Stacey of CamdenChat is doing a feature now in which she highlights different sites from the O’s blogosphere, and she was nice enough to include CamdenCrazies in the series. She said some flattering things and then had me answer some tough questions. I included some excerpts below (me trying to explain WAR and UZR, plus some general O’s stuff for next year as well as my all-time favorite Orioles), but you should head over to read the full version, and bookmark CamdenChat while you’re at it.
“2. As a numbers guy, what is your go-to stat to quickly judge a player’s performance? If you had to rely on just one, what would it be?
If I had to rely on just one it would be Wins Above Replacement (WAR) from FanGraphs, since it gives you the best total picture of a player’s value quickly (though it’s not without flaws). They have a great primer on it. WAR is an end product of a lengthy process, in which other stats are used to build up towards a player’s total value.
For offense there’s weighted On-Base Average (wOBA), which assigns a run value weight to how many singles, doubles, etc. a hitter had and gives a rate stat on the OBP scale that encapsolates offensive production. To get an idea of how a player’s past production reflects on his true talent level, I tend to look at things like Batting Average On Balls In Play (BABIP) and Home Run per Fly Ball rate (HR/FB) to see if a guy might have been “lucky” or “unlucky”, walk and strike-out rates to get an idea of what kind of hitter he is (along with the standard slash stat BA/OBP/SLG).
Defensively, the first thing I’d look at is Ultimate Zone Rating (UZR) which gives a measure of how many runs above/below average a player was with the glove compared to others at his position (more on UZR in question #3). What position a player plays needs to be adjusted for as well, since being an average defensive shortstop is more valuable than being an average defensive first-baseman. WAR takes that into account as well. All that is compared to average, so then there’s another step to relate it to replacement level – you’re freely available guy that anyone can sign (like a Brandon Fahey) – which not only gives you an idea of how much a player contributes above what someone who would need to take his place could do but allows the conversion to dollars to set the replacement level (0 WAR) at the league minimum salary, which is handy. There are also baserunning stats (stealing, but also going first-to-third on a single and things like that), amongst other things, that aren’t all included in WAR but are good to add on.
Pitchers’ WAR is based off of innings pitched and Fielding Independent Pitching (FIP) which uses the things a pitcher is more directly in control of – strike-outs, walks, and home runs. Beyond that, there’s BABIP and HR/FB rate (for “luck”), as well as the very cool Pitch/FX data which gives you information on pitch velocity and movement, amongst other things. In general, I want a pitcher with a lot of K’s, few walks, and a high groundball rate (which keeps the homer runs down). Obvious stuff, really.
That was probably longer than you wanted, but it would really defeat the principles I try to use in what I do to only look at one stat. Though yeah, WAR is generally the way to go.
3. UZR is a stat that gets a lot of flak at Camden Chat, most notably because Nick Markakis was rated so poorly in 2009 and because Luke Scott has been rated positively in the past. Can you give us dummies the pros and cons of UZR in layman’s terms and the best way to use it in evaluating a player’s overall performance? How does it compare to the plus/minus used in the Fielding Bible in your opinion?
FanGraphs saves the day again. Not directly related to the question, but FanGraphs might very well be the best baseball site on the internet now.. Can’t recommend it enough. I wouldn’t nearly do the measure justice by trying to explain it briefly, but basically it breaks the field into a bunch of zones and sees how players do with balls (not just out/hit but weighting the hits by their run values like with wOBA) in each zone compared to other players of the same position, adjusting for how hard the ball was hit, GB/FB/LD, and other things
Pros: it’s the best freely available measure that gives an idea of how good a player is defensively at his position relative to average, converted into runs (which is the base unit across various statistical measures for different aspects of the game – that way we can see if a player’s glove is good enough to make up for a bad bat, or vice versa); and it separates out value from range, errors, arm rating for outfielders, and ability to turn the double play for infielders, which makes it easier to get an idea of a player’s skill set.
Cons: fielding is hard to measure, and so even though it’s best thing we’ve got it isn’t perfect; there is some error in the way it measures things (for example. it uses a limited number of zones, so things get lumped together that aren’t exactly the same); there is also the issue of sample size, in that a full season of fielding stats is not equivalent to a full season of offensive stats. That’s why it’s especially important to look at more than one year of data and regress to the mean. (I’m not as good about this as I should be, but I’m trying)..
Using Nick Markakis as an example, he had a +2.4 UZR in 2007 (that is, he was about 2.4 runs better than the average RF), a +12.1 in ’08, and a -5.8 in ’09. Instead of saying he was all over the place talent-wise – OK, great, and bad – it’s better to look at all three (weighting the more recent years more; say, 5-4-3) and then (since the it’s still just a sample) regress it to the mean (we have some information about how good Nick is defensively, but it’s more likely than not that his actual talent level is slightly worse that we’ve measured – and closer to average – than that it’s better) and say he’s a very slightly above average RF (maybe +2 UZR). So I guess if you want to talk about how good Markakis was defensively in a given year you can cite the UZR while keeping in mind the measurement error (so if it’s -5.8 in ’09 and we think Nick’s actually a +2 RF then it might be fair to say that he was below average defensively but probably not -5.8 runs below average; though technically, it’s possible that a player can have a -10 UZR but actually have had a good defensive season (+10) that was really poorly measured). I just go with the UZR as is when I’m talking about WAR since that’s what goes into it, but when making qualitative judgments I try to hedge back a bit. If you want to talk about how good Markakis is defensively you should really look at multiple years and regress and all that (though certainty is never guaranteed). Also never hurts to look at Tom Tango’s Fan Scouting Report to see how people view various players’ fielding abilities as a realism check.
Comparing UZR to +/-, both use the same principles (breaking things into zones and assigning values to things that happen in them). The way they establish the zones and do the valuation is slightly different, but they both work off of the same base data so they should be fairly comparable. UZR uses larger zones so those individual sample sizes are bigger, which is nice. It also adjusts for the GB/FB tendencies of the pitcher, since groundballs from a groundball pitcher and groundballs from a flyball pitcher – even if both are classified as “fast grounders” (or whatever the exact tag would be) – are slightly different in general (easier to field a GB from a GB pitcher, on average). So I go with UZR (especially given that it’s free). If you keep in mind the limitations of the stats then it doesn’t really matter which one you use.
I really hoped that helped. I’ve been reading about this stuff for a while, so I’ve gotten a little bit of intuition about this but I’m still not really good about being precise when talking about it. Remember that a stat only measures that which it measure (yaye tautologies!), but the way we use and interpret it can be good or bad. If you’re giving UZR flak, then that says more about you than it does about UZR. [This is like me saying win-loss record to value pitcher is bad, instead of saying that the way people use win-loss to value pitchers is bad. It's not the stat's fault that some people don't have a good grasp of what it's telling you.]
That was very, very long. And still not even close to long enough. If Field/FX really gets up and running then that will certainly help in measuring defense.
Part 2 – The Baltimore Orioles baseball club1. Who do you think will spend more time in the OF in 2010, Nolan Reimold or Felix Pie? Is that who you think SHOULD spend more time in the OF in 2010?
If Reimold comes back healthy from his surgery and nobody gets traded, then I think he will. If Luke Scott goes then Reimold may see DH more with Pie playing left. It does work to Felix’s advantage that he can play CF though. As for who should, I’d say that it depends on a lot of factors (sorry). If only one guy can play at a time, then I’d give Reimold more time in LF with Pie filling in for Jones in center and Markakis in right – so that they may end up with similar amounts of total time in the outfield. If the roster doesn’t really get upgraded then Reimold at DH with Luke Scott at first and Pie in left is probably the best alignment. I’ll have more to say about this closer to Spring Traning, and I’d be interested in seeing what others think of the playing time splits as part of my 2010 projections.
2. Which pitcher do you see making the biggest strides in 2010?
Matusz looked pretty good last year for if he just pitches that well for a full season it’ll be valuable. Bergesen may be able to get his groundball rate up a tick (which would be great), but his ERA is likely to look worse even if he pitches at about the same level (not enough K’s). David Hernandez was pretty awful, so if he just becomes just “bad” it would be a stride forward. In the end I guess I’ll go with Tillman – largely by default – since he wasn’t good last year but has the tools to be a quality starter. He doesn’t have to come close to reaching his ceiling to show big improvement. I could see him upping his strike-outs and lowering the walks, and hopefully he can get his home run problems under control.
3. Who is your all-time favorite Oriole?
Mike Mussina. Loved watching him pitch. Had an extensive arsenal and could throw any pitch in any count, including a unique-ish one (the knuckle curve); great control; that was really smart in general, and he seemed like a good guy. I was always convinced – even before really I started to look at stats – that he was really one of the best pitcher in baseball (behind Maddux and the Big Unit for sure is all I remember). I really hope he gets into the Hall of Fame one day, because he deserves it – he was only the 5th best starter of his era, but it was an era filled with some of the best of all-time (Clemens, Maddux, Johnson, Pedro). Later on, I also liked him from a contrarian angle since I thought he was underrated due to the “never winning 20 games” thing. When he won 20 in his last season I was really happy for him, and I thought it showed a lot of character for him to retire to be with his family even though he could have made millions of more dollars and tried to hang on to get to 300 wins. It might be crazy, but Moose’s 35 is the number I associate with the Orioles even more than Ripken’s #8.. Unlike a lot of O’s fans, I was never that mad at him for going to the Yankees. I’m not really sure why though. If there was one guy I could hang out with for a day – even more than some of the sabermetrically inclined players of today – it would be Mike Mussina.”