JABO: The Guy Who Breaks FIP
If there was a seminal moment in the movement towards statistical analysis in Major League Baseball, it may very well have been the 2001 publication of Voros McCracken’s research on Defense Independent Pitching Stats, which he shortened to DIPS. In a series of articles over the course of a few years, McCracken demonstrated that Major League pitchers were more or less equal when it came to preventing hits on balls in play. While there were huge and sustained differences in walks, strikeouts, and home runs allowed, the same did not hold true for the rate at which balls were converted in outs when a pitcher gave his defense a chance to get involved. Whether it was Pedro Martinez or Aaron Sele, roughly 30% of all balls in play went for hits, with minimal variation between pitchers.
McCracken’s idea was so antithetical to general wisdom about evaluating pitching that the subject became a primary source of research in the analytical community, but guys like Tom Tippett (now working for the Red Sox) and Keith Woolner (employed by the Indians) mostly ended up confirming McCracken’s original thesis. Despite a conclusion that seemed absurd, for the most part, pitchers really didn’t appear to have much control over whether balls in play went for hits or outs.
Tippett did manage to find a few types of pitchers who could somewhat break the mold, most notably knuckleballers, and further research showed that the data suggests more that pitchers have “little control” rather than “no control” over their hit rates on balls in play, but even with a more muted conclusion, the reality is that most big league pitchers end up in the range of a .280 to .320 Batting Average on Balls in Play, or BABIP as it is usually called these days. There are differences in pitchers in that range, but by and large, that’s mostly where everyone fits.
Everyone, that is, except Chris Young.
Read the rest on Just a Bit Outside.
Dave is the Managing Editor of FanGraphs.
I seem to remember another guy with an absurdly low career BABIP, over more IP than Young…Mariano something.
Andy Messersmith? Catfish Hunter? Tommy Byrne? Hoyt Wilhelm? Mike Norris? Sid Fernandez? Bob Turley?
The only pitcher named Mariano in the history of Major League Baseball had the last name of Rivera, but his babip of .263 was higher than Chris Young’s .249.
Also keep in mind that league average BABIP has fluctuated over time and that reliever BABIP is a few points lower than starter BABIP.
He didn’t say it was lower than Young’s — he said it was absurdly low, over more IP than Young.
The problem with Mariano, and several other elite pitchers, is that very weak contact can actually results in a fair number of hits, either infield dribblers or those Ric Flairs just over the infield. I seem to remember a pretty famous one of those from the 2001 World Series. It was off… Mariano something.
A big reason why, and I’ve said it here many times, is because of his IFFB rate. He induces a ton of weak fly ball outs, and his career 15.2% is one of the highest rates for a SP. He’s also an extreme FB guy who pitches in big parks. The flyball rate helps to lower his BABIP as well, and since he pitches in big parks he won’t be giving up as many HR’s. He’s even managed to do a good job at suppressing home runs (7.9% HR/FB), another sign he’s just amazing at inducing weaker contact.
He’s a unique pitcher for sure.
Definitely unique. It’s not just one thing.
Since 2010: .240 BABIP, 80% LOB%, 7.4% HR/FB%, 8.8% popup%, 0.24 GB/FB. All of that contributes to the 1.87-run gap between his xFIP (5.16) and ERA (3.29). Also witchcraft.
I don’t actually think that his low HR/FB is a function of weaker contact, but rather also related to his IFFB rate. HR/FB is a percentage of all fly balls, not just outfield ones, so it stands to reason that if a lower percentage of his fly balls are to the outfield, a lower percentage are home runs.
We are now seeing actual data on pitcher exit velocity allowed, and the names at the top of the list make a lot of sense, suggesting pitchers do have some control over hard-hit balls. However, the effect that has on a pitcher’s results isn’t entirely clear. Chris Young has the 59th lowest batted ball velocity allowed out of 154 pitcher who have allowed 75 balls in play this year. I think it’s clear at this point he doesn’t allow a crazy-low BABIP because he crazy-weak contact. After all, a pop up hit at over 100 MPH still gets caught over 90% of the time.
It doesn’t explain the whole thing. Even if you remove the popups, Young is elite at preventing outfield flyballs from leaving the park.
Since 2002 (the beginning of the data) with a minimum of 100 GS: Young ranks 9th of 229 in HR/FB%, 13th of 229 in HR/OFFB%.
I think Brian Cartwright may be right: flyball/popup pitchers like Young are simply inducing contact at a very steep angle. That means popups, easy outfield flies, and few homeruns. That angle-of-contact variable would seem to be entirely separate from batted ball velocity.
My model, which uses ground ball, popup and strike out rates, predicts a .253 BABIP for Young for 2005-2014, when he actually had .239. Both the actual and the predicted are the lowest in the dataset.
As Mark points out, Young is an extreme fly ball pitcher who turns many batters into Melvin Upton Jr.
Yeah, part of what makes him unique is that he gets so many outfield popups, which aren’t categorized by our stats.
Think of the distribution of vertical angles in three groups.
First you have grounders, second balls in the air to the outfield, and third infield popups (using Gameday data). Please notice I’ve omitted line drives.
For the balls in the air to the outfield, those close to the ground (0 to +25) will produce a lot of hits and a few homeruns. From +25 to +40 will produce fewer hits but more homers. Above +40 are basically outfield popups.
Regardless of which pitcher is on the mound, all batted balls at a given vertical angle will have a consistent hit and homer rate. The pitcher controls the outcome by controlling the distribution of vertical angles.
Knowing what percent of batted balls are on the ground, and what percent of balls in the air are popups, tells us a lot about the distribution of the balls in the middle – whether they are more likely in the line drive zone, the fly ball zoen, or the outfield pop up zone.
Chris Young’s lack of ground balls and large amount of infield pop ups tells us that his balls in the air to the outfield are much more likely to be at the higher angles which produce lower hit rates.
I have a linear regression formula which gives a .60 r^2 for BABIP on balls in the air to the outfield – pitchers do control that. For anything on the ground the r^2 is below .15.
Do you publish that anywhere?
I had an article in the 2012 Hardball Times Annual on the subject, and am currently working on follow up research.
My intent was to create a regression mean for my Oliver projections using other things that are known about the batter or pitcher, and realized that if actual hits and/or homeruns are entered as zero, it becomes a FIP type stat
Young was in the high-FB% group in this article I wrote back in February – http://www.hardballtimes.com/are-groundball-pitchers-overrated/
He has an incredible amount of rise on his fourseam fastball – tops in baseball this year among pitchers with at least 30IP, at 12.5 inches, and the best career average in the PitchFX era (min 500IP, ahead of Jered Weaver). The rise, location (up in the zone), perhaps combined with his height and extension, have contributed to his ridiculous 15.4% IFFB%. Second best starter (since 2007) is Ted Lilly at 13.7%.
What’s interesting to me is that Clayton Kershaw has always been near the top of the list of fastballs with the most rise, and yet the ground ball rate on his fastball has been average-to-above for a while now. I think location is a critical factor in explaining batted ball distribution, but something else we don’t usually consider is release point. Kershaw has a high release point, and tends to pitch down in the zone. Even though his fastball has a lot of rise, that downward plane might actually produce more ground balls.
I think low release point, high fastball rise, and high zone location all produce fly balls. My hypothesis would be that if Young was very short, and yet could still produce the same fastball rise, and pitch to the same part of the zone, he might get even more fly balls.
But Kershaw also gets a lot of weak fly balls and popups.
I love Chris Young (pitcher). Seeing players — particularly pitchers — succeed in weird and unique ways is really fun. I love watching Max Scherzer and Matt Harvey as much as the next person but dudes who have success barely cracking 85 are so fun to watch and root for.
Others have touched on it, but this is largely a product of spending most of your career in Petco and Safeco. Before this season, the last year where Young had a woba at home higher than on the road (in any meaningful sample size) was 2006, his first year with Padres.
Road woba – Home woba
2006:-77
2007: 61
2008: 75
2009: 154
2012: 63
2013: 125
I know that woba does not equal babip, but it is illustrative regardless. Extreme flyball (and pop up) pitcher in anti-HR parks = low babip.
Home runs don’t count against BABIP, but they count against wOBA in a big way. Are those differences explained by a difference in home run rate?
Home runs do count against BABIP in the sense that if you turn one into an in-play fly ball that it now has only about a 15% chance to fall in for a hit. So they do have an effect on BABIP due to that removal.
But yeah your point about wOBA as being a bad proxy for BABIP still stands.
But yeah your point about wOBA as being a bad proxy for BABIP still stands.
I’m so dumb.
There is a case to be made that the K-rate in FIP should be replaced by K+IFFB (and constants adjusted accordingly). Koji Uehara also has a .250 MLB-career BABIP driven by a lot of IFFB. He does it by having a higher than normal amount of spin on his fastball, which causes it to “stay up” more than hitters expect.
For Fangraphs’ WAR, it is.
Don’t beat yourself up about it, BM. We all do it sometimes.
I’m doing it right now.
Interesting read, but I felt Dave could have focused on the *why far more than the *rest of season predictions.
Here’s my take on it:
– If Chris Young’s 4-seamer was a few MPH faster, it would move completely over hitters’ bats, he’d be getting a league-average number of strikeouts, and we wouldn’t be talking about his BABIP.
– If his 4-seamer was a few MPH slower, hitters would be squaring up on it too much for him to stay in the majors, and we wouldn’t be talking about him at all.
– At the “sweet spot” velocity that he actually throws the 4-seamer, hitters make contact but on the lower part of the ball, leading to tons of pop-ups and high-angle fly balls.
How do you break something that’s already broken? There’s tons of examples of guys outperforming their peripherals, FIP, BABIP, etc. There is an art (or maybe a science, for you nerds) to inducing weak contact. http://twtwsports.blogspot.com/2015/05/danville-kentucky-folks-by-now-youre.html