Strikeout Rates: Bumgarner & Gio Gonzalez
In early February, I wrote a post for “FG+” that endeavored to sniff out starting pitchers who either over-performed or underperformed their inherent strikeout ability. Inspired by Zack Greinke, Tommy Hanson, and Anibal Sanchez, among others – there were many starters that saw pretty massive fluctuations in their strikeout performance between 2010 and 2011. Being the curious human I am, I wanted answers. Or at least a direction on the path of finding an answer.
Specifically, this little study sought out to create an “expected” strikeout rate by controlling for their prior year K%, current year (2011) K%, SwStr%, and fastball velocity (definitions of correlations and the r-squared of the model at the bottom of this post). The results were pretty interesting, should you happen to have access to the full piece.
Most of our fantasy starters have four games under their respective belts at this point. That is, unless you’re Brandon McCarthy or the strike-throwing-machine Bartolo Colon – who, by the way, was nicknamed “Boogie Bear” while in Anaheim, a fact I would like to keep alive in our collective consciousness. But I digress.
The chart below is rather unsightly, so I apologize, but it’s going to have to serve our purposes nonetheless. The first column is the model predicted K%, followed by 2010 K%, 2011 K%, 2011 FBv, 2011 SwStr%, the difference from predicted K% and then the following columns are 2012 stats with differences from 2011.
| Name | xK% | 10 K% | 11 K% | FBv | SwStr% | + / - | 12 FBv | 12 K% | SwStr% | SwStr% +/- | FBv +/- | K% +/- |
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Zack Greinke | 0.214 | 0.197 | 0.281 | 92.5 | 0.106 | 6.70% | 92.7 | 0.289 | 0.088 | -0.018 | 0.2 | 0.008 |
| Cliff Lee | 0.2 | 0.22 | 0.259 | 91.5 | 0.093 | 5.90% | 90.7 | 0.225 | 0.067 | -0.026 | -0.8 | -0.034 |
| Tommy Hanson | 0.204 | 0.205 | 0.263 | 91.2 | 0.1 | 5.90% | 89.5 | 0.23 | 0.092 | -0.008 | -1.7 | -0.033 |
| C.J. Wilson | 0.178 | 0.2 | 0.225 | 91 | 0.083 | 4.70% | 91.4 | 0.19 | 0.069 | -0.014 | 0.4 | -0.035 |
| David Price | 0.196 | 0.218 | 0.238 | 94.8 | 0.084 | 4.20% | 95.2 | 0.159 | 0.064 | -0.02 | 0.4 | -0.079 |
| Madison Bumgarner | 0.187 | 0.182 | 0.226 | 91.7 | 0.092 | 3.90% | 90.3 | 0.11 | 0.06 | -0.032 | -1.4 | -0.116 |
| Paul Maholm | 0.104 | 0.121 | 0.141 | 87.4 | 0.057 | 3.70% | 87.7 | 0.13 | 0.062 | 0.005 | 0.3 | -0.011 |
| Jake Arrieta | 0.142 | 0.116 | 0.178 | 92.4 | 0.074 | 3.60% | 93.3 | 0.208 | 0.069 | -0.005 | 0.9 | 0.03 |
| Ian Kennedy | 0.185 | 0.207 | 0.22 | 90.3 | 0.088 | 3.50% | 88.3 | 0.2 | 0.094 | 0.006 | -2 | -0.02 |
| Ubaldo Jimenez | 0.186 | 0.239 | 0.219 | 93.5 | 0.075 | 3.30% | 92.1 | 0.139 | 0.054 | -0.021 | -1.4 | -0.08 |
| Clayton Kershaw | 0.241 | 0.25 | 0.272 | 93.4 | 0.111 | 3.10% | 92.6 | 0.253 | 0.109 | -0.002 | -0.8 | -0.019 |
| Yovani Gallardo | 0.209 | 0.249 | 0.239 | 92.7 | 0.09 | 3.00% | 91.4 | 0.226 | 0.085 | -0.005 | -1.3 | -0.013 |
| Anibal Sanchez | 0.213 | 0.187 | 0.243 | 91.7 | 0.109 | 3.00% | 90.1 | 0.238 | 0.092 | -0.017 | -1.6 | -0.005 |
| Justin Verlander | 0.229 | 0.237 | 0.258 | 95 | 0.102 | 2.90% | 94 | 0.261 | 0.108 | 0.006 | -1 | 0.003 |
| Felix Hernandez | 0.202 | 0.232 | 0.23 | 93.3 | 0.088 | 2.80% | 91.8 | 0.261 | 0.087 | -0.001 | -1.5 | 0.031 |
| Gio Gonzalez | 0.2 | 0.201 | 0.228 | 92.5 | 0.095 | 2.80% | 93.3 | 0.297 | 0.102 | 0.007 | 0.8 | 0.069 |
Overall, the predicted drop in K% for this particular group was about 3.9% and the observed drop from this group is just shy of 2% – so the model did alright thus far. If you throw out Gio Gonzalez, it’s 2.5%, but then again, that’d be cherry picking, right?
So let’s take a look at a couple of outliers. Both Gonzalez and Madison Bumgarner jump out at me in this list as being poster boys for head scratching nerdom, although on opposite ends of the K% spectrum.
For Gonzalez, he had a K% of 20.1% in 2010 and 22.8% in 2011. The available research on switching leagues suggests that moving to the National League ought to increase your K/9 by about .5, which is significant, but it wouldn’t account for what Gonzalez has been able to demonstrate in 2012. So far, he’s striking out roughly 30% of batters faced, but his overall swinging strike rate is up just a tick from 2011. His fastball definitely has more zip on it, however – at 93.3 mph, which represents the highest average fastball in his career.
His strikeouts are coming almost exclusively on his curveball and change, which he throws about 40% of the time (25% curveball), each generating just shy of a 20% whiff rate. But with “just” a 10.3% swinging strike rate and a whiff rate that’s good, but not spectacular on his strikeout stuff, you can’t expect that K% to sit up in the 30 percentile range for too much longer (in fact, nobody in baseball had a K% over 30% in 2011). That’s not to say you should run out and try to sell high on Gonzalez, because his results have been pretty filthy, and there’s not much to point to good fortune as the cause. The big knock on Gonzalez was his unseemly walk rate and he’s kept that well under control so far — and if that’s a trend he can put to use going forward, it could be a great year to own him.
Bumgarner, however, has far underperformed what was expected of him relative to strikeouts. The model predicted, based on his 91.7 mph fastball, 9.2% swinging strike rate, and 2010 K% that his actual K% should settle in around 18.7%, closer to what we saw in 2010. But despite quite good results in wins, ERA, and WHIP, Bumgarner’s strikeouts have gone on hiatus with a paltry 11% K rate so far in 2012.
What’s perhaps more concerning is his velocity is also off about 1.5 mph and he’s inducing just a 6% swinging strike rate overall. Over his first four starts, Bumgarner has managed to leave almost 85% of runners on base and opponents have just a .231 BABIP despite nearly a 52% ground ball rate.
Last year, Bumgarner used his slider about 28% of the time, generating a 12% whiff rate, and it was his most valuable pitch at roughly two runs above average per 100 pitches. This season, he’s going to the slider even more — over 36% of the time — but he’s only garnering whiffs on 9% of them, which might be why his O-swing% has dropped about 5% and his Z-contact% rate is up to 93% in 2012. Looking graphically at the slider, it looks like it has lost a couple inches of vertical drop and maybe an inch of horizontal movement when compared to last season:
From a fantasy perspective, I’d exercise caution on both Gonzalez and Bumgarner. While Gonzalez is probably pitching above his head right now, his move to the National League and newly discovered control could make him pretty valuable. It’s not a terrible play to try to see if anyone wanted to buy on him, but chances are you would have a difficult time convincing anyone that his 1.52 ERA and 0.85 WHIP are at all sustainable.
For Bumgarner, I think there’s no reason to panic. I’d give it several more starts to see if he can re-master the movement he once had on his slider and if the velocity starts to creep back up. But if you own Bumgarner, if his velocity doesn’t come back and the slider remains relatively unchanged, it might be a good idea to use his name recognition in a few trade ideas if you have the depth to absorb it on your staff, and other needs to fill.
The 2011 K% correlated with the 2010 K% at 0.754; with Fastball velocity at 0.543; with SwStr% at .825 – all statistically significant at .001.
The regression model using expected K% as the dependent variable, controlling for the three variables above had an adjusted R-squared of .761, statistically significant at .001. Independently, it accounted for more variance in K% than any single variable on it’s own.
Sorry if I’m missing something but I don’t see Gio on your chart. Don’t want to be a pain about it, but since you wrote about him so much in the piece it would be good to see his numbers on there.
gah – copy and paste flub at 2 am. Working on fix. Thanks.
Given the well publicized velocity decline of Lincecum to start the season, perhaps we should start looking at factors specific to SF. E.g., is the weather colder? is there some aspect to their training regimen (such as the emphasis on long toss that was mentioned in a previous post)? etc. etc.
Obviously, there’s not yet reason to think there’s correlation here, but it is an interesting data point.
I believe Vogelsong is down in velocity as well.
Through 4 starts last year Bumgarner had only 9 k’s in his first 17 ip. He had a brutal first couple weeks in AAA the year before. The guy is a slow starter.
agreed. I saw that as well.
I’m really liking this analysis. I’ve looked at Morrow’s Pitch F/X data and I think he’s in the same boat as Bumgarner. His four-seamer’s velocity is down 3mph. It also lack a lot of the movement that made it his miss-bats pitch. Additionally, his change-up’s velocity increased 2mph. Also, it moves less than last year.
Thoughts on Morrow’s drop in K? Do you think he’ll change his ways, or are we looking at a different pitcher this year?
I’ll definitely check it out. In general, I think we should expect that most pitchers graduate increase their velocity as the season moves along. Looking at Morrow’s data in our Pitch F/X, I see just a .6 drop in overall fastball velo. Would need to do more digging on movement though as his K’s are simply not there. Interesting.
Word on the street out of Toronto is that Farrell is stressing that his SPs try to go deeper into games (~7 IP) by “pitching to contact” more. In other words, reduced pitch counts and Ks in exchange for better real-world results. Fantasy stats be damned.
His contact rate in the strike zone is off the charts, too, though. And it’s not like he’s throwing tons more strikes.
I think we need another start or two before we know what’s up with Morrow.
I was trying to double check the R-Sqr. I ran a regression between the expected decline (xK% -2011K%) vs experienced decline (2012K%-2011K%) and I got a much smaller Rsqr (about 3%). Is that how you ran it? If so, I may have made a mistake (I also took out Gio since I think he is a more explainable outlier).
if you go to the FG+ piece, all of that is explained. I didn’t want to bog this post down with all the math. If you’re not a FG+ subscriber, I can send you an e-mail with the particulars.
How do you find the FG+ piece, I’m a FG member, but I dont see the piece your talking about and would like to see the math
there’s a link to the article in the first line of the post.
I’m betting the reason for Bumgarner’s slider losing movement is that he’s throwing more cutters. Watched two of his starts, and he has a pitch that was 88-90 with movement into RHB, and a 82-85 MPH pitch that he drove at the back foot of RHB. That 88-90 pitch is a cutter, and the 82-85 pitch is his slider, but they’re somehow classified as the same pitch. Is it his arm angle that throws off Pitchf/x?
pretty gross user name.
Looks like everyone is classifying what you’re describing as a two-seamer instead of a cutter, which are pretty similar pitches.
I wouldn’t say so, he has a fastball that tails away from RHB and into LHB as well, which is his two-seamer. Cutter is faster but has less movement than a slider, especially vertically, which would seem to match up perfectly with the loss of movement with Bumgarner’s slider this season. Anecdotally, he was throwing the cutter more frequently than the slider in the starts I watched.
One thing that might help understand Gio’s outlier status is his ridiculous K rate vs. opposing pitchers. Of 7 official AB’s, 6 ended in k’s (I think there was one more PA that resulted in a sacrafice bunt).
Just to follow up, his K rate falls to 25.3% after taking out his PA vs. opposing pitchers.
just a quick question…where is the AL OF tier section?
don’t think they split it out this year. Updates coming in the next week or so, I’m sure… http://www.fangraphs.com/fantasy/index.php/rotographs-2012-fantasy-rankings-outfielders/
I need help, does gio throw a slider? On The Show he has one but they’re not great on having correct repertoires