Ten 2018 Pitcher Strikeout Rate Decliners
On Tuesday, I hopped over to the pitcher side of the ledger to discuss nine fantasy relevant starting pitchers with strikeout rate upside this season. I used my xK% equation and compared what the formula spit out to what the pitcher’s actual strikeout was. Today, I’m going to share the ten pitchers who most outperformed their xK% marks.
| Name | Str% | L/Str | S/Str | F/Str | K% | xK% | K%-xK% |
|---|---|---|---|---|---|---|---|
| Jose Quintana | 61.9% | 29.3% | 15.1% | 29.6% | 26.2% | 22.7% | 3.5% |
| Ivan Nova | 66.3% | 24.0% | 13.6% | 28.0% | 16.7% | 13.4% | 3.3% |
| Chris Sale | 68.3% | 26.7% | 23.3% | 29.0% | 36.2% | 33.3% | 2.9% |
| Noah Syndergaard | 66.7% | 25.2% | 21.9% | 24.8% | 27.4% | 24.7% | 2.7% |
| Corey Kluber | 67.9% | 27.9% | 24.2% | 24.4% | 34.1% | 31.6% | 2.5% |
| Madison Bumgarner | 66.5% | 23.6% | 16.4% | 30.8% | 22.4% | 20.0% | 2.4% |
| Clayton Kershaw | 68.6% | 24.6% | 22.5% | 27.1% | 29.8% | 27.6% | 2.2% |
| Jake Arrieta | 62.5% | 29.6% | 15.0% | 27.5% | 23.1% | 21.1% | 2.0% |
| Carlos Rodon | 60.7% | 27.5% | 18.4% | 28.0% | 25.6% | 23.6% | 2.0% |
| Carlos Martinez | 65.2% | 28.4% | 17.6% | 26.5% | 25.3% | 23.3% | 2.0% |
| League Avg | 63.2% | 26.4% | 17.9% | 27.9% |
One of just two starters who outperformed his xK% by more than 3% is Jose Quintana. Though he also outperformed significantly in 2016, he never showed any such ability before that. Did he suddenly discover something not captured in the equation heading into 2016? I doubt it. That said, his xK% did hit a career high, which is interesting considering he also threw the lowest percentage of strikes in his career. The good news is that some of the strikeout rate regression is going to be offset by pitching a full season in the National League.
It’s sad when even a 16.7% strikeout rate could be deemed fortunate. That’s the situation Ivan Nova finds himself in, as he enjoyed no bump despite pitching his first full season in the NL. However, in my historical xK% spreadsheet going back to 2011, Nova has actually outperformed his xK% every single season. And he’s done so by an unweighted average of 1.9%, which is huge. I don’t know what he’s doing not being captured, but at least NL-Only leaguers could hope that outperformance continues, because any further drop will make him worthless in even deep formats.
I’m going to group Chris Sale, Noah Syndergaard, Corey Kluber, and Clayton Kershaw together. Aside from Syndegaard, all of them were among the league leaders in strikeout rate. Regression equations by design are notoriously poor at matching the extremes, and these pitchers represent it.
Syndergaard’s strikeout rate was high, but not extreme enough to totally ignore his much weaker xK%. Of course, this came over a small sample size because he missed most of the season with a lat tear. Since there’s seemingly no discount in early drafts, I don’t think there’s really a reason to pay market price. There’s more downside than upside.
Madison Bumgarner is yet another consistent xK% beater, doing so every season since 2011. However, this was the first season he beat his xK% by more than 1.8%, so this is clearly worrisome. Then again, it’s likely that the shoulder injury he suffered in a dirt bike accident affected his performance, so perhaps this is nothing to worry about.
In three of the last four seasons, Jake Arrieta has outperformed his xK% by at least 2%. What are these guys doing that xK% is missing?! I’m concerned that his swinging strike rate tumbled to its lowest mark since 2013. We’re still waiting for him to sign, so that will obviously affect his fantasy value this year.
Carlos Rodon is dealing with a shoulder problem and is going to get a late start to the season. That alone is enough to avoid him, but he also lands on this list! I wouldn’t touch him.
Carlos Martinez has alternated matching his xK% and outperforming it by at least 1%. This was his best outperformance yet. I continue to believe he’s dramatically overvalued and worry about that ERA-SIERA discrepancy. Since he’s clearly being paid based on his ERA and the assumption of drastic improvement this season, I would not want to be paying the going rate.
Mike Podhorzer is the founder of ProjectingX IQ, an advanced fantasy baseball analytics platform that transforms projection data and in-season performance signals into actionable intelligence. He is the 2015 Fantasy Sports Writers Association Baseball Writer of the Year and three-time Tout Wars champion. He is the author of the eBook Projecting X 2.0: How to Forecast Baseball Player Performance, which teaches you how to project players yourself. Follow Mike on X@MikePodhorzer and contact him via email.
Has this ever been back-tested, and if so, what were the results?
I may have just published one review, of the 2017 articles. The results were terrible, but the sample was tiny, obviously. Also, Steamer performed no better, so it was just a weird collection of pitchers.
https://www.fangraphs.com/fantasy/reviewing-the-2017-starting-pitcher-strikeout-rate-upsiders/
https://www.fangraphs.com/fantasy/reviewing-the-2017-starting-pitcher-strikeout-rate-downsiders/
I’ve never done a historical study, but the R-squared is 0.93, which is insanely high. There are some consistent outperformers and underperformers, but trying to account for them a year ago with a new equation failed.
A high R-squared does not mean you have a predictive model. As I commented in your first article, has your model been tested for multicollinearity, for example? Are your independent variables correlated? I would say so. Your model is like saying someone who is 1) very tall and 2) has big hands likely has big feet. 1 and 2 are highly correlated and will give a false sense of what the R-squared result is telling you.
My bad, the equation isn’t meant for predictive purposes, it’s backwards looking. I use historical xK% marks to guide my K% projection, since it’s much better than actual K% when dealing with small sample sizes.
The variables are the different strike types. A pitcher good at inducing swinging strikes has no bearing on whether he’s also good at inducing called strikes. They are separate skills.
Forgot that my original article linked to in the intro has all the predictive metrics, xK% in Y1 to X% in Y2 compared to K% in Y1 to K% in Y2. My metric was ever so slightly better overall (even though I didn’t develop it to be predictive), but the gap widened for pitchers with less than 50 IP, which is where the real value lies.
https://www.fangraphs.com/fantasy/introducing-the-new-pitcher-xk-updated-for-2017/
given Q and arrieta’s inclusion, is it possible the cubs found a way to outperform this or is it just a coincidence? and what are the elites doing that allows them to outperform?
Remember Quintana was also a White Sock half the season, so you can’t solely chalk it up to the Cubs. I’d have to look at Cubs pitchers historically to really answer that.
given Q and rodon’s inclusion, is it possible the white sox found a way to outperform this or is it just a coincidence?
I’m interested to know why you think C-Mart is over valued? His SIERA and FIP are about a half a run higher than his ERA, which doesn’t strike me as wildly high.
That increases the risk, because you’re relying on the luck metrics (BABIP, HR/FB, LOB%) to remain better than average and hoping these are skills C-Mart owns. I will always prefer to bet on the underlying skills like strikeout, walk and ground ball rates over those metrics. We need far more data to determine if a pitcher truly owns such skills, rather than having benefited from good fortune.
Sierra only .2 higher than era last year and xfip almost identical, doesn’t seem like that’s too far off, right? Verlander’s marks were way further off than CarMo last year
It’s SIERA, not sierra, and drafters clearly aren’t buying him based on last year’s ERA, but rather a significant improvement. No one is paying the price of a 16th selected starter expecting a 3.74 ERA and 1.30 WHIP like he posted in 2017.
“Regression equations by design are notoriously poor at matching the extremes, and these pitchers represent it”
This wouldn’t be a problem if you modeled these variables without assumptions of linearity. Curious as to why you choose to do so.
I don’t know any other way to model things! I’m a novice at this. I use Excel’s regression tool and take whatever it spits out. Many in the comments suggested other model types, like quadratic, random forest, etc, and tested it out a year ago, but nothing was any better. I don’t know how to run those myself anyway.
So, the list of overperformers (decliners) is 4 of the top 5 pitchers in the majors (only Scherzer is missing) and 8 of the top, what, 20? Maybe 25? Meanwhile the list of underperformers (surgers) includes nobody who would be considered in the top 40 or 50. It isn’t even like it’s “Hey, the one list is a little better than the other”. It’s not even close and it really doesn’t matter whether you look backward or forward at them. The overperformers list is a LLLLLLOOOOOOOOTTTTT better than the underperformers list, much more than just simply stealing a few strikes.
TO my eye, the overperformers list does skew a little older, but when you see such stark discrepancies, I think you need to look at the formula again.
Mike, is the K analysis able to reflect in any way what pitchers do in specific counts? For example, a guy might exceed his expected totals if he has a pitch-to-contact mentality early in the count, but then always goes for swinging strikes and the K with 2 strikes. Just spit balling, but approach might have something to do with consistently being higher or lower than expected.
Yup, sequencing. That’s the one thing I know is missing, but impossible to incorporate.
Why do you think Carlos Martinez is overrated? He’s an extreme GBer with great K rate, 26yo, and he’s slowly increased speed up to near 96. It was over 96 3+ years ago, but presumably he reduced it for control reasons. And his pen support was lousy last year (though perhaps that has not changed). Just curious.
Read my comments above. I don’t dislike him, just think people are paying for his 2015 and 2016 ERA, even though his SIERA suggests that was the result of good fortune, not underlying skills. He may very well have the ability to outperform his SIERA, but we don’t have nearly enough of a sample yet to prove this is the case.
Hmmmm, that’s quite funny, maybe I don’t think he’s overvalued! I had him as my 16th ranked starting pitcher in LABR Mixed, and sure enough, his ADP is also 16th among starters.