Carl Pavano’s 2010: Trading Whiffs for Grounders
Although Carl Pavano had nearly identical FIPs in 2009 and 2010 (4.00 versus 4.02), he achieved them in quite different ways. Pavano, who re-signed with the Twins yesterday, did a great job of limiting walks in both years. But in 2009 he combined that command with average-ish strikeout and ground-ball rates. In 2010, however, had just 4.8 Ks per nine (two per nine fewer than in 2009 and sixth lowest in baseball), but induced grounders on over 50% of balls in play.
Usually such a change in strikeout and ground-ball numbers is the result of a drastic change in pitch usage, see Joel Pineiro’s 2009. But it looks to me like Pavano’s pitch usage has not changed that much from 2009 to 2010:
| Pitch Type | 2009 | 2010 |
|---|---|---|
| Two-Seam Fastball | 0.372 | 0.384 |
| Four-Seam Fastball | 0.196 | 0.188 |
| Change up | 0.229 | 0.230 |
| Slider | 0.203 | 0.197 |
Instead the change comes from each of his pitches getting different results in 2010. Almost across the board his pitches have higher GB% and lower whiff (swinging-strike) rates. The only exception is his four-seam fastball had slightly higher whiffs in 2010. Swinging strikes and ground balls generally trade off, but it is interesting to see to play out so clearly for one player from year to year.
The difference is the most striking with Pavano’s slider which went from 37.5% GBs in 2009 to 59.2% in 2010, accompanied by a decrease in whiff rate (per pitch) from 12% to 10%. It looks to me that this was caused by location in the strike zone. Here is a plot showing the density of his slider to right-handed batters in 2009 and 2010, from the catcher’s perspective (so negative x values are inside and positive away).
You can see that he is getting his slider low and away much more in 2010. This is where a pitcher wants his slider to end up, and a location that will induce lots of weak contact on the ground.
Looking towards 2011, if Pavano can keep limiting walks and getting grounders he will be a valuable asset to the Twins even if he strikes out just a batter every two innings. But the margin for error for low-strikeout pitchers is small, and a little bump in walks or fly balls could be big trouble.
Hat tip to Millsy for the idea of using smoothScatter in R to make these types of pitch-density graphs.
Dave Allen's other baseball work can be found at Baseball Analysts.
Millsy does some great work at his blog. I’ll try smoothScatter next time. Looks great.
This sort of sounds to me like his GB% is a great candidate to regress, and that he may get hammered this year.
Surely where a pitcher places his pitches is mostly skill. This is why you make projections based on things like GB rate, because they don’t usually just reflect random variation around the mean.
why does everybody seem to assign a high ground ball rate to pitching success? the majority of great pitchers have been fly ball pitchers.
Adam,
All else being equal a pitcher would prefer ground balls to fly balls. More double plays and fewer extra-base hits. In reality pitchers who get lots of strikeouts generally get lots of fly balls. Since strikeouts are even better than ground balls these are generally the best pitchers. Guys who get lots of strikeouts and ground balls (Felix Hernandez, Josh Johnson, Ubaldo Jimenez) are very good.
Here is a nice explanation:
http://baseballanalysts.com/archives/2010/03/categorizing_st.php
Dave,
Sounds good in theory. Here’s the reality: Of the 26 HOF pitchers for which data is available, 18 had higher than league average fly ball rates. The only strong groundball pitchers were Whitey Ford (1.32) Bob Lemon (1.51) and Gaylord Perry (1.33). Nobody else was above 1.2 for their career. Of those recently retired who are can’t miss HOF candidates, Clemens (1.15), Glavine (1.23) and Maddux (1.86) are above league average (which is around 1.07): Pedro (0.90), Johnson (0.97), Smoltz (1.02) Hoffman (0.67) are below. If you look at the next tier of “maybes” you get Kevin Brown (2.15) as the only ground ball guy. The rest are fly ball pitchers.
If you look at the Cy Young winners, near greats, and HOF pitchers, over 60% of them are fly ball guys.
Here’s a list of Cy young winners and potential HOF candidates who are recently retired (not including the pitchers listed above)
Fly ball: Sabathia (1.05), Lee (0.75), Morris (1.07), Greinke (0.90) Colon (1.03), Zito (0.79), Hentgen (0.87), Peavy (0.97), Gagne (0.79), McDowell (0.91), Welch (0.88), Saberhagen (1.00), Viola (0.90), Hernandez (0.99), Davis (0.94) Bedrosian (0.73), Scott (1.03), Sutcliffe (0.90) Stone (0.87) Cone (0.78), Santana (0.80), Wells (0.99), Schilling (0.94),
Ground ball: Lincecum (1.19), Webb (2.91), Carpenter (1.47). Hoyt (1.10), Vukovich (1.25), Drabek (1.13), Hershieser (1.88), Gooden (1.16), Denny (2.11), Valenzuela (1.12), Flanagan (1.24), Hernandez (1.88), Halliday (1.79), Rivera (1.33)
If you are drafting from the ground ball list, and I’m gonig to draft fro the fly ball list, I’m going to win, in both quality and quantity. It gets even more pronounced when you look at HOF only.
HOF ground ball: Blyleven (1.11), Ford (1.32), Drysdale (1.19), Lemon (1.51), Newhouser (1.1) Niekro (1.13), Perry (1.33), Wihelm (1.16)
Only 8, and half of them are close to league average.
HOF fly ball: Ryan (0.92), Gibson (0.99), Paige (0.78), Koufax (0.69), Seaver (0.99), Hunter (0.66), Bunning (0.74), Carlton (1.00), Eckersley (0.67), Feller (0.79), Jenkins (0.99), Fingers (1.00), Marichal (1.03), Wynn (0.66), Palmer (0.80), Roberts (0.71), Spahn (0.99), Sutton (0.89)
If I’m chosing fly ball over ground ball, I win 18-8. If I pick all pitchers with a ratio under 1.2, I win 22-4.
I probably should have kept my mouth shut (easy to kick butt drafting pitchers when fly ball guys are so undervalued,) but the current meme that a ground ball is better than a fly ball is the complete opposite of reality. There are plenty of reasons. Lets start with the fact that a ground ball requires three successful actions to make an out–a catch, a throw, and another catch. A fly ball requires only a catch. Or you could start with the fact that there are 9 fielders who can turn a fly ball into an out. Except in extremely rare circumstances, (say, Ernie Lombardi getting throw out by an outfielder) there are only 5 players who can turn a ground ball into an out. Or you could mention that there are no bad hops it the air–there is sometimes wind, but that can work to take a hit away as well. I think people got fired about ground ball rates because there are currently some outliers active in the game. When Maddux enters the Hall, he will be the greatest ground ball pitcher in history to enter the Hall.
I am curious about the relationship between extreme ground ball guys and arm injuries.
oops, made a mistake above. If I am chosing HOF pitchers with ratios under 1.20, I win 23-3, not 22-4. That’s a Ron Guidry type year. I hadn’t mentioned Guidry above. He was a fly ball guy (0.84). Thought it would be piling on.
So I’m trying to wrap my head around these new fangled stats. What I don’t understand is how FIP can stay the same between two years when K/9 decreases and GB rates increase. If it is fielding independent how can it (FIP) not increase when more outs seem to be coming from fielding put outs rather than K’s?
More ground balls means less home runs.
If you take Pavano’s rates from 2009 and compare to 2010, he lost 46 strikeouts in 2010. That costs him 0.42 runs/9 in FIP. He saved 5 HR, which gained him 0.28 in FIP, and he saved 6 walks, which gained him another 0.08 in FIP.
Makes sense, thanks!
Is there a way to see how many fielding outs a pitcher records in a season?
Steve, IP * 3 – k?
because FIP is very limited. K’s only account for about 25% of the outs in a game, so we’re missing out on nearly 75% of the pitchers actual performance. This is a major flaw in sabermetrics, WAR, and Fangraphs.
Thank you for your uninformed opinion. Please go buy a copy of The Book, read up, and come back when you understand why you are so, so wrong.
Hey Dave,
Thanks for the shout out. It looks like you were able to custom format the axes, which I was having a hell of a time figuring out with the smoothScatter function. Did the default just work out that way for your plots because the range was the same for the two (2009 and 2010), or were you able to figure out a way to custom format them easily?
Thanks.
Hey Millsy,
I usually supress the default axes by adding the argument xaxt=’n’ and yaxt=’n’ in the plot or smoothScatter or whatever command. Then I add the axis back in exactly where I want them with axis.
Thanks. I’ll have to try that.
I had originally tried “smoothScatter(y~x, xlim=c(-x,x), ylim=c(-y,y))” to no avail, which usually works just fine for regular scatter plots.
Wonder if the decrease in K’s might be related to velocity. He’s placing his slider down and away with more consistency, might he be sacrificing some zip to get it there more precisely? It would probably help explain the higher contact rate, as velocity and swing-throughs are generally thought to correlate pretty well.
It’s a shame there isn’t anywhere to see that information quickly and easily.
http://www.fangraphs.com/statss.aspx?playerid=790&position=P#pitchtype
Doesn’t look significant to me, so I guess not. Thanks for the snark and stay classy, whoever you are..