Archive for Sabermetrics
Jacob Luft at SI.com uses PrOPS to break down the luckiest and unluckiest players this season. If you’re unfamiliar with PrOPS you can read further here and here. Players who have been lucky, and putting up numbers better than they way they have hit the ball are likely to decline over the rest of the season, while unlucky players ought to improve. Luft explains:
Because breaks tend to even out as the sample size of data grows, PrOPS is a powerful tool in figuring out which players will benefit and which ones will suffer as their statistics regress to the mean. For example, Reds outfielder Austin Kearns posted a real OPS of .785 last season, but his PrOPS was .840, indicating that he was a better hitter than his statistics were giving him credit for. This season, Kearns’ real OPS is almost identical to his PrOPS from last season: .852. If you look back at the top 25 underperformers for 2005 as calculated by PrOPS, you’ll find others who have improved this season, including last year’s leader, Jason Giambi, and Mike Lowell.
Jacob indicated to me that PrOPS influenced him to pick up Austin Kearns for his fantasy team this year—thankfully that’s worked out well, so far. I’m not a fantasy player, but I’m curious how many people find PrOPS helpful or unhelpful. Drop me a line if you have a comment. Thanks to Jacob for furthering awareness of the statistic. Also, thanks the The Hardball Times for keeping track of it.
It’s no secret that the Braves pitching is awful right now, which is the main reason for the team’s recent woes. With the departure of Leo Mazzone, who has received much credit for the excellent Atlanta pitching staffs during the Braves division title streak, it’s a fair question if his absence is part of the explanation. While acknowledging that it’s somewhat unfair to look for any impact this early in the season, I’m going to do it anyway. When the O’s and Braves meet at the end of the month for an interleague series, it’s going to get talked about quite a bit anyway. The small size caveat applies, interpret at your own risk.
In my earlier analysis of Leo Mazzone for the Baseball Analysts, I acknowledged that one of the more difficult problems in assigning credit and blame to Mazzone is that he was not the only stable factor involved in the Braves’ pitching success. Leo came on board with the Cox-Schuerholz regime; therefore, it’s entirely possible that it was something in the organization other than the pitching coach that caused Atlanta pitchers to improve when they arrived and decline when they left. After much reading on the subject—since I published the study I’ve read nearly everything written on Mazzone’s method—I believe that there is something to his unique approach. Now that Mazzone has moved on, we have the opportunity to view the coach and organization separately.
Today, I want to look at a group of pitchers who pitched for Mazzone last year in Atlanta and are still pitching in Atlanta this year. This holds many things constant, since the main difference is the lack of Mazzone. Sure, the players are one year older—bad for older pitchers but good for younger ones—and other factors such as weather, personal matters, etc. may also have an effect; but, this should be fine for a rough, and admittedly premature, analysis. So, how have these pitchers performed this year versus last year? Here are the ERAs and Fielding Independent Pitching (FIP) ERAs for these pitchers.
ERA Pitcher 2006 2005 Difference Horacio Ramirez 5.06 4.63 0.43 John Smoltz 3.73 3.06 0.67 Tim Hudson 3.93 3.50 0.43 Macay McBride 4.08 5.79 -1.71 John Thomson 4.84 4.47 0.37 Jorge Sosa 5.58 2.55 3.03 Kyle Davies 6.12 4.93 1.19 Chris Reitsma 9.11 3.93 5.18 Mean 1.20 Median 0.55 FIP ERA Pitcher 2006 2005 Difference Horacio Ramirez 4.22 5.15 -0.93 John Smoltz 3.39 3.14 0.25 Tim Hudson 3.47 4.22 -0.75 Macay McBride 3.93 1.34 2.59 John Thomson 4.67 3.39 1.28 Jorge Sosa 5.54 4.13 1.41 Kyle Davies 6.01 4.29 1.72 Chris Reitsma 6.76 2.82 3.94 Mean 1.19 Median 1.35
I wouldn’t put too much stock in the precise magnitude of these numbers, but it’s clear that the direction of the overall quantitative impact of this year versus last year has been negative. It’s exactly what we would expect if Mazzone was responsible for the improved pitching. The sample size is certainly too small to yield any statistical significance, but it’s still fun to observe. Even if the effect is real, I think it’s also unwise to say that this is the result of the differing methods of Roger McDowell and Mazzone. Even though Mazzone’s methods have brought results, McDowell may also have good methods that haven’t had time to bear fruit. I do think the pitching coach transition is part of the problem. Transitions to new work environments are always difficult. It could be that the Braves pitchers are still adjusting to a new advisor. Mazzone is not working any miracles in Baltimore, or at least it’s not obvious if he is, considering the they have the second worst ERA in the league. But, the Orioles don’t have the quality of arms that the Braves do—Smoltz and Hudson are far better than any pitcher on the O’s—and the organization is in much more flux.
As a Braves fan, I see the loss of Mazzone possibly affecting two pitchers: Reitsma and Sosa. Both seem to have lost none of their stuff, but consistently make placement errors with pitches. That’s something that Mazzone’s throwing program is designed to help. Thomson’s recent struggles also seem similar. But again, it’s too early in the season to say anything definitively. I would love for the O’s and the Braves work a trade involving some of these pitchers to see if Mazzone can work his magic. I don’t think either side has much to lose.
Addendum: Rick Maese of the Baltimore Sun has some more about this on his blog. He points to signs that the O’s pitching may be turning around. I guess we’ll see soon enough. I don’t think Mazzone just happened to be good for 15 seasons in Atlanta, so while some may be bailing on Mazzone, I wouldn’t bet against him.
In the larger picture, this book apparently mirrors the trend in all sports of putting more of an emphasis on numerical evidence and drawing conclusions based on numbers, like Linkstigator John Hollinger does or like they do at 82games.com. I understand the movement and in some ways I appreciate the way numbers make everything so cut and dried. But at the same time, I find it impossible to rely solely on numbers. For instance, a couple of weeks ago Sam sent me an article from sabernomics.com about Braves outfielder Jeff Francoeur. Francoeur started the season slumping, and this article went back and analyzed all sorts of numbers and determined that “There is a real problem,” that the Braves were “going to have to gut this one out.” The author also called for everyone to stop pretending that Francoeur is an All-Star.
Since then, Francoeur warmed up, and he’s now up to 10 home runs and 36 RBIs (good for 6th in the National League).
So what happened? Well, Francoeur made adjustments. He figured out what he was doing wrong and fixed it.
And I guess that’s illustrative of my main beef with the reliance of numbers in evaluation. I loved Moneyball and thought it was a great read. Given the circumstances on Oakland — no money to spend, etc. — it’s worked really well.
But at some point, don’t we have to just look a guy and watch him play and say whether or not he’s going to be good or bad or a good fit for our team?
Since then—refering to the time since Whitaker posted the article— Frenchy has posted a .200/.213/.267/.479 line in 47 PAs. For the season, he’s put up an ugly .247/.260/.426/.686 .
As to the insinuation that I only looked at numbers, I’ve watched nearly every at-bat of Francoeur’s this season. I know that if you throw it high and inside, he can’t hit it; and I haven’t seen any stats on his hitting zone this season. You don’t need to: Frenchy looks bad up there. He is talented and can hit the ball a long way when he connects, but he is an out-machine right now. My eyes tell me this, and the stats confirm it. I suspect hit OPS will not stay below .700 for the entire season, but he’s not an All-Star yet. If you think otherwise, play The Jeff Francoeur Game.
It’s hard not to wonder, after reading “The Wages of Wins,” about the other instances in which we defer to the evaluations of experts. Boards of directors vote to pay C.E.O.s tens of millions of dollars, ostensibly because they believe—on the basis of what they have learned over the years by watching other C.E.O.s—that they are worth it. But so what? We see Allen Iverson, over and over again, charge toward the basket, twisting and turning and writhing through a thicket of arms and legs of much taller and heavier men—and all we learn is to appreciate twisting and turning and writhing. We become dance critics, blind to Iverson’s dismal shooting percentage and his excessive turnovers, blind to the reality that the Philadelphia 76ers would be better off without him. “One can play basketball,” the authors conclude. “One can watch basketball. One can both play and watch basketball for a thousand years. If you do not systematically track what the players do, and then uncover the statistical relationship between these actions and wins, you will never know why teams win and why they lose.”
The WoW is by three economists—Dave Berri, Martin Schmidt, and Stacey Brook— who have been doing some of the best research in sports economics over the past few years. I happened to end up at a dinner with Dave Berri at the Western Economic Association meeting last July, where I learned about the book. I thought the book had a lot of potential, given the work these economists had done. It reminded me of what William Easterly did in The Elusive Quest for Growth with his amazing academic work on economic growth.
Because of our similar research interests Dave sent me a few finished chapters, and it was even better than I thought. I’ve mentioned it a few times over the past few months, because of my excitement over what I have read. I think the readers of Sabernomics will enjoy its discussion of not only baseball, but other sports as well. The chapters on basketball are exceptional. Now that I have the entire book in front of me, I find that it has exceeded my already high expectations. Sometimes I wonder if my opinion is the exception or the consensus, but it turns out to be the latter. In addition to Mr. Gladwell, the book has also received praise from Alan Schwarz (cover blurb) and Tyler Cowen.
I’ll try and post a more thorough review of the book soon, but I’m a bit tied up with some other projects, so I can’t make promises. If you read the book, or find other reviews, and want to post comments in this thread, go right ahead. Also, check out the authors’ blog.
Recently, two such researchers, Arthur De Vaney and John Charles Bradbury, both of them trained economists, separately developed a remarkably counterintuitive proposition: Steroid use in baseball over the past 10 years had nothing to do with the home run records set during that time.
They offer no opinions about steroids, and no opinions about who used them and how often and who’s lying and who isn’t. They’re just looking at the numbers. …
A lot of people aren’t going to like this idea. They are convinced that Barry Bonds achieved his records by cheating. Here’s the weird part: It is perfectly possible that Barry Bonds cheated but that the cheating had nothing to do with the records. So if you are worried about the taint on the game brought about by illicit drug use, fine. Boo away. But of you think the records are tainted — well, check your math. Everything you know could once again be wrong.
Thanks to Bryan Donovan and the guys at The Hardball Times, you can now get player PrOPS which are updated daily. And, just like the other stats at THT, you and filter the data in many different ways. Bryan has really done an amazing job with all of the stats.
I had hoped to write a PrOPS article this week, but I’ve had a bit more to do than I anticipated. So, I’ll quickly mention one minor change. This version of PrOPS is based on data from the 2002-2005 seasons. Unlike previous versions, this PrOPS is park-neutral, instead of showing what a player ought to do because of his home park. This makes comparisons across teams a little easier.
And if you’re wondering, “how accurate is PrOPS?” you need to purchase a copy of The Hardball Times Annual 2006.
I received a lot of feedback on my Jeff Francoeur post from the other day. Most recently, Jacob Luft of Sports Illustrated picked up on it (thanks for the mention and kind words, Jacob), so I thought I’d add one other quick note on Frenchy.
I’m in the process of generating the initial 2006 PrOPS numbers for all players in the league. It’s going to take me a few days to make sure everything is working properly, but I thought I’d throw out Francoeur’s numbers now. I’ll post all the numbers over at THT as soon as I get them ready. If you’re not familiar with PrOPS, check out my initial post at The Hardball Times or see my article in the 2006 THT Annual. PrOPS gives players credit for how they hit the ball rather than on the outcome of the hit balls.
Jeff Francoeur AVE OBP SLG OPS Actual: .210 .222 .352 .575 Predicted: .269 .282 .441 .723 Difference: -.059 -.060 -.089 -.148
He’s been hitting the ball better than his numbers indicate, but an OPS of .723 is still not very productive.
I’ve been wanting to post some these thoughts for some time, even before the season started. But, I kept putting it off, thinking that The Natural would prove me wrong. And I was hopeful that the kid had something to teach me. I’ve hinted at my thoughts on Francoeur publicly, and discussed them more in private, but I guess I should go ahead an put my thoughts in a single post. Jeff Francoeur’s 2005 was a fluke. And it was flukey for more than one reason. That doesn’t mean he won’t be a very good baseball player one day, but his current performance in 2006 should have been expected. I don’t care how much “make-up” you have, if you can’t lay off bad pitches you’re not going to excel as a hitter.
Second, I believe the evidence indicates that part of Francoeur’s fast start was the result of poor scouting. In Mississippi, Francoeur posted a line of .275/.322/.487/.809 against double-A pitching. The funny thing is that his performance in the minors was slightly worse than is PrOPS numbers in the majors. Less capable minor league pitchers knew something that major league pitchers didn’t, or the Braves had Jeff on some bizarre hitting program. The way Jeff tailed way off over the rest of the 2005 season is consistent with major leaguers getting good scouting reports on the guy.
This leads to another interesting question: why did it take so long for major league teams to figure him out? There were certainly scouts watching him in double-A, why didn’t they pass along what the minor league pitchers were doing? My guess is that Francoeur’s jump surprised everyone, and that scouts were not scouting him like advance scouts typically do. Instead, they focused on his raw ability and promise. Scouts saw his poor plate discipline and just reported, “he’s not ready yet, fire it in there.” And well, that was very bad advice. And because Francoeur is blessed with amazing power, when he got pitches he could hit he hit them along way. He didn’t fluke his way to 14 home runs, you have to be gifted to hit home runs. But I think with good advance scouting reports he would not have been nearly as successful—maybe half of those homers go away. In fact, one thing teams may have learned from this experience is that unexpectedly pulling up kids from the minors can yield benefits, because other teams lack the information to get these guys out. Instead of contenders looking to get Joe Randa through a waiver-wire deal for the playoff push, maybe teams should pick up a talented prospect whom no one expected to see.
And why is it that minor league pitchers figured him out? Well, look at the incentives for the pitchers in double-A versus those in the majors. If a double-A pitcher wants to move up, he has to get outs. The best way to do that is to prepare for the guys you’re going to face, especially the best players on the team. These pitchers saw he liked to swing at everything—a friend of mine who watched him in high school said this was no secret then—and they stayed away from the zone without fearing the free pass. But for major league pitchers, Francoeur was just another rookie. Why worry about him when you’ve got to face the Jones boys? And that’s when Frenchy’s window for success opened.
The problem is that now that the window has closed, what are the Braves to do? He’s nearing the 100 PA mark, without having walked even once. And he’s leading the league in swinging at first pitches, so his pledge to work on plate discipline is not going so well. Also, he’s only had five extra-base hits, so he’s not hitting for power when he does hit the ball. This isn’t a bad-luck, small-sample-size slump. There is a real problem.
So, what should the Braves do? Some people think he should be sent down. I don’t think you can do that now. He’s been in the big leagues too long. If he goes to Richmond, all he’ll be thinking about is how to get back. I think the mental fatigue would be too much. The Braves are just going to have to gut this one out, and let him learn on the job. But, it is time to stop pretending he’s already an All-Star. Moving him down in the batting order might reduce some of the pressure, and he could split some time with Diaz and (gulp) Jordan. He’s still an excellent defender and baserunner, too. There are plenty of players in the league who are no worse. Most of them don’t get to play as much, though.
UPDATE: This post is old, and uses a fairly weak method for measuring peak age. I have conducted a thorough analysis published in Journal of Sports Sciences that finds peak age occurs around 29–30.
— — —
Judging the peak age of baseball players is an interesting problem. Over the course of a player’s career, many factors may affect his numbers that have little to do with age: injuries, differing park factors, changes to the run environment of the league, long-run contract incentives, etc. I have looked at the peak ages of pitchers and hitters before (see the right sidebar for links), and I’m happy with the results of found, despite the potential problems. But, I’ve thought about another way to find peak age. Instead of looking at the changing performance of players with long careers, I wanted to look at those with short careers.
Players with short careers typically play only during their peak years. Only when they are at their best are these players providing “major league” level of talent. Both before and after, their skills are not good enough to keep them in the league. Therefore, I’m going to put all of the stats aside and only look at the average age of players with three years or less of major league experience. The average of these short-timers should tell us something about when players peak. I used a recent sample of players from 1980-2002. By excluding 2003-2005 I exclude young players with three year careers who may have much longer careers.
For both hitters and pitchers, the median age of players with careers of three years or less is 26. The mean age for hitters is 26.02 with a standard deviation of 2.37 years. The mean age for pitchers is 26.33 with a standard deviation of 2.82 years. Interesting. I haven’t though much about this other than I thought it would be a neat study to do. Comments, as always, are welcome.
I’ve been really busy lately—I know who hasn’t?—which has significantly postponed some things I wanted to do. For example, I have been wanting to post the SSPS Projections for 2006 for a month. I’m only going do so for hitters this year. If you want to know what’s in SSPS (Sabernomics Simple Projection System) see the introductory post from last season.
I’m not going to defend it against other systems, it’s just more information. Use it as you wish. I don’t predict playing time or make projections for new teams. The projections assume that hitters play in the same home park as they did in 2005. For players who played on more than one team, the projections assume the player played for the team on which he played the most.
Here’s a list of the top-20 projected hitters (by OPS) by league. Enjoy!
Addendum: Something is screwed up with my projections for the Nationals. I’ve removed them, and I will add them when I get a chance. Sorry.
NL Rank Player Team05 AVG OBP SLG OPS 1 Adam Dunn CIN 0.282 0.412 0.583 0.995 2 Todd Helton COL 0.321 0.425 0.552 0.977 3 Chipper Jones ATL 0.300 0.411 0.560 0.972 4 Albert Pujols SLN 0.315 0.411 0.560 0.971 5 Derrek Lee CHN 0.299 0.384 0.572 0.956 6 Andruw Jones ATL 0.287 0.371 0.577 0.948 7 Jim Edmonds SLN 0.276 0.390 0.551 0.940 8 Lance Berkman HOU 0.294 0.400 0.519 0.919 9 J.D. Drew LAN 0.289 0.397 0.518 0.916 10 Morgan Ensberg HOU 0.281 0.381 0.534 0.915 11 Ken Griffey CIN 0.288 0.389 0.525 0.914 12 Tony Clark ARI 0.276 0.358 0.555 0.913 13 Jose Cruz ARI 0.269 0.380 0.533 0.912 14 Javier Valentin CIN 0.293 0.396 0.516 0.912 15 Russell Branyan MIL 0.265 0.381 0.527 0.908 16 Troy Glaus ARI 0.272 0.366 0.538 0.904 17 Jason Bay PIT 0.288 0.387 0.508 0.896 18 Dustan Mohr COL 0.269 0.342 0.551 0.892 19 Preston Wilson COL 0.284 0.357 0.535 0.891 20 Carlos Delgado FLO 0.284 0.373 0.517 0.890 AL Rank Player Team05 AVG OBP SLG OPS 1 Mark Teixeira TEX 0.316 0.393 0.588 0.981 2 David Dellucci TEX 0.294 0.397 0.578 0.976 3 Jason Giambi NYA 0.281 0.414 0.545 0.959 4 Travis Hafner CLE 0.293 0.398 0.559 0.956 5 Alex Rodriguez NYA 0.295 0.397 0.557 0.954 6 David Ortiz BOS 0.293 0.396 0.555 0.950 7 Manny Ramirez BOS 0.285 0.380 0.546 0.926 8 Mark DeRosa TEX 0.295 0.372 0.548 0.921 9 Kevin Mench TEX 0.314 0.377 0.533 0.910 10 Alfonso Soriano TEX 0.302 0.359 0.549 0.907 11 Richie Sexson SEA 0.275 0.380 0.526 0.906 12 Michael Young TEX 0.317 0.380 0.521 0.902 13 Richard Hidalgo TEX 0.289 0.358 0.538 0.896 14 Rod Barajas TEX 0.302 0.360 0.535 0.895 15 Hank Blalock TEX 0.302 0.363 0.526 0.889 16 Jhonny Peralta CLE 0.286 0.370 0.517 0.887 17 Paul Konerko CHA 0.270 0.361 0.520 0.881 18 Gary Matthews TEX 0.298 0.368 0.510 0.879 19 Adrian Gonzalez TEX 0.290 0.348 0.524 0.872 20 Jonny Gomes TBA 0.273 0.353 0.515 0.868