When Plate Discipline Sticks
A few days ago, Jake Leech asked me if Zack Cozart’s 2017 improved plate discipline would stick into 2018.
https://twitter.com/Stroke_19/status/931525718667943936
Cozart saw quite a bit of improvement with his K%-BB% dropping by 6% points.
Note: I like using K%-B% to get an overall value for a hitters plate discipline. Earlier this year, I investigated what early season stats point to a true breakout. K%-BB%, along with launch angle (FB%), were the two key factors to focus on.
| Season | BB% | K% | K%-BB% |
|---|---|---|---|
| 2016 | 7.3% | 16.5% | 9.2% |
| 2017 | 12.2% | 15.4% | 3.2% |
| 2018 (Steamer) | 8.8% | 15.6% | 6.8% |
The Steamer projection has his K%-BB% regressing closer to his 2016 values than the ones from 2017. This is how projections work with previous season stats having some weight along with some regression.
From the Tweet thread, I wondered if some underlying plate discipline values (Pitch Info swinging and contact values using inside and outside rates) could point to a hitter keeping more of the gains.
I took all the hitters (min 200 PA) since 2008 who saw their K%-BB% increase by 7.5% in one season and then their eventual value in year three. I started with hitters who saw a 6% change like Cozart. I couldn’t find a nice year-to-year trend. I did find one at 7.5%. The 6% values generally lined up but the 7.5% values were more obvious.
Once I found the hitters who improved their plate discipline, I grouped the hitters by who kept the improvement going forward into three groups. Here are the results.
Note: To make sure the results weren’t from the largest changes regressing the most, I found the year one to year two changes. The average changes were within 0.3% (8.6%, 8.9%, 8.7%) of each other.
| Average | O-Swing% | Z-Swing% | Swing% | O-Contact% | Z-Contact% | Contact% |
|---|---|---|---|---|---|---|
| Kept Changes | -2.4% | 0.1% | -2.4% | 8.0% | 3.9% | 4.9% |
| Regressed Some | -1.5% | -0.5% | -1.5% | 3.8% | 2.4% | 3.0% |
| Heavy Pullback | -0.9% | -2.0% | -1.9% | 5.8% | 1.8% | 3.4% |
| Median | O-Swing% | Z-Swing% | Swing% | O-Contact% | Z-Contact% | Contact% |
| Kept Changes | -2.3% | 0.1% | -1.6% | 6.8% | 3.9% | 5.1% |
| Regressed Some | -2.2% | -0.6% | -1.8% | 3.8% | 2.4% | 3.1% |
| Heavy Pullback | -1.3% | -2.5% | -2.2% | 5.3% | 2.0% | 3.0% |
The two categories which showed consistent progression were O-swing% (outside swing rate) and Z-contact% (strike zone contact rate).
With these two findings, I went down several paths such as combining their values and performing regression. In the end, I kept the values separate and decreased each a bit to a 2% point drop for O-swing% and a 3.5% increase for Z-contact%.
I found all the hitters with at least a 7.5% point drop in K%-BB% with the 2%/3.5% change (Good Indicators group). Then I compared them to the rest with the 7.5% drop (Bad Indicators).
| Year1 | Year2 | Y2 – Y1 | Year3 | Y3 – Y2 | |
|---|---|---|---|---|---|
| Good Indicators | 16.3% | 7.2% | -9.0% | 9.4% | 2.2% |
| Bad Indicators | 16.1% | 7.0% | -9.0% | 11.1% | 4.1% |
Both groups saw some regression but the good indicator group didn’t regress as much. Examining the results from 2016 to 2017, five players fall into the Good Indicators group who should keep most of their gains for 2018.
| 2016 | 2017 | Difference | |||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|
| NAME | PA | K%-BB% | O-Swing% | Z-Contact% | PA | K%-BB% | O-Swing% | Z-Contact% | K%-BB% | O-Swing% | Z-Contact% |
| Steven Souza Jr. | 468 | 27.4% | 32.1% | 76.0% | 617 | 15.4% | 24.9% | 79.4% | -12.0% | -7.3% | 3.4% |
| Eddie Rosario | 354 | 22.3% | 44.3% | 79.4% | 589 | 12.1% | 40.2% | 81.9% | -10.3% | -4.2% | 2.5% |
| Joey Votto | 677 | 1.8% | 23.2% | 84.7% | 707 | -7.2% | 18.4% | 88.5% | -9.0% | -4.8% | 3.8% |
| Marwin Gonzalez | 518 | 18.5% | 36.9% | 89.1% | 515 | 9.7% | 28.8% | 91.1% | -8.8% | -8.2% | 2.0% |
| Anthony Rendon | 647 | 8.0% | 27.6% | 87.3% | 605 | -0.3% | 20.1% | 91.8% | -8.4% | -7.4% | 4.5% |
Two stars and three breakouts from last season are on the list. Here some quick thoughts on each:
- Joey Votto: He is the league’s closest thing to a hitting robot. He took the league’s best plate discipline to a whole new level. Good luck pitchers.
- Anthony Rendon: He transitioned from a .270/.350 (AVG/OBP) hitter to borderline elite one at .300/.400. Even if he regresses some, the high OBP and power will keep him accumulating counting stats.
- Steven Souza Jr.: The changes transformed him from borderline unplayable to a decent sleeper. The one issue to remember is that he’s more valuable in an OBP league with his sub-.250 AVG.
- Marwin Gonzalez: I’m afraid his value going to jump as he is qualified at 1B, SS, and OF. Also, it helps that he’s productive and everyone saw him in the postseason.
- Eddie Rosario: I see why owners might not buy in but he had his 2017 breakout season with his BABIP dropping from .338 to .312. He’s being more selective at the plate and driving the new juiced ball all over the field. Believe.
Getting back to Cozart, he just didn’t meet the criteria to keep his changes into 2018. He’ll keep some but as his Steamer projection shows, he’s likely to see his plate discipline pull back. Instead, players making and keeping plate discipline changes need to start with a notable plate discipline change along with the ability to lay off pitches out of the strike zone and hit the pitches in it. Basically keeping the improvement comes down to: don’t swing at balls and hit strikes.
Jeff, one of the authors of the fantasy baseball guide,The Process, writes for RotoGraphs, The Hardball Times, Rotowire, Baseball America, and BaseballHQ. He has been nominated for two SABR Analytics Research Award for Contemporary Analysis and won it in 2013 in tandem with Bill Petti. He has won four FSWA Awards including on for his Mining the News series. He's won Tout Wars three times, LABR twice, and got his first NFBC Main Event win in 2021. Follow him on Twitter @jeffwzimmerman.
I am so confused on the blurb about Marwin. Perhaps it needs some editing?
Makes sense to me. Increased value for multiple positions plus he got a lot of added attention because of the postseason.
The swing at strikes and take balls is definitely good advice but I think the not Swinging at balls is more important.
I did some Research looking at Zone swing minus O swing https://www.fangraphs.com/community/is-z-o-swing-a-better-indicator-of-plate-discipline-than-o-swing/ and while it was slightly better than just O swing the effect wasn’t strong. however I found that that might be a Little artificial since on base almost exclusively correlated with O swing but there was some positive effect of more Zone swing percentage on power while chasing less had no positive effect for power.
so it could be that it is a coincidence that O swing and z-o come out so Close because they affect different things but in the end it is about the same. that might mean that certain hitters relatively benefit more from lowering O swing and others from increasing z swing but I did not look into that.
Not swinging at balls is more important. It only takes a 2% change
Thanks for the analysis, Jeff. I’ve been enjoying this series of articles.
Thank you so much! I was trying to do this EXACT analysis on Cozart myself!
One question – the 200 PA threshold I understand – need to get names into the sample – but it seems light. Is that where the metric begins to gain some statistical significance? Because if so, we may want to look at Cozart half-by-half – he exceeded that in both halves and seemed to improve:
1st half, 284 PA : 11.6% BB, 16.9% K
2nd half, 223 PA: 13.0% BB, 13.5% K
Those second half numbers are inflated by insane July (14.1% BB, 7.8% K%!!) and less-insane but still great August (14.0% BB, 11.6% K) numbers and a 98 PA September netted a 9.2% BB and 16.9% K.
To add to the intrigue, his batted ball profile changed very significantly but he remained equally as productive:
1st half: 21.8% LD, 40.6% GB, 37.6% FB, 11.8% HR/FB, RC 143
2nd half: 16.7% LD, 35.2% GB, 48.1% FB, 19.2% HR/FB, RC 138
Fly ball revolution? LD% isn’t sticky, and HR/FB isn’t sticky yet, but at those FB and GB looks to have legitimately shifted some. And once he lofted 46% FB in June, he didn’t dip below 45.2% again for the season. He was pulling the ball more (51.2% 2nd half vs. 47.0% 1st half) but that doesn’t seem sticky. But NOT going oppo does seem to have some weight, but need more info (19.8% 1st half vs. 12.4% 2nd half). I think he changed his profile to take advantage of the juiced ball and the 20+ HR power is real.
What does it all mean? I don’t know. I got frustrated and gave up! I plugged in the first half numbers and called it a day thinking that offered regression from the second half but growth from the past. And figured if he somehow ends up with the Reds again with his donkey pal Votto in his ear, I’d bump him up another round or two.
After your analysis I’m thinking I might be a tad optimistic on the BB% gains and actually a little pessimistic on the K% gains. But I wonder if his second half was used, or if the 200 PA threshold for comps was raised to 400 PA or so, how the conclusion of “stickiness” would differ.
Excellent article though – I cannot describe how much it helped me connect a couple dots I was too stupid to see! Thanks!
A lot there. One note, follow GB% changes. FB% and LD% meld some. GB% is an outstanding proxy for Launch Angle. Also, the 1H to 2H changes do give me more hope for him.
Great tip about FB% and LD% melding, makes sense. Appreciate the response, Happy Thanksgiving!
Cozart can elevate the ball and his plate discipline isn’t terrible but he hits the ball very soft. Not billy hamilton soft but well below average.
With the right angle and if you pull the ball enough which he does you don’t need insane exit velo but he is very much on the low end of velo for a guy hitting 15+ hr.
I think that band box helps him a lot. In cincy I think he can hit 20 again but if he moves to a larger park HRs could plummet.
For what it’s worth, Cozart has been slowly and gradually increasing his walk rate for the past 5 years: 4.2%, 4.6%, 6.5%, 7.3%, 12.2%. The strikeout rate has been relatively consistent, between 13.6% and 16.5%. I think it’s reasonable to expect about Steamer’s 8.8% is about right, with a small chance of it being a bit higher and a very small chance of it being much lower (and likely not lower than 7.3% at worst).
The real question is the BABIP. Last year (.312) was the first it has been above .274 in 4 years and the first time it has ever been above .285.
Does that continue? His IFFB rate hasn’t changed, His LD rate hasn’t changed, and his FB rate hasn’t really changed. I expect whatever he gains with the increased walk rate is mostly given back with a pretty big drop in BABIP back to the .270s. Assuming he keeps some/most of the power, that puts him around .270/.355/.490.