Baseball, like many things, does not always yield answers easily. While fans and analysts have generally been good at describing what happened, the why of something happening is tricky. Over the past few years and decades, the industry and third-party observers have added stats upon stats and reams upon reams of baseball-related data. Some of these lend themselves well to correlating the whats and the whys, but much like most non-scientific data, definitively evidencing causation tends to prove elusive. That’s really another way of saying: if you’ve accessed this article to find the smoking gun behind Julio Teheran’s recent run of success, I’m afraid you’ve come to the wrong place. Instead, what I present here is just an idea, and far from a perfect one. As time goes on, we will get more data, which will either refute this idea, or provide more evidence in support of it. Either way, there probably won’t be anything definitive. Success is fleeting and multifaceted, and perfectly capturing the complex relationships that deliver it may not be in our grasp. So it goes, but we struggle against the tide anyway, and have fun doing it.
Last week, I asked the TC community to give us their thoughts on Teheran’s recent good run. We got some (one) fanpost
s. I also got some emails, and I’d like to thank everyone that loaned their thoughts to this topic and shared it with us. The main things noted were:
- Command — but this is very difficult to verify, without access to ball-hits-glove data, which are not (yet?) public; and
- Pitching more like Marco Estrada (more fly balls) as opposed to less like Marco Estrada;
What I present below is not really all too different from these. I think it’s the same nexus. Does that mean this is definitely something? Well, maybe. I’d still love some ball-hits-glove data, no doubt about that.
Before discussing how Teheran might be doing what he’s done, it’s important to clarify specifically what he’s done. Why is this important? Because if you inaccurately describe the outcome, trying to tack on a relationship to that outcome can lead to problems. So, here’s the “difference” in Teheran that I think we’re trying to capture.
Going back to 2015 (conveniently, the first year that good Statcast data on things like xwOBA are available), Teheran has alternated good and bad years. His 2015 was the worst year of his career (104 ERA-, 117 FIP-, 108 xFIP-). But then, in 2016, he had the best year of his career (77 ERA-, 92 FIP-, 100 xFIP-). Last year, it was back to 2015 with a nadir-type performance (104 ERA-, 116 FIP-, 115 xFIP-).
To start 2018, Teheran issued two clunker starts. On Opening Day, he allowed four runs with a 3/3 K/BB ratio against the Phillies. The performance was far closer to his 2017 aggregate than his 2016 aggregate. Then, he followed that up with one of the worst starts of his career, getting mauled by Washington for five runs (three homers) with a 2/3 K/BB ratio while getting just seven outs. But then, something happened.
His next two starts after that were good: three runs in 12 total innings, with just one homer allowed and a 16/7 K/BB ratio. After that, he got even better: seven shutout innings, 6/1 K/BB. I’m choosing to skip his “injured” start against the Phillies, both because it’s not clear why he was even making that start in the first place, and because what he did after that just continued his trend from the start before. In his most recent starts, he’s fired 13 scoreless innings, with a strong 13/2 K/BB ratio. In his most recent outing, he didn’t issue a single free pass, the first time that’s happened since August of last year. If you exclude his injured start, he hasn’t allowed a run in three straight outings. I know that’s technically not a “real” stat, but he’s actually never gone three straight outings without allowing a run before, so that’s kind of interesting.
You can also quantify his recent run of success via Game Score, which you can glance at here. Again, excluding his injured start, Teheran has now had three straight outings with a Game Score (version 2) of 70 or above. He’s never done that before either, though there was a really good stretch in June 2016, another in May 2016, and some other stellar spans in early 2014 as well. If you go back further (and keep excluding the injured start), he’s had five starts in a row with a Game Score of 50 or above, indicating an average or better start. This isn’t the first time this has happened (he was really good between April 20 and May 24 in 2016) or even the second/third time (five starts in August 2015; April-May 2014), it’s still noteworthy.
So, there’s your “what.” Teheran’s 2018 has been good, aside from his first two starts, so far. He was also good in 2016. He was not good in 2015, nor in 2017. For an exploration of the “why” to make logical sense, the cause has to be:
- Present in 2016 and/or his recent 2018 starts; and
- Not present, or diminished, in his 2015 and 2017 starts.
If those two criteria aren’t met, then the relationship is pretty vague — probably too vague to serve as a sound conclusion.
There’s one other thing I want to talk about, with regard to the “what,” and it’s something that comes up a lot. Beside Teheran’s performance, which has bounced back and forth as detailed above, he’s had one trend so far: his velocity. In short, it looks like this:
In short, Teheran’s fastball velocity has declined somewhat each year. The declines aren’t large, but we know that fastball velocity is at least reasonably important for success (both intuitively and through past research), so the fact that Teheran had a better season in 2016 despite an average velocity decrease, and is doing so again despite a big velocity drop is noteworthy (and no, the “injured” start does not account for the entire velocity decline; Teheran’s average fastball velocity in his most recent start was 88.8 mph). And, while yes, Teheran does tend to pick up velocity as the season goes along, his fastball has never been this weak before. The point is, whatever the “why” explanation is tagged as being, it needs to be cognizant of this downward velocity trend, not so much as part of the explanation, but just for the sake of nuance and holism.
The Why: A Proposal
So, here’s my attempt at venturing an explanation for Teheran’s 2016 and recent-starts-in-2018 success. I looked at a few things before settling on this explanation one of them was whether he had finally managed to curb the damage lefties were doing against him. What I was hoping to see was that better success against lefties was the reason for his success in 2016 and recently, because we know that it proved to be his undoing in 2015 and 2017. That case, well, it’s hard to make. So I moved on.
(You can see that sometimes his FIP by handedness moves in lockstep, and other times it doesn’t, but there’s no real trend of “his FIP is the same against lefties or righties and it’s only one fluctuating split that makes a difference.)
So, without handedness, I moved on. Specifically, I moved on to pitch location. For comparison, this is what the “composite” Teheran’s pitch location looked like, for 2015-2017, by handedness.
These heat maps are from the catcher’s perspective. For lefties, you see essentially a strong focus only on pitches down and away, though towards the strike zone edge and drifting somewhat into the zone. For righties, that focus is actually more middle-away and radiating both to up-and-in and down-and-away. For lefties, both the inner part of the plate and up-and-away were not popular targets; for righties, Teheran avoided the very top of the zone as well as down-and-in pitches.
But, as indicated above, we know that 2016 was definitely not the same as 2015 or 2017. So, does splitting these locations by year tell us anything?
There are a lot of differences, of course. What do you notice, when comparing 2016 versus 2015 and 2017? For lefties, I notice more pitches in the zone (especially low and away), but also a tendency to use more of the strike zone, especially up. For righties, I don’t know if the approach is too different. The same point still seems to be targeted, though 2017 seemed to differ in that there was less consistency and command, as both the light halo way outside the strike zone became more pronounced and even the locations in the zone varied, with a lot drifting to right down the middle. We could look at 2018 as a whole, but with a small sample size accrued to date, it tells a fractured, too-entangled story.
As noted up above, 2018 as a whole doesn’t do a hypothesis much good. We’ve seen two poor (2015/2017-esque) starts, and starts afterwards that might be different in some way. So, we need to parse the heat maps by start, as done below.
Again, there’s a lot to unpack there. But here are my general thoughts:
- Starts 1 and 2 were probably not going to be successful, just based on command. In both, pitches to lefties didn’t really hit the zone enough to be worth offering at. In Start 1, balls to righties were also not going down-and-away so much as they were just going down, contrary to Teheran’s tendencies to date. In Start 2, he was filling up the zone against righties, and got hammered for it.
- Start 3 is where things start to evolve. The pitching to lefties is very erratic, but the most common location is well-placed, and there’s at least some integration towards the rest of the zone as opposed to only missing outside. Righties, meanwhile, featured a fairly pinpointed set of pitches, which is (probably?) a good thing.
- That lack of location variability to righties then happened for batters of both handedness in Start 4. Low and away to each was the order of the day.
- Start 5 is where things get more interesting. Gone is the pinpoint, small heat map nature of Starts 3 and 4. Lefties got more pitches in the zone, more of which were up. Righties saw an interesting tilted V shape, with more pitches up but also attempts to get hitters to chase both up-and-away and down-and-away.
- Start 7 is where we start seeing a very different set of locations. Everything is up. For lefties, anything belt-high appears to be fair game, with pitches no longer rolling down-and-away but rather just straight outside, or up-and-away. For righties, Teheran’s axis of attack completely changed. Up-and-in, or above the zone, were the foci. The low-and-away corner, a mainstay of his approach from 2015-2017 and every prior 2018 start, ended up being tossed aside.
- Lastly, Start 8 kind of looks like the result of mushing Start 7 with some of the earlier starts. Against lefties, it features the vertically-higher baseline of Start 7, with more emphasis on rediscovering that low-and-away corner. Against righties, the locations are still focused away, but away across all heights, not just low-and-away.
Since height appears to be the dominant difference, we can also look at these patterns using actual quantitative location data, rather than heat map data, courtesy of Pitch Info readings at Brooks Baseball.
This is just some quick data on Teheran’s four-seam fastball location. In this case, 0.00 reflects the vertical center of the strike zone, so negative numbers are below, and positive numbers are above. The dun shading reflects notably low average fastball locations, while the blue shading reflects notably high ones.
You can see that this array doesn’t actually really neatly align with the “first two starts have something different from the remaining starts” threshold I set forth earlier. Yes, the first two starts had average fastball heights either in line with 2017, or lower. And then the third start featured more elevation, commensurate with 2016. But then the fourth and fifth starts, which were still good, featured lower fastballs, with Start 5 really avoiding the sort of vertical locations Teheran had previously focused on with his fastball and really working it down instead. And then you come to his last two starts, which are consistent in that the average fastball location is much higher. It’s not a very neat story, unfortunately.
The below isn’t a neat story. It’s suggestive, but not definitive. As I said at the beginning: correlation, not causation.
This relationship also holds if you remove 2018, or if you remove just the last two starts of 2018. Are higher fastballs tied to Julio Teheran’s success? Well, maybe. We need more data to feel more strongly about it. But there is something suggestive here. If he had used higher fastballs, rather than lower fastballs, in his fourth and fifth starts of the season, the suggestiveness would be even stronger, but he didn’t.
There’s been a lot of discussion over the past year or so about high fastballs. The theory is that with the new, jumpy ball, and hitters being well aware of the launch angle revolution, high fastballs are a tonic to cure what’s ailed pitchers: balls lower in the zone getting crushed. To be fair, I’m not sure the narrative is this easy, but there’s a good Jeff Sullivan/Fangraphs piece here with some information on this front. Using Baseball Savant, I calculated that in 2015 and 2016, the rate of pitches in the bottom third of the zone or below was about 51 percent, which fell to 48 percent in 2017 and currently sits around 49 percent. wOBA and xwOBA on contact on those pitches shot up from 2015 to 2016, from .341 to .352. But, in 2017, it fell to .345. I don’t think that’s an easy narrative, either.
For Teheran, he was perhaps a bit late to this particular party. In 2015 and 2016, his rate of those same lower pitches was 46 percent. But, recall that 2015 was Bad Julio and 2016 was Good Julio. While the league was moving away from lower pitches in 2017, he was moving towards them, increasing his rate from 46 percent to 48 percent. He was rewarded with another Bad Julio season, but since there wasn’t any change in this rate between 2015 and 2016, I’m not sure the idea that he faltered in 2017 just because he threw more pitches lower really holds up logically. 2018 only makes this more complicated: his rate of pitches in this part of the zone was actually about 60 percent, much higher, through his first five starts. It only cratered in his last two. To that end, I think it’s probably more some location-specific calculus at play here than simply the proportion of pitches he’s throwing below the belt.
Still, there’s a reason the league has been moving, to some extent, to higher fastballs. For Teheran himself, throwing his harder stuff higher could have a lot of advantages, including playing to his outfield defense, avoiding any actual gains hitters have made in driving pitches at the bottom of the zone, and reinforcing his fly ball/pop-up tendencies. Most critically, though, throwing his fastballs higher may be an attempt to make up for whatever velocity he’s losing, to the extent that hitters find it harder to catch up with those pitches.
I looked through a fair bit of video and it was hard (for me, maybe not for you) to find a good example of the advantage of this kind of pitch. What I came back to was the below. This was a 90-mile-per-hour fastball that Teheran threw to Brian Goodwin. It wasn’t overpowering, and if it was down, it probably gets hit. I don’t know if it was the horizontal location, the vertical location, or both that made it work. But given that Goodwin swung right through it, maybe the vertical location had something to do with it?
In any case, we’ll see what Teheran does next. Maybe the “what” isn’t real, and he will go back to being ineffective. I hope not. Maybe the “why” of higher fastballs is spurious, and he will pitch differently going forward, and have success using something different. Or maybe, as messy as the data are, this is the start of a trend. As with pretty much everything in baseball, the beauty is between the interaction of the forecasts, the predictions, and the could-bes and what actually happens. This extra beauty is something I’m now excited to watch for in Teheran’s future starts.