Jeff Samardzija is the Best and Worst He’s Ever Been

If you’ve paid any attention to the San Francisco Giants, you’ll know that they stink something awful right now. The parts generally are no greater than the whole. Jeff Samardzija, he of the 5.44 ERA, is not blameless here.

In an alternate universe, though, he could be. Some in(s)ane factoids about Samardzija: Only Chris Sale has as many starts as Samardzija in which he struck out more hitters than he completed innings (6). (In Sale’s first start of the season, he went seven and struck out seven. So close.) Samardzija is also one of only six starters with four-plus starts of eight-plus strikeouts. And among pitchers who have thrown at least 75 innings since August 8, 2016*, Samardzija’s 3.12 xFIP ranks 7th-best, behind only Carlos Carrasco, Noah Syndergaard, Clayton Kershaw, Sale, Michael Pineda and James Paxton. That is elite company.

*Why August 8? I was trying to see who has been better than Ivan Nova since he was traded to the Pirates. Nova shows up 8th on that list above. Seeing Samardzija’s name directly before his floored me.

Samardzija is striking out the world yet has little to show for it. His advanced stats (28.7% K, 5.2% BB) suggest excellence, and his peripherals (11.8% SwStr) affirm them. In short, his 3.43 FIP and 2.87(!!!) xFIP depict a much more effective starting pitcher. It’s his strand rate (LOB%) — a catastrophically bad 58.1% — that has done him in. Normalize it, and he’s sitting pretty with a mid-3.00s ERA.

All that said, I’m here to investigate what changed. Once upon a time, Samardzija was a touted prospect, cracking multiple top-100 lists in 2009. The strikeouts lived up to the hype, yet the results lagged. Then the K’s eroded, and the results eroded further. They K’s are back, and they’re back with a vengeance.

The splitter has returned. Almost half a decade ago, Ben Duronio asked, How good is Jeff Samardzija’s splitter? I was tempted to recycle the title to bring everything full circle. Samardzija’s splitter was good and now is, again, good, and he’s using it twice as often as he did in 2015 and the first half of 2016 when it suffered from diminished effectiveness. There seems to have been a well-intentioned effort to improve it, too, as indicated in this February piece from CBS Sports. It’s a mere two paragraphs — nay, a paragraph and a half — but it suggests that whatever Samardzija did to restore his splitter’s effectiveness in spring training may have worked.

What exactly it was that he did, though, it’s hard to say. Perhaps his work in spring training merely served as an extension of his late-season gains last year: starting August 8, when his overall performance suddenly turned for the better, his splitter velocity improved by 0.8 mph, coinciding with about an inch less of horizontal movement on the pitch. The changes seemingly have transformed the pitch; what once was thrown for strikes only 55.7% of the time with a 12.4% whiff rate is now thrown for strikes 71.2% of the time with a 22.6% whiff rate.

The hitch: hitters are swinging at the pitch a ton. While this development has induced tons of fruitless hacks — only Rick Porcello‘s fastball and Masahiro Tanaka’s splitter have been swung at more often, min. 100 pitches thrown — it has produced tons of fruitful hacks as well. The balance leans in favor of the former — he Shark’s splitter exists in rarified air, as it is one of few pitches miss bats more often than they allow balls to enter play — yet it still allows an adequate amount of damage simply on the basis of frequency. Much this contact appears troubling, too: his splitter has allowed a 30.8% line drive rate (LD%). However, his overall hard and soft contact allowed (Hard% and Soft%, respectively) are among the best we’ve ever seen from him, so those line drives may not be especially efficacious. The .194 isolated power (ISO) that his splitter has allowed is not great, but with such a high line drive rate, you’d expect worse.

Eventually, something will have to give. Maybe hitters learn to lay off, resulting in fewer whiffs, walks, and balls in play. All outcomes become a little more muted. Alternatively, if hitters continue to offer at the splitter, the line drives should slip from their alarming frequency. With time, as the line drives recede, his splitter could eventually resemble its late-2016 prowess when it allowed a meager .128 ISO and .195 batting average against.

Still, there exist other holes in Samardzija’s arsenal, holes that an improved sinker alone cannot remedy. Hitters have always teed off on his sinker, the pitch he features most prominently. It has allowed a .339 batting average on balls in play (BABIP) since the beginning of 2012, an issue that has been exacerbated in 2017. (His splitter has only allowed a .249 BABIP in that same span, by the way, with a better ground ball rate in 2017.) There’s clearly more optimization that can occur here.

Regardless of his gains, though — regardless of all of this — no one suffers a sub-60% strand rate forever. Fewer than 5% of qualified pitchers have suffered even a sub-66% strand rate in the last 15 years. Alas, I guess it’s possible that Samardzija suffers bad luck for another 160 innings, but it’s much, much likelier that his ERA creeps back toward something more palatable, to say the least. Should he finish the season with a 4.00 ERA, that’s a 3.56 ERA from here on out. I expect better; a 3.70 ERA on the season would mean closer to a 3.20 ERA onward. That’s pretty dang good.

It’s easy to write off Samardzija for his perennially disappointing results relative to his “stuff.” But luck has not been on his side this year while the skills have. Even if xFIP isn’t the most predictive of the fielding independent metrics these days, there’s a reason why analysts love to bet on their xFIP darlings. The Shark was always a mixed league starter, but if he continues to strike out 10 (and walk only 2) per nine innings and benefit from favorable strand rate regression, he could quietly become a legitimate mid-rotation asset — or better.