Some strange statistical anomalies show up in my SABR research presentation attendance this year. For example, somehow I ended up seeing everything in the Catalina Room and didn’t see anything in the Pacific Room. I also didn’t see anything on the New York Yankees except Herm Krabbenhoft’s one about correcting the RBI record for Lou Gehrig and Hank Greenberg. And I didn’t see any stats analysis until the third & last day of presentations, which is also unusual.
But that’s baseball, sometimes it just doesn’t go as expected.
Last night’s game at Dodger Stadium went in some ways exactly as expected For example, everyone was predicting a low-scoring game given that the Dodgers and Padres are both a bit anemic in their lineups this year. The score was 1-0. And that one run took three hits to bring in. And then there are the things no one expects, like the “closer” (Broxton is on the DL) Guerra coming in and loading the bases on a leadoff double and then two consecutive hit-by-pitches… and then wiggling out of the jam with two strikeouts and a line drive caught in center field.
Tonight we’re off to see the Angels and I don’t know whom to root for. My usual policy is to root for the home team whenever I’m not seeing the Yankees on the road, but I like the Mariners and I have never liked the Angels. I suppose I’ll see how I feel when I get there.
I saw four research presentations today. I was trying to see all 5 time slows, but couldn’t quite get down there early enough for Mark Pankin’s presentation which asked the question “Did having no lights hurt the Cubs?” I’ll have to ask for the answer if I see him in the hallway.
EXAMINING HOME FIELD ADVANTAGE
Phil dug into one of the burning questions, which is thee home team wins in baseball 54% of the time. But WHY?
Phil quickly ran through some of the potential reasons and why they have been discounted in previous studies: fan enthusiasm, ballpark familiarity, home cooking (comfort in surroundings, one’s own bed, family), travel stress on the away team, batting last, etc…
But what about umpires? A recent book (Scorecasting, by Tobias Moskowitz and Jon Wertheim) was released earlier this year and attributes pretty much all of the Home Field Advantage to umpiring/refereeing bias in all sports. They found that more strikes are called for the home team in baseball, for example, and that this effect increases the higher leverage the situation is! I.e. with the game on the line, the home team is more likely to get the call their way than the visitor would, and that the opposite is also true, that in low-leverage situations (like a blowout) the visiting team was more likely to get the call.
However, Phil didn’t think this could account for it all, and he analyzed low-leverage situations (where the lead was 4+ runs) and did not find as much of an effect as Scorecasting did. Mitchel Lichtman (MGL) also did an analysis and found a slight positive effect but not as big as Scorecasting. Three studies have been done using Pitch F/x data. John Walsh in “The Hardball Times Annual” found that home team was favored total by 0.8 pitches per game, which would equal about a sixth of a run or a seventh of a run. That would only be about a third of the total home-field advantage seen. J-Doug in “Beyond the Box Score” found umpire strike-calling accounted for about a sixth of HFA, and Dan Turkenkopf found it to be about one-eighth.
So what else can it be? Phil tried to look a umpire or referee-independent stats and performance, including free throw shooting in basketball, and found them very difficult to separate out. But there were some. Like for example, in speed skating there turns out to be a home field advantage! That can’t be a refereeing bias there.
Phil is forced to conclude after all his sharp analyses that a) we need to come up with some better ways to tease out umpire-independent performance and b) there is probably some universal biological or intrinsic process at work. One theory is that it may be a territoriality thing, i.e. testosterone or “mother bear defending her cubs” that kicks in when people are playing on their home turf. That’s beyond the ability and scope of most baseball statistics folks to test.
The complete slides to Phil’s talk are up on his website: www.philbirnbaum.com.
UMPIRE TREAMENT OF ROOKIES
by Pat Kilgo and a team of researchers
This was a top-notch presentation as stats ones go, and I doubt I can do it justice here because the full effect can only be gotten from looking at the graphs and slides that demonstrated the proof. In short, there are all these anecdotes about umpires giving the benefit of the doubt to either a veteran pitcher or a veteran hitter, or a good player over a not so good one. Famously umpire Bill Klem supposedly told a young pitcher who protested he had throw a strike that “Mr. Hornsby [the batter] will tell you when it’s a strike.” Reggie Jackson said toward the end of his career umpires would tell him that if he didn’t swing they wouldn’t consider it a strike. On the other side Bob Uecker tells of being a rookie and complaining that a called strike wasn’t, and being told not only was it a strike, “So’s the next one.”
Kilgo and his team tested this using Pitch f/x data. Using all available data, they examined 738,496 judgement calls on ball/strike calls and counted as strikes any ones that any part of the ball touched the strike zone. (Pitch f/x can have an error of up to half an inch.) A ball that was called a strike was called a “false strike,” and a strike that was called a ball was a “false ball.”
They found the total umpire accuracy rate on these judgment calls to be 84.9%. The false strike rate: 7.4%, false Balls: 7.7.
Now, if the anecdotes are true, we should see higher false strike rate for veteran pitchers, and higher false balls for veteran batters. In fact, their data demonstrate this quite clearly. Generally speaking the older a pitcher is (more years in the league), the more likely he is to get the calls his way. Likewise with batters, and also catchers in the batters box appear to get an advantage. For example, of the top five guys favored with false balls while batting, three of them are catchers. Of the pitchers, check out how many of them are closers:
Livan Hernandez 13.8% false strike rate
Mariano Rivera 13.6%
Jake Peavy 13.6
Brian Wilson 13.5
Pretty cool. They also showed some effect of star pitchers versus non-star.
STARTING STAFFS AND PITCHING ROTATIONS
David, the founder and president of Retrosheet, always digs statistically into some common myth or assumption about the game. Sometimes he uncovers a bombshell, other times perceptions are upheld. In this one, there are some surprises and some common themes upheld. I can’t come close to recreating the numerical sleuthing that results from his step by step journey through the numbers, but I can repeat the part that surprised me, which was the irrefutable demonstration that the five man pitching rotation has been the standard since 1910. You always hear about the supposed days of the 1930s, 1940, 1950s, when “men were men” and pitched every third day if necessary, etc… Well, it would appear that the last time that was actually a common occurrence was around 1901. For ONE HUNDRED YEARS the 5-man starting staff has been the most common configuration by a long shot.
Also a bit against “common wisdom” is that the more consistent teams, i.e. the ones who used the same guys in the same order all the time, were NOT necessarily the better teams or the more successful teams. There was no correlation at all.
STARTING PITCHER RATING SYSTEM
The final presentation of the conference was this one by Vince Gennaro, another previous winner of the best presentation award and newly minted SABR president, on his work to devise a system that can actually rate the value of starting pitchers. Vince (who is a Yankees fan like me) started his presentation with the announcement that Derek Jeter had just gotten his 3000th hit, off David Price, a home run. (Ironic since Jeter typically doesn’t hit home runs.) I already knew because my mother had texted me the second it happened. (And as I type this later she just texted to say he went 5-for-5. Jeter defies statistical explanation.)
In building his system, Vince took into account several things that stats like ERA and wins just don’t, like: does he go deep in games and save the bullpen? Is he equally effective against right and left-handed batters? What kind of hits come off the bat from this guy? This rating system rewards “efficiently prevented runs.” It does not only rely on “outcome” stats (like how many hits allowed) but also “process” stats like strikeout percentage, swings and misses, fastball velocity, and release point. The system also rewards consistency of game to game performance. (He got a big laugh when he showed the chronological graph of Jake Westbrook’s performance which seemed pretty up and down… until compared with A. J. Burnett’s. Burnett. Jeez.)
Overall the system is still evolving, but Vince demonstrated the importance of saving the bullpen (not only reducing the number of bullpen innings but the effect of keeping the bad guys in the pen on the bench) and also the value of ground balls over fly balls. None of these individual facts are mind-blowing, but figuring out how to add them all together into a rating system that is consistent with rewards and valuing what a player is worth is the interesting part, and the part that is still evolving. Given that the conference opened with Scott Boras essentially saying that he already has some system sort of like this proprietary for his own clients it was interesting to “look under the hood” and see how Vince is tinkering with the same sort of machine. I’m looking forward to seeing how this evolves in the future.
And now it’s time for the ballgame! Off to Anaheim we go.
(Did you enjoy reading this blog entry? Please consider buying me a hot dog.)