R. Yeatts / Getty Images
Yu Darvish currently has the second-lowest xFIP (2.52) of any pitcher qualifying for the ERA title. xFIP is a version of fielding-independent pitching that normalizes for fluctuating home run rates league-wide, and is recognized as a good predictor of future ERA. You know that Yu Darvish is good – what you might not know, is that he’s also made an enormous adjustment since last year. According to Pitchf/x, Darvish has made the most significant change[i] with his mix of pitches from 2012 to 2013 – he’s throwing nearly 20% more sliders.
Not every pitcher needs to change things up – some are just fine keeping the same mix of pitches from year to year. See below for a comparison of how Darvish and Clayton Kershaw changed their mix of pitches from 2012 to 2013.
Kershaw has changed very little, and remains his dominant self – he has a 1.85 ERA thus far this year, a 2.58 FIP, and a 3.27 xFIP – meaning the stats say home runs will catch up with him at some point (and in fact, his home run per fly ball ratio has gone down from 8.1% last season to 5.1% thus far this year). Meanwhile, Darvish – who boasts a big repertoire – has sacrificed a bit from everything, but most notably his fastball, in exchange for more sliders. The natural follow-up question is: how is the experiment working out for him? A good way to answer: using linear weights.
Linear weighting is a measure of a pitcher’s runs saved by pitch; it can also be normalized on a per-100-pitches thrown basis (w/C). So even while Darvish is throwing his slider 20% more in 2013 than he did in 2012, he’s saving 0.57 more runs per 100 sliders – it’s a more effective pitch for him even though he’s throwing it more often. The change in pitch mix has also made his second-most popular pitch, the four-seam fastball, much more effective on a per-100 pitch basis – it’s actually made the fastball an asset, rather than a liability. (If you’re alarmed by the drastic drop in changeup effectiveness, he barely ever throws it, so don’t worry.) He’s losing effectiveness on the cutter, but even that pitch is still working in his favor. One of the reasons the new mix is more effective for Darvish is because hitters are making much less contact on him. Note that Darvish’s velocity hasn’t changed much; his slider is only up 0.2 mph from last year, fairly stagnant at 81.8 mph (though his curveball is 4 mph slower than it was last year – giving him a nice 93 mph-66 mph differential between his fastball and curveball).
Hitters are swinging at slightly more pitches outside the strike zone this year against Darvish, but the big difference is in contact rate – overall, hitters are making contact 7% less in 2013 than they did in 2012, and despite the change in mix, Darvish is still pitching in the zone as much as he was last year.
His slider is only called a ball 27.5% of the time, in contrast to 42.5% for his fastball. Hitters swing at it 52.3% of the time, in contrast to only 36.5% for fastballs; and they whiff on sliders 21.8% of the time, versus 12.4% for fastballs. He has a 41.75% whiff-per-swing on the slider, only bested by his curveball. For all his pitches combined, Darvish’s whiff-per-swing percentage is 36.1%, good for best in the majors.
As you’d expect, his K/9 has gone up from 10.40 to 12.28, good for best in the league. On batted balls, his line drive percentage is down from 22.2% to 17.0%, which makes a big impact on saving runs, too. But it isn’t all good – Darvish is giving up more fly balls, and his home run rate has increased even more than that – his home run per fly ball percentage has increased from 9.1% to 15.8%. While this is something that the xFIP metric corrects, Darvish’s home park is hitter-friendly; the Ballpark in Arlington sees 2.35 HR per game thus far in 2013, good for fifth-most in the AL.
While the home runs are disconcerting, that’s one bad development against many good ones: more strikeouts, fewer walks, and fewer line drives. From an even broader lens, here’s how Darvish’s 2013 compares with his 2012 campaign.
For the full 2013 season, ZiPS projects Darvish to end up 18-8 with a 3.27 ERA and 3.05 FIP, which are significantly better than his 2012 stats. If Darvish stays consistent, that tradeoff will certainly be worth the change in his pitching mix – and if he does so while bringing his home run rate down, his more conventional stats will reflect that he’s now one of the best pitchers in baseball.
[i] Disclaimer: the biggest actual change Pitchf/x recorded is Jake Peavy’s significantly fewer four-seam fastballs in favor of two-seam fastballs. But sliders are more interesting than switching between fastballs, no?