitters for 2016.

As you can see, Martinez uses his changeup about 60 percent more often than his slider against lefties, but the slider about four times as often as the changeup against righties. No surprise there; that’s Pitching 101. You’ll also note that he throws his four-seamer quite a bit more than his sinker against lefties, and his sinker slightly more than his four-seamer against righties. That finding is a bit more interesting.

One of the reasons why most pitchers use their breaking balls more against same-handed batters and their changeups more against opposite-handed ones is that breaking balls tend to move to the glove side, away from a same-handed batter, whereas changeups tend to move to the arm side, away from opposite-handed ones. By the same logic, one might guess, a pitcher would throw his four-seamer more to same-handed hitters, because that pitch moves much less to the arm side than does a sinker or two-seamer. Martinez is no exception to those stereotypes, by the way:

I imagined, for a moment, that this might be a unique quirk, something that would set Martinez apart from almost anyone. Nope. Seeking out similar pitchers (use both the four-seamer and the sinker often, right-handed, have a changeup and a breaking ball), I found that they followed a similar pattern. Rick Porcello, Jeremy Hellickson, and Yordano Ventura all were less extreme examples of Martinez’s plan of attack. I then wondered whether Martinez was using the four-seamer to get in on the hands of left-handed batters. Here’s the Brooks Baseball plot of all his four-seam fastballs against lefties, for 2015 and for 2016:

So, not really. He did start to do that more last season, but it’s not even clear that it was intentional. Mostly, Martinez throws his four-seamer up and away from lefties, usually above the zone, trying to induce swings and misses. As uses for a four-seamer against opposite-handed hitters go that’s a pretty smart one, and it’s very effective for Martinez. It certainly isn’t a plan for which the sinker is suited.

The scope of that utility is fairly small, though. Few batters will chase that pitch unless the count is already at two strikes, and it’s a tough pitch to trust in a 2-2 or 3-2 count. Far more common, it would seem, are the situations in which a sinker that starts in the middle of the zone and ends up low and away would induce an ugly swing and, at worst, a weak ground ball. (He does do this, as you might have noted in the usage table above, to try to get grounders in hitter’s counts.)

Conversely, Martinez does throw the sinker mostly in on right-handed batters, trying to saw them off or get them to swing over the top. The pitch really is nasty, so it often works. Truthfully, though, I don’t think his fastball preferences based on the handedness of opposing hitters are about the fastballs themselves. It seems to me that they’re really about the pitches before and after those fastballs.

On our generalized pitch tunneling tables, you won’t find Martinez near the top of any leaderboards. He doesn’t repeat his release point exceptionally well. On average, there isn’t an overwhelming difference between the locations of consecutive Martinez pitches when they reach the tunnel point, nor in the distance that they break between that point and the time they reach the plate, nor in their flight time between release and when they reach the tunnel point (the way our tunneling data controls for velocity changes).

We use the ratio of post-tunnel point break to differential in location at the tunnel point as a shorthand for tunneling efficience. In essence, the more two given pitches diverge after the tunnel point, and the less differentiated they were when they reached that point, the better the chance that the batter will be fooled and unable to adjust. And Martinez is ordinary in that regard. He seems, through the wide-angle lens, to be a guy who has trouble disguising his offerings and using sequencing to baffle opposing batters.

That’s not really the highest use of this data, though. It works for players at the extremes. It shows you how much John Lackey’s variation of arm slots stands out from other starters. It shows you how much late break Clayton Kershaw has (the most of any pitcher with at least 1,200 pitch pairs in our tunneling database last year), and that Jon Lester repeats his release point better than anyone else in baseball, and that Bartolo Colon gives hitters the closest approximation of Greg Maddux’s icky column of milk (Colon had the lowest average tunnel differential of any qualifying hurler last year).

For most pitchers, though, the real insight tunneling provides can only be found when you break down those pitchers as individuals, and specifically as a collection of pitch pairings. Here’s the tunneling data on every two-pitch sequence Martinez used at least 100 times last season:

Carlos Martinez Pitch Pairs, 2016

Brk:Tun = Break:Tunnel Differential | Pl. Diff = Plate Differential | Rel Diff = Release Differential | Tun Diff = Tunnel Differential | FT Diff = Flight Time Differential | P-T Brk. = Post-tunnel break | Rel:Tun = Release:Tunnel Differential

I’ve ranked these by the break-to-tunnel ratio, because that’s the best estimator (if we must pick one) of the efficiency of a sequence in these situations. Two pitch pairs interact really well for Martinez, according to that stat: the four-seamer and slider, and the four-seamer and changeup. That’s in a vacuum, though, and again “in a vacuum” is a disqualifying qualifier if you’re trying to do good analysis of tunneling.

Martinez’s break-to-tunnel for a [four-seamer, slider] sequence ranked 44th of the 109 pitchers who used that sequence at least 100 times last year. He ranks 29th if you sort that same list by sheer post-tunnel break, because his fastball hums and rides to the arm side, while his slider is a big sweeper. However, he also had the sixth-widest differential between the two pitches at the tunnel point, meaning that batters had a better chance to pick up the slider (coming just after the four-seamer) off Martinez than almost anyone else in the league.

Meanwhile, of the 54 pitchers who used a [four-seamer, changeup] sequence at least 100 times last year, Martinez’s Break:Tunnel ranked eighth. He had the 10th-smallest tunnel differential of those 54 pitchers. Here, I’ll stop to infuse some Statcast information. Martinez’s four-seam heater had an average spin rate of 2,130 rpm, which ranked 203rd of the 235 pitchers who threw at least 400 four-seamers. That’s roughly the story with Martinez, most of the time. His sinker (Statcast calls it a two-seam fastball) spun at an average rate of 2,046 rpm, good for 88th out of the 107 pitchers who threw at least 300 of those. His slider’s spin rate was 2,145 rpm, 123rd of the 144 guys who threw at least 300 sliders or curves (Statcast calls Martinez’s slider a curve; Statcast doesn’t know this guy well).

He’s a low-spin pitcher, is what I’m saying. It’s why he gets such great natural sink and induces a lot of ground balls. However, his changeup’s spin rate was 2,010 rpm, 24th-highest of the 119 pitchers who threw at least 200 changeups last year. So, whereas many hurlers get tumble on their changeup by throwing it with much slower spin than their other pitches, Martinez is among those who manipulate the spin axis instead. The particular spin axis Martinez achieves allows his change to move dramatically to the arm side:

(I mean, did you see that?!)

Because the pitch still has so much spin, and because some of that is backspin, Martinez’s changeup doesn’t sink much. Now, we return to the smallness of the tunnel differential when he throws a four-seam fastball, then a change. That comes from the fact that he throws both pitches with similar spin. Because the fastball is thrown so much harder, and because its spin axis is about 50 degrees less than the changeup’s, the four-seamer rides higher and runs much less.

The same can’t be said of Martinez’s sinker-change sequence. You might have noticed that that sequence doesn’t even appear in the table above. It’s no accident. His [sinker, changeup] pairing simply isn’t anywhere near as effective as [four-seamer, changeup], from a tunneling standpoint. He actually has a smaller release and tunnel differential for the [sinker, changeup] sequence than for [four-seamer, changeup], but the deception there just doesn’t pay off the way one might hope.

Martinez achieves a much bigger velocity gap between fastball and changeup with the four-seamer as the set-up pitch. The post-tunnel break differential is also more than twice as large for the sequence led by the four-seamer as for the one led by the sinker, so even if a batter isn’t perfectly fooled, he might underestimate just how far from the four-seamer’s path will veer, and swing and miss anyway. That alone is enough to explain why Martinez throws his four-seamer to lefties so often. It sets up his changeup in a way neither of his other pitches can.

A similar thing is happening with his sinker and slider. Overall, he still throws the slider on the back of his four-seamer more often than after his sinker, and we don’t have breakouts of this data by handedness (yet!), but I think it’s safe to say that if we could isolate plate appearances against righties, we’d find more instances of [sinker, slider] than [four-seamer, slider]. The four-seamer and slider are each electric offerings, but they diverge too soon. The average tunnel differential when Martinez throws [sinker, slider] is about 15 percent smaller than when he throws [four-seamer, slider], which is a huge difference. It’s the difference between being one of the easiest slider-after-fastball guys to pick up in baseball, and being a bit above average at keeping hitters in the dark on that front.

The spin axes of the sinker and slider are 180 degrees or so apart, which gives hitters something to try to pick up right out of the hand. Despite its filthiness in isolation, Martinez’s sinker isn’t a perfect pitch because of the way it interacts with his other ones. Now, let’s duck under the rope and back out into the light of day, and remember that the pitch does plenty in isolation to be a worthwhile part of a repertoire. Besides, four pitches turn over a lineup better than three. If Martinez can use the four-seamer and slider on a righty the first time through, for instance, that hitter might be looking slider and think he has it—only to be screwed into the ground by a sinker running in and diving toward his kneecap. I’m not speaking hypothetically, either. Martinez really does that:

One last observation: Martinez seems to have a hard time going from his breaking or offspeed stuff to his fastball, especially the four-seamer. His release and tunnel differentials tend to be much wider when going from slow to fast than when going from fast to slow. That’s telling.

If you followed all of that, you have a much more nuanced idea of how Martinez pitches, and why. There’s considerable danger in that. You might get overly confident in him, because everything we come to understand better we come to like better. We trust the thing more. We tend to believe we now understand the thing perfectly, but often we face another round of adjustment, re-learning, and new discovery just around the next corner. Martinez needs better command. His effective velocity is considerably less than his raw velocity, because he’s a short guy with little extension. He could benefit from pitching more under control, physically, as evidenced not only by his walk rate and by the large distances by which he sometimes misses within the strike zone, but by those struggles in repeating his delivery when ramping up from an offspeed pitch to his fastball.

Martinez pitches with a plan. He uses his four-seam fastball to set up opponents, but his sinker to violate their expectations and to bail himself out when the pressure is on. He gets to his slider by throwing his sinker for strikes. This does help me understand him. His approach yields a maximum of early ground balls and leaves some strikeouts on the table, and while DRA might not love that approach I see the merit in it. If Martinez can stay healthy, I think he’s a better pitcher than the one PECOTA sees. He has such a deep and devastating repertoire, and such a good idea of the best way to deploy it, that his ERA seems a fairer reflection of the player he is than DRA does.