Archive for the ‘playing with stats’ Category

Thing Fifteen: Solving the Twins’ Outfield

April 28, 2009

To this point the blog has, if nothing else, justified its name, with this being the fifteenth new thing in fifteen days. And yet, aside from the occasional cheap shot at Alexi Casilla or Delmon Young, I’ve completely avoided talking about my own favorite team. The main reason for that is that my goal is to write one relatively succinct, digestible thing per day, and as I’m sure you’ve seen, I’ve struggled with that a few times already; if I start writing about the Twins, odds are I’m going to just prattle on forever. But I’m afraid that’s a chance I’m going to have to take today. It’s just time.

The general thinking is that five outfield/DH types — Young, Denard Span, Carlos Gomez, Michael Cuddyer and Jason Kubel — are all good enough to be playing every day somewhere, but only four spots are open to them. So the question coming into the year was: who’s the odd man out?

Well, so far, Gardy has done his best to answer that with: “well, nobody! Or everybody, depending on how you look at it!” Through the first 20 games, he’s started the following combinations (left-center-right):

Young – Span – Cuddyer: seven times
Span – Gomez – Cuddyer: six times
Young – Gomez – Span: four times
Young – Gomez – Cuddyer: two times
Kubel – Span – Cuddyer: one time

All told, Span has started six in left, eight in center, and four in right; Gomez has started 12 games, all in center; Young has started 12 games in left and one at DH; and Cuddyer has started 15 in right and two at DH. Kubel has essentially been the full-time DH, starting against both righties and lefties, though two others have spelled him there in addition to Cuddyer and Young.

Let’s take a look at who these guys are. Two career numbers for each player are given below; the first is wOBA, a system that’s about as good as any for assigning one number to the offensive value of a player, and it works on essentially the same scale as OBP (.300 is bad, .340 fine, and .400 great); the second is UZR/150, which attempts to measure how many runs a player saves or costs his team per 150 games played against the average at his primary outfield position, relying on play-by-play data.

Michal Cuddyer (.339, -6.3): the elder statesman of this group (but still a week or so younger than me), Cuddyer had an excellent year with the bat in 2006 (.282/.362/.504, 24 HR, .370 wOBA), but slipped in 2007 and was hurt for most of 2008, and is off to a terribly slow start in 2009. He has a reputation among Twins fans as an excellent outfielder, but fans often confuse excellent arms with excellent outfielders; Cuddy has a cannon, but doesn’t get around well at all. His defensive numbers through his first 15 starts this year are bizarrely good (26.4 UZR/150), but his real ability tops out at about a minus-five-run right fielder. He hits righties well enough to justify playing every day for most clubs, but his real talent is hitting lefties, against whom he’s a career .280/.368/.439 hitter.

Carlos Gomez (.287, 18.7): Just 23 years old, Go-Go can be both a delight and absolute torture to watch. He swings from his heels (often falling to his knees off a particularly ambitious miss), never walks, is prone to mistakes on the bases, and, in 2008, would often bunt (often foul) with two strikes. But he might be the fastest player in baseball, and he absolutely is the best defensive centerfielder in baseball. As such, he needs only to get on base about 30% of the time, as he did in 2008, to be a useful everyday player. With his youth and talent (and he has a very nice swing on the rare occasion that he keeps it within reason), he still has the potential for much more than mere usefulness.

Jason Kubel (.338, -20.0): He’s a better hitter than his career wOBA suggests; that’s brought down by a poor first year back from surgery in 2006. He had a .345 wOBA last year and is tearing the cover off the ball in the early going this year, at .417. A typical lefty, Kubel has a career OPS 120 points higher against righties than against southpaws. With his reconstructed knee, he moves like he’s about eighty. A team without Justin Morneau might try him at first base, but he has no business “running” around the outfield.

Denard Span (.364, 12.0*): a former first-round pick, Span had pretty much obtained “bust” status heading into 2008, and then suddenly exploded. With an excellent 2008 in both the minors and majors and a similar start to 2009, it seems safe to conclude that Span did suddenly become a player: great eye at the plate, good bat control, good instincts on the bases, some gap power. He can apparently hit left-handed pitching despite being a lefty himself. He’s not quite the centerfielder Gomez is, but he can more than hold his own out there, and is an incredible asset in either corner.
* The 12 UZR150 is a reasonable guess; he hasn’t played enough games at any one position to really trust the numbers. What’s clear is that he’s an excellent defensive player at any of the outfield positions.

Delmon Young (.321, -15.8): Bill James recently wrote that Young must be the worst percentage player in baseball, and at this point, frankly, you could almost take “percentage” out of that label. Young, like Gomez, is just 23, but unlike Gomez, he has shown few flashes of potential and no currently useful Major League skills. He’s hit around .290 in both of his two full seasons, was once considered the #1 prospect in baseball, and had 93 RBI in 2007. That’s enough to convince some people that he’s a useful or promising player. Watch him every day, though, and you see something different. In the field, Delmon looks uninterested at best, clueless at worst; he frequently misses routine plays and routinely makes even minimal challenges into adventures (or doubles, or triples). He hasn’t balanced that by showing any power, hitting a total of just 23 HR in 1220 AB in 2007-08, and he’s drawn just 52 unintentional walks (against 232 strikeouts) in that same period. Even his minor league stats are largely underwhelming. I tend to believe that any player the scouts loved as much as they once loved Delmon must have something going for him, and maybe Delmon will show that something someday. But right now he’s here, and here is very, very, very, very far from there.

So what should he be doing with these guys? I see a few things that should be just blindingly obvious:

  • Span should be starting somewhere every day. Not only is he the best overall player of these five, which he clearly is because of his defense; he might even be the best pure offensive player among them. Whatever else you do, if Span is healthy, pencil him in in the leadoff spot and one of the outfield positions. He’s already sat out two of the first 20 games, and that’s two too many.
  • Young should not be starting anywhere on a contending team. Look, I get the argument. He’s a promising player, or people consider him as such, and needs to be playing every day. But if this team intends to compete in the Central — and this year, every team in the Central figures to compete in the Central — Delmon has no place on’t. Let him start every day in Triple-A (where, it should be noted, he’s never exactly proved himself), coach him heavily on defense and pitch recognition, and hope you don’t have to call him up before he’s ready because of an injury.
  • If you’ve got a flyball pitcher in the game, Gomez has to be in the game too. The thing about the Twins’ five outfielders is that only two of them are good defensive outfielders. So if you’ve got a guy on the mound who gives up a lot of fly balls (and that’s most of the Twins’ rotation), your best chance to win is to have both Gomez and Span in the outfield, even if you take a hit on offense.
  • Kubel shouldn’t DH versus lefties. Even in 2008, his best offensive year, Kubel had just a .704 OPS against left-handed pitching, worse than the overall OPS of Nick Punto and about equal to Casilla’s. Unless Gardy has some reason to believe Kubel has completely come around in that area — and I really don’t think he does — that’s just not a designated hitter.

So here’s what I’d do (assuming demoting Delmon isn’t an option):

Against RHP: Span LF, Gomez CF, Cuddyer RF, Kubel DH
Against LHP: Span LF, Gomez CF, Young RF, Cuddyer DH

So yeah, first, I’d play Gomez every day, flyball pitcher or no. I really think his defense is just that good, and, like a lot of people do with Young, I want to see him play every day to see if his bat will come around. Moreover, Span blanketing left allows Gomez to shade toward right, minimizing the damage done by playing Cuddyer and/or Young, who can pretty much just guard the line.

Second, I’m never putting Young in left, where his numbers have been uniformly terrible (I’ve watched him miss a relatively easy foul fly against the Rays as I’ve been writing this). For some reason, his numbers from about a season’s worth of playing right field with the Rays are above average (6.0 UZR/150 in 163 career G). That might just be a blip (and probably is), but it might also be that he had to depend less on his range and more on his strong arm in RF than he does in LF. There seems to be less foul territory in right field in the Dome, and the fence is closer. At least by putting him there you’d be giving him a chance of being a useful player, rather than just watching him flail helplessly around in left every day (as I have to currently).

Third, a Kubel/Cuddyer DH platoon is actually an above-average DH, whereas the current Kubel/Kubel setup is a serious weakness against lefties, especially in a lineup where your two best hitters are lefties.

Is this really worth spending all this time thinking over? …Well, yes, by somebody (probably not by me, but what can you do?). A Span/Gomez LF-CF would save about 40 runs on defense over the course of a season compared to a Young/Span one, which makes about four wins. And you give a little bit of that back on offense, but honestly, until Delmon actually shows something, it’s not all that much (and then Gomez takes a little back again on the basepaths). Four extra wins in the 2009 American League Central could very well mean the playoffs. To Gardy’s credit, he knows what his best defensive outfield is, frequently subbing Gomez in for Young and shifting Span to left in the late innings of close games. Now someone needs to explain to him the kind of difference having that for nine innings could make.

All that said, if Joe Mauer doesn’t come back on May 1 and knock the ball all over the park for 130 games, it’s not going to matter. But they might as well put their best lineup out there until we know for sure…

Getting in His Head, Tangibly

April 23, 2009

On Monday night, in what I’ve determined looking back must have been the Pirates-Marlins broadcast, the local (almost certainly Marlins’) broadcasters were chatting about the importance of the leadoff man getting on base, and the commentator said something very much like this: “Also, of course, you can never know how much impact that is going to have on the pitcher, because he’s going to be distracted by the fast runner on first base, um, so you don’t know what kind of effect that’s going to have on his pitching.” The play by play guy agreed, calling it “one of those intangibles that is just so important on the baseball diamond.” That’s totally paraphrased just to give you an idea, but I do know he used the word “intangibles.”

Of course, if someone is talking to you about baseball and uses a form of the word “intangible,” the odds are very, very good that you’re in the process of being told a lie. Baseball people have this way of equating things that they do not know with things that cannot be known. And that’s silly.

The impact that a good baserunner being on base might have on a pitcher’s effectiveness at pitching to the next batter may be difficult to measure (and it is — very, very difficult, or at least very, very time consuming), but it’s as “intangible” as the chair I’m sitting in. Do pitchers pitch worse with Jose Reyes on first base than they would with Jose Molina on first base? Better with Carlos Delgado taking his lead than with Carlos Gomez? It’s a pretty simple question. The fact that nobody has spent hundreds of hours digging through game logs to find the answer yet (as far as I know) doesn’t mean that answer is “intangible”; it means the answer isn’t important or interesting enough to spend all that time on.

So here’s a little exercise; it’s not at all meant to definitively answer the question, but just to show that the issue can in fact be, um, tanged (real word! …but not at all the root of “intangible”). There’s no way I’m going to go out and figure out how individual pitchers performed with various individual runners on base. But how about this: if speed on the bases makes pitchers less effective, shouldn’t we expect guys who hit behind speedsters to do better with runners on first than batters who hit behind slow or average runners do with those base-cloggers on first?

Chase Utley is a good example. The Phillies’ most common 2008 lineup had him batting in front of Rollins and Victorino, both top ten in stolen bases. When one of them wasn’t in front of Utley, they were replaced by Jayson Werth (who stole 20 out of 21) or fleet-footed backups Taguchi and Bruntlett. We can assume that very nearly every time Utley came up with men on base, those men were very good baserunners. If Utley did better with guys on base than without, and the difference is more than the typical difference between hitting with the bases empty and occupied (and there is usually some difference regardless; pitchers just do better throwing out of a windup), that might start to suggest that a “speed guy” getting on base really does impact the pitcher’s effectiveness.

Of course, a pitcher has enough to worry about with Utley at the plate, and that’s too small of a sample size anyway. So I’m going to take a bunch of guys in the National League who are likely to have hit with a lot of speed guys on first base in 2008 (a pitcher might be just as scared of a good runner on second base, but counting that performance would be more likely to catch other runners than the ones I have in mind), and see if any patterns emerge, though of course the sample size will still be way too small to tell. Batting orders change more than you might realize — Victorino, for instance, usually hit behind Rollins, but also spent time hitting behind anti-Rollinses Pat Burrell and Ryan Howard; Castillo hit mostly behind Reyes but also behind Brian Schneider. So this isn’t perfect, or even especially meaningful. But I’m confident that, when each of these players came up with a guy on first in 2008, there was a better than 50% chance that that runner was a threat to steal 20 or more bases.

Baseline: The 2008 National League put up a .731 OPS with the bases empty, and jumped up to .761 with a runner on first, so your average hitter will increase his OPS by about 30 points with a runner on first. If the fast-runner-distraction effect is real and significant, these guys should do considerably better than that as a group.

Utley: Batted second or third in every game he started, behind at least one stolen base threat approximately 100% of the time.
– Bases empty: .886
– Runner on first: 1.014
– Difference: +128 points

Victorino: Batted second behind Jimmy Rollins 81 times; led off (so rarely came up with a runner on) 14 times; batted 5th or 6th 37 times.
– Bases empty: .819
– Runner on first: .797
– Difference:
-22 points

Jeremy Hermida: Batted second in 87 games, almost all behind Hanley Ramirez. Also batted third, sixth, seventh and eighth a handful of times each.
– Bases empty: .718
– Runner on first: .773
– Difference: +65 points

Luis Castillo: batted second behind Reyes in 58 starts; batted seventh or eighth about 20 times.
– Bases empty: .629
– Runner on first: .643
– Difference:
+14 points

Andre Ethier: batted second 80 times and fifth a bunch of times; hit behind Furcal, Pierre, Kemp or Martin about 80% of the time, but also occasionally behind Kent or Manny.
– Bases empty: .935
– Runner on first: .623
– Difference: -312 points

Ryan Church: hit directly behind Beltran or Reyes about 80% of the time; Delgado occasionally came between him and Beltran.
– Bases empty: .703
– Runner on first: .971
– Difference:
+268 points

Ryan Theriot: hit second 104 times, usually behind Soriano.
– Bases empty: .752
– Runner on first: .786
– Difference:
+34 points

J.J. Hardy: seems to have batted behind either Weeks or Hart in about 3/4 of his starts.
– Bases empty: .832
– Runner on first: 1.052
– Difference:
+220 points

Average [fake, unweighted average, just a basic add-and-divide of the above]: +49 points

So these eight guys, overall, did get a bigger bump from having a guy on first than the 30 points the league as a whole got. What does that tell us? Well, absolutely nothing. Take away Ethier, and it’s a huge difference; take away Hardy or Church, and it’s a smaller-than-average bump. But these guys as a whole could be benefitting from pesky baserunners getting on base in front of them. To figure out whether or not they are would take a much, much longer and more sophisticated study, one which neither you nor I have the time or inclination (or, in my case at least, skill) to get into.

But here’s the point: it could be figured out. And someday it probably will be, if it hasn’t been already in some study I’m not aware of. The day when people who are paid to write or talk about baseball generally stop referring to certain very tangible things as though they’re mystical and unknowable just because those people can’t be bothered to take the time to know (or even honestly think about) them will be a very happy day.

Fun with Small Sample Sizes

April 22, 2009
  1. The Yankees sit at 8-6, but are on pace to score 810 runs and allow 972. This would make their expected (Pythagorean) record about 66-96.
  2. Then again, if Chien-Ming Wang were allowed to make 30 starts at his current pace, he’d give up 230 runs (in just 60 innings). This would be a record since 1901, narrowly edging out Snake Wiltse’s 1902 effort (in 300 innings). The record since 1950 is Phil Niekro’s 166 in 1977 (in 330 innings).
  3. Miguel Cabrera (through Monday, prorated): .489/.538/.787, 635 AB, 149 R, 310 H, 54 HR, 162 RBI
  4. Carlos Quentin: 87 HR, 162 RBI, 150 R…12 2B, 0 3B
  5. Brian Giles is hitting .151/.211/.189 (through Monday) and is on pace for twelve runs scored, zero homers…and 87 RBI. That’s how you know RBI is an awesome and totally not at all context-dependent stat.
  6. Washington Nationals (through Monday): 27-135 (.167), 770 RS, 1040 RA, Pythagorean W/L: 57-105.
  7. Raul Ibanez: .383/.442/.830, 176 R, 68 HR, 149 RBI, 14SB/0CS, about four defensive runs saved. Which totally makes sense considering the following hilarious evidence (from Lookout Landing): 1, 2, 3, 4, etc. So, yeah…it’s a long season.