2007 Defense through PMR
DMZ · January 13, 2008 at 8:03 am · Filed Under Mariners
Leafing through the Probabilistic Model of Range numbers for 2007 (check out the whole list), some things that jumped out at me:
Sexson’s surprisingly not near the bottom at his position.
Raul Ibanez is. He sucks defensively. He’s ahead of Manny Ramirez, Chris Duncan, and Pat Burrell. He’s behind Barry Bonds.
So did Jose Guillen. 5th-worst in the majors. Have fun with that, KC.
Beltre does well.
Ichiro is the second-best in center.
Betancourt’s a little below average.
could Sexson really be as bad as we think he is?
Great info. The only playoff team worse then us was the Diamondbacks and they were right behind us. Plus they defied about every prediction stat this year so I don’t think that means much. 6 playoff teams are in the top 10 with only the Indians and Diamondbacks falling to 16th and 19th on the list.
We rid ourselves of Ibanez and Sexson for good defenders and hope Yuni stops making errors and we have a real defense with a chance at competing.
The best defensive 1st basemen, Doug Mientkiewicz, happens to be a free agent. Had a .789 OPS and only cost the Yanks 1.5 Mil last year. Looks like its worth a shot at a lefty 1B with great D. Unless we want to keep bad defense and a .694 OPS…
I really like PMR, because PMR is constructed the way that I think is the most logical way to evaluate defense. How many types of balls are hit into a given fielder’s area, how hard are those balls hit, how many did he field in comparison with an average defender, what is the value of the differences between the fielders performance and the performance of an average fielder.
That’s about as straightforward and direct as you can approach the issue.
#2
There does come a point where a good defender is only so valuable. Doug Mientkiewicz sucks at the plate and his defense, no matter how perfect he played, wouldn’t make up for his weak hitting at a power position.
Has the PMRs been released in a list form yet which allows comparison of individuals within a position? Last time I had checked there were only a few positions released that compared data that way.
I love Yuni’s fly ball graph. Its right along the predicted for most the graphs, then shows flys to left field and it spike way way above the predicted.
The “Run halfway to the wall to catch the popup” is classic Yuni.
Doug Mientkiewicz sucks at the plate and his defense, no matter how perfect he played, wouldn’t make up for his weak hitting at a power position.
Mientkiewicz had a 5.8 VORP last year, 6.2 in 2006, and 0.0 in 2005. If you figure he’s about 10 runs above average on defense (which seems pretty reasonable given the PMR data–and basically agrees with UZR), Mientkiewicz is about 10-15 runs above replacement overall, per year.
Sexson had a -7.1 VORP last year, 25 in 2006, and 45.8 in 2005. PMR has Sexson at about -5 runs this year, Dave put him at -10 to -20 in his season wrap up, and UZR has him at about -10 runs/year from ’03 to ’07. So I’ll say he’s at -10 runs/year–the low end of Dave’s range. That makes him about +60 on offense and -30 on defense over the last three years, or about 10 runs above replacement overall, per year.
Meanwhile, over the last three seasons, Mientkiewicz was paid $7M and Sexson was paid $39M.
Mientkiewicz’s defense can’t make up the difference between him and a hitter who can consistently contribute, say, 30 or 40 runs on offense, but Sexson hasn’t been that kind of hitter over the last couple of years, and it’s certainly questionable whether he will be next year.
According to MGL (here: http://mvn.com/mlb-stats/2007/10/19/did-doug-mientkiewicz-get-joe-torre-fired/#comment-1166), Mientkiewicz is +13 runs in the field, about one win below the average 1B as a hitter, and certainly deserving of a full-time job.
And if you’re still hankerin’ for more defensive metrics, check out the 2007 total zone ratings, by the erudite Sean Smith. They’re explained here and the ratings themselves are here.
That’s an excel file, fwiw.
Basically, they show the same things PMR does –
Sexson’s bad, but only -5; Ibanez sucks, Jose Lopez is actually quite good. Other tidbits of note: Yuni Betancourt scores out almost average, which is much better than he does in some other systems. Oh, and this agrees a bit more with UZR that Ichiro isn’t very good (I would dispute this). TZ also has Jose Guillen as being basically Ibanez’ equal in terms of defensive futility. That might help explain the M’s poor outfield DER, and it’s an explanation that I find much more palatable than, “ichiro sucks,” which are fighting words where I come from.
I agree with most of this, but the problem I have is that I can’t use a system where there’s something I so disagree with (Ichiro) that I throw out those specific results, but not others. Unless there’s a systemic reason for throwing out a result. Is there a good explanation for why Ichiro would come out so badly?
TZ ratings puts Beltre as a – player, which I just don’t believe.
I don’t see any reason to think that defensive metrics are worth much of anything at the moment, and that FSR is probably a more accurate system.
Well, multiple methods tend to focus in on what you’re really after…looking at the different measuring sticks shoul give you a decent idea of what’s happenng…
11 – they put him right at average. I hear ya, I don’t agree with that either, but I tend to use Dave’s confidence intervals around the specific number, so in this case a -1 is more like from a few runs below to a few runs above ave. I tend to think it’s at least a few runs above, but that one doesn’t stick out as impossible like the Ichiro figure.
I like FSR, but I like to use it in conjunction with other data – to counter the Jeter Effect. I’d also like to see some Y-T-Y correlations on FSR, UZR, RZR, PMR, etc. I just want to see how volatile a lot of these numbers are.
10 – i keep asking about that in every defensive metric post. Basically, the only thing I can think of is that the play-by-play data is screwy somehow. That’s why Ichiro is awesome in one PBP metric and awful in another – the PBP data is supplied by two different companies. Now, I don’t know why UZR/TZ could get you reasonable results for most others, but it’s worth pointing out that UZR has not generally been impressed by Ichiro, even in RF.
Well, I’ve been following the evolution of defensive metrics as long as I can remember, and here’s my quick take on this: defensive metrics are complicated, imperfect, and any single season, much less metric, shouldn’t be relied on too heavily.
To form an opinion of a player’s defense, look at every good metric you can get your hands on, and feel free to discard ones that are way out of line for a particular player — if you see Joe Outfielder show up as league-worst in every metric but one, where he’s average, you’re entirely justified thinking he sucks.
But you’re really better off when you can look at two, three years of data across the different metrics, and then you can really start to make judgments about their contributions.
DMZ – That makes complete sense to me. It’s just way more work than I’m willing to go. That’s why I appreciate this site so much, where other people do take the time to go to all the work.
[if you know the spelling’s wrong, don’t inflict it on us]
How much of Yuni’s miserable first half tarnished his overall numbers? I noticed that while he was a real butterfingers last year, he only had 2 more errors then 2006 overall. But the sheer volume of them in before July in ’07 was disappointing to say the least.
16
Or it could have the opposite effect and re-enforce the belief that vetern leadership really is worth it. You dont get much more gritty or veteran than Favre.
I don’t actually believe Brett Favre’s name wasis spelled “Falbvrets?” I simply included every consonant I could think of that could be silent in a French word. My apologies if I left any out. I know it wasn’t the most wildly insightful post ever, but why the hate? Is it the implied slam on “eggheads?” I work in a room full of some of the biggest eggheads in the world every day. I AM and eggead! Post revised below with CORRECT spelling.
I hope Bavasi watched the Seahawks game yesterday. If he refuses to acknowledge abstruce defensive metrics dreamt up by rooms full of eggheads…maybe watching Brett Favre march down the field like a schoolgirl skipping through a field of daisies will make him appreciate the value of defense.
“Abstruce”? “Wasis”? “I AM and eggead”?
I’m not going to edit your post, but even your ostensibly CORRECT post reads like someone talking with a mouth full of mashed potatoes.
And to be clear, that isn’t a good thing.
Generally speaking, we’ve traditionally have very low tolerance for use of the (sp?).
It’s super annoying. (jk) is like that too.
7- Mat made my counter argument for me while I was watching some good upset football.
Mientkiewicz is not ideal, but he is much cheaper and has a better chance to help this team then Sexson. That being said I would only take him on if we could get someone to take Sexson and some or all of his salary.
Maybe I’m blind, but how do/did you compare a player to others at his position? Did you just click on every LF and add up how many outs below average they were for all types of balls. It must be easier than that.
Just wanted to say great website! Kinda new around here and I love all the great information and stats. However, I do not think everything in baseball can be looked at using ONLY numbers. Any system that has Ichrio and Beltre as below average fielders has to be seriously flawed,as both of them are clearly gold glovers.
As a fan of this great game I’ve come to realize there are way too many variables to take into account to get an accurate prediction of a players level of play. Some stats are great and help you paint a picture of a certain player, but others don’t do anything but give you wrong information. Don’t forget to use the most important tool to determine players abilites your own eyes.
[ot]
24 – Don’t forget to use the most important tool to determine players abilites your own eyes.
The problem I have with that is that I only have regular access to the Mariners. I watch all the games on TV (I’m in Oregon), so I see almost their whole season, but not much of anyone else (maybe the hated Yankees on ESPN). I’d like to know how Yuni stacks up to shortstops on other teams. I enjoy watching him, but I know part of it is that he just looks like he’s having so damn much fun out there. And what about Lopez? Is he as bad as the casual fan is being increasingly led to believe? I don’t think so, but I don’t get to watch all the other 2BB in the majors to make a comparison.
A big thing for me is that I think having fans vote for All Stars is stupid, but that’s the system. How am I supposed to know how to rank players that I don’t get to see much or at all? (How are players and coaches supposed to do that as well?)
Then of course, there’s my beloved Dodgers, with a bunch of kids playing that are going to keep getting better. I hardly ever get to see them play anymore, and even if I did, I wouldn’t know how they were stacking up against the rest of the league.
For me, the strength of this site is that it’s populated with people (especially the authors, but there are several more) who keep up on the new ways of using numbers in baseball, but they remain skeptical until something is shown to have value. AND, at the same time, they watch a hell of a lot of games. You’ll have to be on the site during a game sometime.
I know you didn’t completely dis stats, but I do think if you know what you’re doing, you can find a universe of numbers that gives you a pretty good way to compare people. As for prediction, the best you can do in any field is figure out what’s most likely to happen. That’s the whole point of trying to quantify variables. It might not matter to a fan, but it sure should matter to front offices. If they can’t figure out what level a player should be at, they can really get fleeced in trades… Or in deciding what pitcher to put in…
Sorry for the rant.
24
I don’t think people here are trying to turn baseball into binary, behappy. The value of defensive metrics is to quantify defensive contribution and get beyond the ‘great’ or ‘good’ or ‘terrible’ tags that can be applied by your recommende method. If you want to say that a player’s great offensive production is offset by his horrible defense it is not enough to rely on your eyes and say he’s terrible – you need something that can counter batting average and RBI (or VORP or whatever).
As for what you suggest, the FSR mentioned earlier in the thread refers to the fans scouting report, a project run by Tom Tango to compile defensive rankings of all players based on what the fans think. Not quantitative in a runs over average sort of way, but a ranking nevertheless. The problem with it is Ichiro rated as the greatest defensive player in baseball at any position, which he is surely not.
I’m often astounded that so many of the baseball metrics aimed at individuals do so well.
Compare baseball to the stock market, for example, which is, virtually by definition, the sum or multiplicand of millions of numbers.
How many people successfully predict the movement the DOW Jones over the course of a year? By predict, in this case, I don’t mean day-to-day but overall predictions for the entire year. Answer: not so many.
How many people successfully predict the movements of individual stocks over the course of a year? Answer: less.
Baseball analysis has no such limitless universe of raw numbers to choose among, but baseball analysts for my money do a better job than stock analysts.
By subtracting Guillen and adding Jones we’re looking at a significant boost to our outfield defense. This should allow Ichiro to shade over into left field and negate some of Ibanez’s atrociousness, shouldn’t it?
If Jones doesn’t get traded and replaced by a signing of Guillen light.
Baseball analysis is far easier than stock analysis.
There are thousands of public companies, and they each employ hundreds or thousands of people, many of which can have a direct or indirect effect on the stock price — and we know virtually nothing about them. Moreover, there are millions of events every day, all over the world, any one of which has some effect on the stock of some company. Fire in a factory in China? Somebody isn’t getting circuit boards, which means some product isn’t ready for the holiday shopping season, but somebody else’s is. Hurricane in the Gulf of Mexico? Refineries shut down, oil spikes, the airlines get hurt, but it’s a boost to builders with presence in the south. Heck, the chair of the Fed just says a magic phrase at luncheon and the entire market can swing hundreds of points. To predict the market with accuracy you have to predict the world.
Meanwhile in baseball you’re dealing with 750 players (give or take), who play in at most 162 games in a season. They’ve all been through essentially the same filtering process to get to the majors, and they’re all subject to the laws of physics and the limits of the human body (even when, uh, science is giving them a boost). The difference between the best and worst, and between either of those and average, isn’t all that great (what was it Crash Davis said — one hit a week?). There’s only, what, a factor of two between the best batting average and the worst? There’s even less between the best and worst ERAs (given a reasonable number of starts). Players that are really bad aren’t allowed to play (well, unless they’re Richie Sexson or Jeff Weaver). Players can play hurt (like Raul most last year) and they can suddenly get the handle on a new pitch or get into a groove or grow into themselves or whatever, but even then the “surprise” in the numbers is relatively small. To predict a player you don’t have to predict the world, or even the team; you just have to look at what he’s done before and factor in growth and aging within some margin of error. It’s hard to predict wins because there’s more confounding factors, but even then a lot of baseball analysts get within 10% — and that’s still much better than financial analysts and the stock market, because even then the confounding factors are comparatively limited.
I’m not surprised that baseball analysts do a better job than stock analysts. Fortunately, you don’t have to be right all the time to make money.
joser, I agree, except that there’s the occasional 163 game season. Which Colorado was quite happy about.
We all know that defense is pretty underrated in terms of the general public (and most franchises).
I think we need a post called something like “Pitching wins championships and 9 other common misconceptions.”
And quick stats/reasons on why.
I think we need a post called something like “Pitching wins championships and 9 other common misconceptions.â€
And quick stats/reasons on why.
Well, there’s a great book called ‘Baseball Between the Numbers’ put out by the guys at Baseball Prospectus that deals with a lot of that stuff.
Yeah, the Baseball Prospectus book is a good read. Anything that quanitfiably destroys the myths of Jeter gets a thumbs up from me.
[ot]
lailaihei – Head over to LookoutLanding.com and check out Jeff’s “Dispelling Myths†1-3. They’re awesome!
note from Drayer about the “Adam Jones called back” rumor
[ot]
[ot]
It’s time for the Mariners and Reds to submit simultaneous “FU†emails to MacPhail.
41- at least the Ms…that would be nice…
Does is it seem odd to anyone else that the Reds, in the heat of their Bedard negotiations, would hire Walt Jocketty? After all, Jocketty’s most famous trade in St. Louis was his Farm System-for-Mulder deal, a colossal train wreck eerily similar to the Reds potential Farm System-for-Bedard deal.
I wonder if the Reds asked Jocketty, “What the hell were you thinking? Because if we are thinking the same thing now you were thinking then, we will pull the plug on this bad boy, pronto.â€
“After all, Jocketty’s most famous trade in St. Louis was his Farm System-for-Mulder deal”
I’d guess some St. Louis fans might think his acquisition of Mark McGwire was a bigger deal…
31. Baseball analysis is far easier than stock analysis.
Correct, but in addition to the reasons that you mentioned, there’s another even more important one: stock markets are fairly close to being “efficient”. Without going into the Efficient Markets Hypothesis, the bottom line is that markets are SUPPOSED TO BE unpredictable (when they are working efficiently).
Because if people COULD predict which stock was going to “win the World Series” (i.e. be a big bargain that was going to shoot up in value in 2008) they’d buy it right now. And it’s price would rise, no longer making it a bargain.
That automatic feedback is what makes stock prices hard to predict: any predictive information that we have (e.g. Apple’s got good design, Toyota’s got a lead in hybrid cars, etc.) gets incorporated by the market and into the stock’s price. There is good predictive info out there, but if the market’s working right, it’s already been taken into account.
In contrast, if we have good predictive info that the Royals will likely do worse than the Red Sox in 2008, there is no market which causes the Royal’s “price” to fall and the Red Sox’s to rise, i.e. no mechanism which reacts to the info by giving the Royals an easier schedule and the Red Sox a harder one. So our predictive info retains it predictive power.
The problem with talking about how Mientkiewicz is a better player and better value than Sexson is that it’s not really an argument about Mientkiewicz or how teams with playoff ambitions could use Mientkiewicz. It’s about how farcical it is that the Mariners think that they are going to compete for a playoff spot while paying Sexson $15 to put up sub-replacement numbers at a prime offensive position.