Yu Darvish, Now Throwing Harder

Jack did a good job this morning dissecting the adjustments Yu Darvish has made and how those have paid off in a big way, but there was one thing he didn’t mention that Darvish is also doing now – throwing harder.

Here are the average velocities for Darvish’s pitches from 2012 and then last night, per PitchF/x:

Season Team Four Seam Two Seam Cutter Splitter Slider Curve Change
2012 Rangers 92.6 92.7 89.7 87.2 81.6 70.9 82.4
2013 Rangers 94.1 95.2 91.0 85.6 82.7 69.9  

The MLBAM classification algorithm isn’t perfect, so just for comparison, here are those same numbers according to Harry Pavlidis, who manually classifies pitches at BrooksBaseball.

Season Four Seam Two Seam Cutter Splitter Slider Curve Slow Curve Change
2012 93.6 93.3 90.4 89.0 82.9 79.6 70.3 87.4
2013 95.4 94.8 91.3 85.4 84.0 82.1 70.6 86.3

We can also show this graphically. Here’s the spread of all pitches that Darvish has thrown that have been classified as four seam fastballs by PITCHF/x.


You’ll note that not only is the point denoting average velocity higher for last night’s game than it was for any point last year, but that the maximum velocity is also higher than it was at any point during last season. Darvish threw ten pitches that topped 95 last night, and three times, he got over 97.

Perhaps the most remarkable point of all of this is that April velocity is lower than seasonal velocity, and most pitchers throw harder as the season progresses than they do at the start of the year. And yet, according to the PITCHF/x data, Darvish was throwing harder last night than he did in any single start last year. In his Opening Day start last year, Darvish only topped 95 three times, topping out at 95.8. He eventually started hitting 97 on occasion towards the end of the year.

But, again, this isn’t all about the fastball. Darvish’s velocity was up on most of his pitches last night, and as Jack notes, he relied heavily on his cut fastball and slider. Differentiating between those two types of pitches can be difficult for an algorithm, so we’ll just focus on Harry’s classifications for these pitches. When you compare April 2012 to last night, in order to account for the early season diminished velocity effect, you’ll see that Darvish’s cutter was recorded at 91.3 versus 90.7 last year, and the slider was 84.0 versus 82.8.

Now, just because this is PITCHF/x data doesn’t mean that small sample size warnings don’t apply. It’s possible that a calibration issue in Houston could be inflating the velocities relative to what the system records in Arlington, or it could just be that Darvish put a little extra mustard on his pitches as the game wore on, given the whole perfect game thing. Two of his three pitches recorded at 97 mph came after the sixth inning, though there’s no Verlander-style velocity increase as the game wore on.

Still, just because PTICHF/x has Darvish throwing harder last night doesn’t mean it’s a new established norm for him. His maximum velocity spiked in his fourth start of the year last season, then fell off in May. Velocity is fairly consistent from start to start, but that doesn’t mean that there’s no variation, and last night could end up being an outlier. Or it could be a measurement issue with PITCHF/x. It’s still one start. Never read too much into one start.

That said, there are reasons to think that it’s probably partially real, at least. Since the Rangers spring training complex has PITCHF/x cameras installed, we also have data from March for each of the last two years, and Darvish’s 2013 spring training velocity is up compared to his 2012 velocity as well. And, of course, we have the benefit of knowing just how hapless the Astros looked against Darvish last night, and even after accounting for a weak opponent, the results align with the idea that Darvish had his best stuff last night.

For all the talk about how Houston’s line-up compromises the impressiveness of Darvish’s performance last night, keep in mind that he opened 2012 against the Mariners, and they tagged him for five runs in 5 2/3 innings. He struck out only 5 of the 30 batters he faced on Opening Day last year, versus 14 of the 27 he faced last night. That’s two straight years he’s opened against a bad line-up, but his results from last night were far more impressive.

Both from a scouting and results perspective, Darvish was at his best last night. The question now is whether he’s going to keep throwing harder the rest of the season, and whether or not he can actually gain velocity from this current point as the season wears on or if he’s just starting from a higher base. After all, the longer spring training — due to the WBC — may have allowed pitchers to build up more arm strength than usual by Opening Day, so we can’t say for certain that pitcher velocity trends will follow previous years patterns. However, even at last night’s registered velocities, Darvish’s projections would likely have to be revised upwards.

Steamer used a projected 92.8 mph FBv for Darvish as their velocity input, and it still spit out a 3.25 FIP projection despite the fact that he pitches half his games in Texas. As those guys have shown, increasing fastball velocity improves a pitcher’s projection and makes for more accurate projections, and Darvish was throwing well above that 92.8 average last night. If he spends the year throwing closer to 94 than 93, then Darvish may actually be underrated by the preseason forecasts. And they already forecast him as one of the game’s best starting pitchers.

It’s still one start, and one start against the worst team in baseball. Maybe Darvish gets rocked in his next start while throwing 92 and we throw this whole post in the trash. Small sample warnings cannot be overstated at this time of year. This might not mean anything.

But, you know, on the chance that Darvish’s velocity is trending upwards, well, look out, American League.

Print This Post

Dave is the Managing Editor of FanGraphs.

21 Responses to “Yu Darvish, Now Throwing Harder”

You can follow any responses to this entry through the RSS 2.0 feed.
  1. Aggie E says:

    Nice add-on column Dave. Maybe the comfort level after 1 year in the MLB could see an overall improvement in not just BB/9 but also stronger pitches..

    Vote -1 Vote +1

  2. Primi timpano says:

    Perhaps throwing fewer fast balls and more cutters and sliders allows Yu to throw harder. To me, the impressive part of his performance was the lack of walks.

    Vote -1 Vote +1

  3. rustydude says:

    On 8/15/2012, Felix had more velocity on his pitches, as well, during his perfecto! So that seems like it might be a thing.

    Vote -1 Vote +1

    • Dan Rozenson says:

      It could also be a result of a pitcher trying his hardest when he knows he has a perfect game going. Felix’s first 15 fastballs that day averaged 93.0 mph, while his next (and last) 19 averaged 95.1. That is a significant difference.

      Vote -1 Vote +1

  4. Dan Rozenson says:

    Another thing to keep in mind when comparing Harry’s data with the raw version is that whereas PITCHf/x starts tracking at y = 50 feet from home, Harry tries to simulate a more realistic release point of about 55 feet. So comparing MLBAM and Harry’s data in each year is tricky, but comparing year-to-year within each system is more safe.

    Regarding calibration errors, it is possible the gun was off, but it’s usually not off by that much. In both of the games at Minute Maid Park this season, the spin deflection has been suspect but the speed seems about right.

    Vote -1 Vote +1

  5. Shao says:

    For his command issue, I have watched him pitching for years and I thought not many people notice that he was actually rushed at age 25. He dominated NPB for years, which people generally recognized as AAA equivalent. But after the first game he pitched in the big league, he said NPB is not even comparable to AA. He has never been a master of command but I don’t think he is that terrible like what he was last year. He is like a pitcher that dominates A for years and is rushed to major and suddenly finds that hitters at this level can easily send his fastball to the moon if he throws in the old way he did in A. Then he becomes afraid of attacking the strike zone until he finds how to use his cutter.

    I think the scout Jack mentioned is absolute right. He was a thrower because he can easily overpower any hitter in NPB. He has no incentive and needs to think about how to mix his weapons at that level. He was mentally bullied and that caused his command issue last year. Ryu in Dodgers might see the same problem after he got 10 hits in his first game in the major.

    I am happy to see Darvish find his way and I have no doubt he will one day throw a true perfect game in the major.

    Vote -1 Vote +1

    • wobatus says:

      I have a lot of doubt that he will pitch a perfect game in the majors. The odds are quite low.

      Vote -1 Vote +1

  6. CJ in Austin Tx says:

    I’m not sure I would use the next game to test whether Darvish’s velocity remains high. Because of the potential for a perfect game, Darvish was allowed to throw a lot more pitches than he would normally be permitted to throw in his first start of the year. I think it’s possible he may be feeling the effect in his next start.

    Vote -1 Vote +1

    • SucramRenrut says:

      111 pitches is nothing for a near complete game. it’s not like he threw 135. I’m surprised Washington didn’t let him face the next guy considering there was still a CG SO going, and he may have only needed one pitch. But hey, who knows what goes through Wash’s bean?

      Vote -1 Vote +1

      • Aidan says:

        The key part of this is the timing. Some pitchers are better prepped for the beginning of the year, but the workload Yu put up was near three standard deviations above what he had done over his last 5 spring starts, and about 3 dozen pitches more than his highest spring count.

        Vote -1 Vote +1

      • Craig says:

        We know what goes through Wash’s nose…

        Vote -1 Vote +1

      • CJ in Austin Tx says:

        Darvish’s pitch limit going into the game was 90 pitches, according to the Rangers’ broadcasters. They pointed out that he didn’t throw more than 75 pitches in a spring game.

        Vote -1 Vote +1

  7. Jason H. says:

    After one start CC is finished and Yu is now unhittable. I wonder if we could take any given start by a starting pitcher and find some outlier with respect to some aspect of their performance that day, and then write an article about “a possible trend to look out for” with an obligatory nod to SSS.

    Vote -1 Vote +1

    • Andre the Angels Fan says:

      Why are you even reading fangraphs at this time of year?

      Vote -1 Vote +1

    • jim says:

      it’s like you didn’t even read the article

      Vote -1 Vote +1

    • Brad says:

      You make it sound as if Darvish was hittable prior to his near perfect game. Guess what, he wasn’t very hittable before either.

      Vote -1 Vote +1

      • Jason H says:

        Well, he’s given up hits in all of his starts, so he’s certainly hittable. Whether he is “very hittable”, I suppose, is a matter of criteria.

        Vote -1 Vote +1

  8. lexomatic says:

    I think it’s obvious that the reason for the difference in velocity is the innate differences between PTICHF/x and PITCHF/x

    Vote -1 Vote +1

  9. J says:

    In the article you state that most pitchers throw harder as the season goes on. Throw harder or have more velocity? I have always been under the impression that pitchers wear down over a season, and on the graph you show, Yu’s first April start of 2012 appears to have a higher avg. velocity than any start in his second half except for one. It seems avg velocity increases during the first half of the season, as pitchers loosen/warm up and then slightly decreases on avg during the second half due to wear, fatigue, heat, etc. Am I way off here? Thoughts?

    Vote -1 Vote +1

  10. vivaelpujols says:

    Why would not just check out other pitchers in this game and see if they were also throwing harder? Then we wouldn’t have to guess about the park calibration issues.

    Vote -1 Vote +1