More on Plate Discipline
Last year I did a two part series on plate discipline that delved into a few statistics that I thought better represented a batter’s actual plate discipline than your traditional metrics. The stats are in for the 2006 season, so I figured it’d be worth taking another look. Here’s a quick recap of last year’s findings:
Z% (Zone Percentage) – The percentage of pitches a batter sees inside the strike zone. Correlates with walk rate (BB%) and home runs per fly ball ( HR/FB). Batters with more power are pitched more cautiously resulting in a lower Z% and a higher BB%.
OSwing (Outside Swing Percentage) – The percentage of pitches a batter swings at that are outside the strike zone. Correlates with walk rate (BB%). This year, OSwing will be represented as OSwing above the MLB average.
Contact (Contact Percentage) – The percentage of times a batter makes contact with the ball when he swings the bat. Correlates with strikeout rate (K%) and home runs per fly ball (HR/FB). Batters who can’t make contact with the ball obviously strike out more often, and batters who swing “harder” often make less contact, resulting in higher HR/FB and more strikeouts.
So, with the recap out of the way, let’s look at some year to year correlations for all three for the first time.
They all correlate well from year to year, but Both OSwing and Contact correlate extremely well. I consider OSwing about the best measure of plate discipline since not swinging at pitches outside the strike zone is pretty much the definition of plate discipline.
Seeing that it correlates so well from year to year (at least in 2005 & 2006) suggests that players do not quickly develop plate discipline. Perhaps it’s a skill that can be learned over time, but there are few players who saw drastic changes in OSwing from 2005 to 2006. Less than 10% of all players with 300 at-bats in 2005 and 2006 saw more than a 5% change in OSwing from 2005 to 2006.
Name Dif Name Dif Andruw Jones 8.09% Jeromy Burnitz -5.10% Geoff Jenkins 7.13% Dave Roberts -5.34% So Taguchi 6.97% Mark Loretta -5.38% Willy Taveras 6.62% Freddy Sanchez -5.66% Aaron Miles 6.16% Vladimir Guerrero -5.71% Scott Hatteberg 6.08% Jay Payton -5.89% Joe Crede 5.92% Kevin Mench -6.82% Jorge Cantu 5.13% A.J. Pierzynski -7.50% Eric Chavez 5.08% Clint Barmes -8.39%
Contact showed an even higher correlation from year to year than OSwing, also suggesting that players don’t really change their approach from year to year. In fact, there were only 10 players who had more than a 5% change in Contact from 2005 to 2006.
Name Dif Name Dif Corey Patterson 5.10% Brad Wilkerson -8.87% Mike Piazza 5.17% Bill Hall -7.41% Adam Everett 5.51% Nick Swisher -7.02% Reed Johnson 5.71% Chris Shelton -6.37% Troy Glaus 6.71% Craig Monroe -5.21%
Finally, Z% showed the least amount of correlation from year to year, but it wasn’t a poor correlation by any means. The decreased correlation I suspect is due to this metric not being entirely within the batters control. While how a batter is pitched is indicative of his various skills (mainly power and overall plate discipline), it’s still up to the pitcher to decide how to proceed.
Between both Contact and OSwing there appears to be a sort of “sweet spot” for batters. Let’s apply some filters to OSwing and see what happens. Particularly, let’s look at power batters who have a HR/FB greater than 15%.
For the first list, let’s limit the batters to those who have “considerably better” plate discipline than the rest of the league. Let’s call “considerably better” an OSwing of 5% or more than league average.
Name Contact HR HR/FB Jason Giambi 80.97% 37 20.00% Morgan Ensberg 74.65% 23 16.43% Barry Bonds 85.78% 26 16.56% Nick Johnson 84.53% 23 15.97% Pat Burrell 79.50% 29 18.13% Jim Thome 71.92% 42 27.81% Chipper Jones 82.27% 26 19.12% Frank Thomas 86.58% 39 17.41% Carlos Beltran 84.08% 41 21.13% Adam Dunn 70.42% 40 22.22% Troy Glaus 75.66% 38 18.72% Jason Bay 75.26% 35 18.82% Nick Swisher 71.07% 35 17.86% Austin Kearns 74.11% 24 15.29%
If we move to the next list, which I’ll use the same criteria for, but instead of batters who are “considerably better”, this will just be batters who have “above average” plate discipline (OSwing between 0% and 5% above league average).
Name Contact HR HR/FB Josh Willingham 79.31% 26 15.85% David Ortiz 77.92% 54 26.09% Albert Pujols 86.24% 49 22.48% Casey Blake 82.82% 19 16.67% Raul Ibanez 80.26% 33 16.50% Jim Edmonds 73.31% 19 16.81% Travis Hafner 72.73% 42 30.22% Lance Berkman 79.24% 45 24.59% Phil Nevin 72.79% 22 21.57% Jermaine Dye 78.26% 44 25.43% Bill Hall 71.97% 35 19.44% Brad Hawpe 73.98% 22 16.18% Paul Konerko 82.11% 35 17.50% Moises Alou 85.24% 22 17.46% Andruw Jones 72.94% 41 22.04% Richie Sexson 69.40% 34 19.32% Mark Teixeira 79.93% 33 15.94% Alex Rodriguez 74.27% 35 20.23% Mike Piazza 81.88% 22 17.05% Ken Griffey Jr. 80.70% 27 18.00% Manny Ramirez 78.46% 35 23.49% Miguel Cabrera 80.65% 26 15.57% Adrian Gonzalez 79.93% 24 15.69%
Next up are the batters who have “below average” plate discipline (OSwing between 0% and 5% below league average).
Name Contact HR HR/FB Ray Durham 88.15% 26 15.95% Adam LaRoche 76.81% 32 21.19% Marcus Thames 73.19% 26 17.11% Carlos Delgado 74.39% 38 22.89% Ty Wigginton 76.38% 24 16.90% Carlos Lee 86.49% 37 16.09% Ryan Howard 67.49% 58 39.46% Mike Cuddye 76.26% 24 15.69% Aramis Ramirez 84.28% 38 15.14% Juan Rivera 84.38% 23 17.69% Mark Teahen 79.05% 18 16.51% Craig Wilson 70.20% 17 15.74% Craig Monroe 74.79% 28 15.14% Matt Holliday 78.84% 34 20.00% Prince Fielder 76.55% 28 15.82% Vernon Wells 83.45% 32 15.02% Torii Hunter 78.02% 31 18.34% Wilson Betemit 76.67% 18 18.00% Preston Wilson 76.26% 17 16.67% Miguel Tejada 84.33% 24 15.48%
And finally, the batters who have “considerably worse” plate discipline (OSwing of 5% or more below average).
Name Contact HR HR/FB Jeromy Burnitz 72.47% 16 16.00% Ben Broussard 77.00% 21 15.56% Justin Morneau 81.07% 34 16.43% Rocco Baldelli 77.09% 16 16.00% Jacque Jones 73.82% 27 25.47% Alfonso Soriano 73.92% 46 18.25% Jeff Francoeur 76.47% 29 15.26% Vladimir Guerrero 83.15% 33 16.34%
Now that you’ve seen the lists, it seems clear to me at least, that it’s preferable to have above average plate discipline. Some of the guys in the below average list are borderline, but for the most part, it’s just not as prestigious a list.
The “considerably worse” list is fascinating, since some of these players actually get away with such an aggressive approach. Vladimir Guerrero, who swings at pretty much everything, is talented enough to get away with it. Justin Morneau got away with it last year, but it’s worth noting his plate discipline didn’t improve from 2005 to 2006 and his 2005 season was, fairly forgettable. For what it’s worth, his contact rate did rise by about 3%.
It looks like high contact rates may be able to counter poor plate discipline. It would seem to me that the truly “special” players (with exceptions like Vladimir Guerrero) have that rare combination of power, plate discipline, and contact rates. You see this in players like Barry Bonds, Jason Giambi, Frank Thomas, Carlos Beltran and of course Albert Pujols. Of course, this isn’t the be-all-end-all filter, since there are a few players who sneak in like Casey Blake, who I wouldn’t consider particularly special.
So far we’ve looked at players who had at least 300 at bats, but maybe it’s possible to identify some breakout players from batters who had less than desired playing time.
Here are the players in 2006 who had an OSwing greater than -2% below average and a HR/FB over 12%. I relaxed the HR/FB filter slightly since being able to hit for power might not be quite there yet in younger players.
Name Contact HR HR/FB Hideki Matsui 87.70% 8 12.12% Gabe Gross 76.60% 9 14.06% Chris Snyder 80.31% 6 12.24% David Dellucci 73.62% 13 14.44% Greg Norton 76.61% 17 17.89% J.J. Hardy 85.78% 5 13.89% Derrek Lee 81.04% 8 15.09% Corey Koskie 77.05% 12 17.14% Damion Easley 83.13% 9 14.06% Aaron Guiel 77.69% 7 17.07% Luke Scott 78.83% 10 14.71% Jason LaRue 71.65% 8 15.69% Wes Helms 77.88% 10 14.71% Freddie Bynum 72.73% 4 15.38% Ben Johnson 72.95% 4 12.90% Michael Napoli 68.34% 16 17.20% Chris Duncan 77.25% 22 29.33% Scott Spiezio 80.91% 13 13.83% Corey Hart 75.84% 9 12.16% Russell Branyan 63.90% 18 22.50% Dave Ross 71.51% 21 23.86% Yorvit Torrealba 78.96% 7 16.28% Ryan Doumit 74.43% 6 14.63% Marlon Anderson 81.94% 12 13.79% Daryle Ward 77.90% 7 15.22% Josh Bard 85.16% 9 15.79% Carlos Quentin 74.38% 9 18.00% Joe Borchard 70.60% 10 17.54% Cody Ross 76.85% 13 14.77% Adam Melhuse 73.71% 4 14.81%
This is by no means a “magic bullet” list, but I’d consider it one of many starting points for narrowing down possible breakout players. There are certainly a few players on this list such as Josh Bard, Chris Duncan, Luke Scott, and others that appear to be quite promising. It’s also a reminder that injured players shouldn’t be forgotten such as Derrek Lee, Hideki Matsui, and David Dellucci.
If you were to look at the same list last year, out of about 30 players, you’d have identified Frank Thomas, Jim Thome, J.D. Drew, Mark DeRosa, Milton Bradley, Matt Murton, Ty Wigginton, Nomar Garciaparra, Marcus Thames and Curtis Granderson. So, about one third of the players ended up being at least decent to excellent sleepers.
If you’re still with me, we’ll look at one last list of filters, which I’d consider a sort of potential breakout power hitter list with already established players. I’ll filter on players with an above average OSwing, a contact rate between 70% and 85%, a HR/FB greater than 7.5%, and players all under the age of 30.
Name Contact HR HR/FB Ryan Langerhans 76.50% 7 8.54% Jhonny Peralta 73.48% 13 9.22% Bobby Crosby 77.88% 9 9.09% Jonny Gomes 70.95% 20 13.33% Rickie Weeks 73.58% 8 9.09% Jose Bautista 77.72% 16 11.59% Chris Shelton 73.46% 16 12.60% Edwin Encarnacion 80.15% 15 12.10% Curtis Granderson 70.84% 19 11.66% Matt Murton 83.57% 13 13.54% Jeremy Hermida 78.88% 5 6.17%
Last year, using the same filter, yielded a group of 15 batters who hit 190 home runs in 2005 and 266 home runs in 2006. The group included Carlos Beltran, Grady Sizemore, Mark Teahen, Nick Swisher, Brad Hawpe, and others.
The bottom line is, that since stats like OSwing and Contact do correlate so well from year to year, it would definitely make sense to include them in a projection system (instead of me boring you to death with random filters). I’d say Contact is arguably better than using strikeouts, and OSwing is really unlike any of the traditional statistics that would go into projections.
Print This Post
Wow, how relieving is it to see OSwing%, and Contact% which such high year-to-year correlation. I was literally holding my breath.
I don’t know why i haven’t seen HR/FB yet, that seems like an awesome measure of power. That coupled with Contact% and FB/(FB+GB) should make for a pretty good HR projection system, right? Or does HR/FB not correlate well, year-to-year?
Finally, last question, it doesn’t seem that Contact% and OSwing% correlate very well with each other. Maybe that’s in the study you did last year (you should link to it in this story). Does that seem right?
These pitch-by-pitch studies are the most interesting right now, and i think your approach so far has been elegant. I’ve been waiting for this entry since last spring when you did the initial studies.
Thanks Dave!
I was pretty happy myself to see OSwing and Contact correlated so highly from year to year. HR/FB from ‘05-’06 has about a .61 r^2. I don’t know off the top of my head what other “power stats” look like from year to year, but HR/FB seems as good a quick measure of power as any.
The thing about HR/FB is you have guys like Jacque Jones who have a very high HR/FB, but fail to get the ball in the air on a consistent basis. I feel that it suggests untapped power better than some of the other metrics.
You’re correct that OSwing and Contact have pretty much no correlation. They’re two entire separate skill sets.
Anyway, glad you’re enjoying the pitch-by-pitch articles. I’ll continue to roll them out occasionally. I’d like to start up the Daily Graphing series again if I find the time, which will end up containing some more player specific pitch data.
What is the current league average and range in OSwing? I just want to confirm that when you use the term “above average” or a “+”, the player is swinging at pitches outside the strike zone LESS often than the average, and vice versa. Also, when you give a “percentage” change or difference from the average, is it the actual number of percentage points from the average within the range?
Alan, the average OSwing in 05 was 20.33% and in 06 it was 24.29%. I’m not sure why it increased so much, but if I suspect it has less to do with the players, and more to do with the pitch charting. That’s why I decided to go the above/below average route.
Anyway, when I say a player is above average I do mean he swings at pitches outside the strike zone LESS, as in “above average plate discipline”. The “percentage” change is just the difference in OSwing percentages, so it’s not an actual percent difference, it’s the difference of the percentages.
David, Thank you for that information. With regard to your last list of potential break-out power hitters, how far above average is Matt Murton’s OSwing%…?
Your article’s last paragraph states that it definitely would make sense to include these new statistics in a projection system. However, my understanding is that to obtain a complete player-by-player multi-year set of data, Baseball Info Solutions would charge a lot of money, making that ambition cost-prohibitive.
Alan: Murton is a little under 2% above average in the OSwing department. I liked him a lot last year, and even though he didn’t meet my lofty expectations, I still like him a lot. If I put my prediction hat on: I wouldn’t be shocked to see him hit 20-25 home runs this year.
Michael: It’s true that getting access to pitch location stats isn’t always financially feasible. But if you could get access to them, I bet it’d be possible to improve your projections model. How much it would improve is an entirely different story. There’s a good discussion on InsideTheBook.com on how even the best projection models are only slightly better than a very simple system like Marcel.
It’s possible that including pitch location data in projections could be better, but not cost effectively better.
Either way, there are a number of major (and minor) baseball outlets who might be able to take a risk on the data. If ESPN, Fox Sports, or CBS decided to delve into it (doubtful), they wouldn’t even notice the bill. Maybe it’d even be worth it for BP or BaseballHQ? I know BaseballHQ had a series of articles this past season using 2005 location data for a few pitchers, so it’s at least on their radar.
Raw pitch location data (2005 and forward) will be available for retired players on FanGraphs at some point in the near future. There’s a possibility some current aggregate stats (like OSwing) will be available for all players too. Contact on the other hand can be calculated using Retrosheet event files.
David,
Great article.
I’m curious about Corey Patterson and his 5.10% increase in Contact. Given Patterson’s HR/FB slightly improved (11.6 – 11.9), can you conclude if he swung more selectively or that he just made more contact? Did his OSwing improve as well?
You indicated OSwing and Contact have pretty much no correlation as they’re two entire separate skill sets. Is this true on a micro level or am I trying to connect too much?
Either way, is it fair to suggest that Patterson’s increased contact rate is evidence of an increased skill? Given the high correlation of contact rates, is there any evidence that players who see a large increase in their contact rate continue at that higher rate in the future?
Thanks.
Corey Patterson’s OSwing was considerably worse in 2006 than it was in 2005. He just missed the 5% change list and his OSwing is one of the highest (worst) in baseball.
Did he shorten his swing up this year, because that would validate the increase in contact. I know he was working on that at some point while he was with the cubs, but maybe it didn’t really click with him until this year? He really doesn’t have any plate discipline, but since he’s fast, I’m guessing as long as he makes contact, he’ll beat out some of the grounders to first.
I’d be interested in seeing his ‘04 OSwing since his 2004 was more in line with 2006. Remember this is only 2 years of data and I’ll be sure to revisit this after the 2007 season.
Contact does have a high correlation with strikeout rate, so for a quick and dirty answer, I guess you could ask the same question of strikeout rate: does a single year increase mean an increase of skill? I think more often than not, yes.
Fantastic research. What sabermetrics is all about.