Thursday, August 06, 2009

My Intern loves Stats, too

Working at America's Favorite Radio Station, we have the occasion to experience quite a few interns that come and go through our station on their ways to careers of their own. Well, under normal circumstances, I would never reprint one of their essays that they bounce off me. But, TC Fleming is not a normal intern (in my estimation). He is a young, promising sports dork. And as an older sports dork, I should know.

Anyway, he wrote an essay this week that I wanted to share with you. It is all about the strides the Rangers havemade this season in the defensive department. I am not going to lie to you, this might be a bit deeper than some would like to go - given the enormous amount of Bill James-type statistics he uses, but if you enjoy that sort of data analysis, then you very much might enjoy this.

TC the intern looks at why the Rangers are better in 2009. It is all about the gloves:



These Rangers have been able to finally cash in the promise past teams never delivered on because the organization has finally stopped focusing entirely on bats and started to give credence to pitching and defense. I get the impression that most folks hold that sentiment about the 2009 Rangers (I get this impression mainly from Norm’s call with Dwight in North Dallas). This view of the situation is only half right: the Rangers have made astounding leaps in their defense, some really incredible stuff. However, the pitching—especially the starting pitching—is not terribly different from last year. They’re better, but only by a little. In the rush to find a nice, clean reason for the fact the Rangers have a lot more wins than last year, I don’t feel like there’s being an adequate separation being made between the contributions of the pitching staff and the contribution of the team’s gloves. I would like to use this forum to demonstrate such a separation.

Now, if you’re the sort of fella who likes baseball but doesn’t have time for too many numbers, then just trust me on this one: just about nothing has changed with the pitching staff. The pitchers are pitching the same. It’s just that the guys behind those pitchers are covering for their hurlers much better than in years past. The Rangers have a defense that is top five in the major leagues (closer to five than one, but still, top five).

But if you are the sort of fella who likes baseball and can tolerate a little stat analysis, then I think you might like these next five minutes of your life.

And if you’re not the sort of fella who likes baseball, you must be pretty frustrated with this blog.

In 2008, the Rangers allowed 5.96 runs per game. In 2009, the Rangers have allowed 4.41 runs per game. The two elements to run prevention are pitching and defense (and luck, but I’ll throw in a word about that at the end). So if we show, as I’m about to, that the pitching is pretty damn similar to last year, then the defense must hold the majority of responsibility for that difference in runs allowed. That’s not all my “the defense is so awesome now” is resting on, but that’s a good bit of it.

It’s been demonstrated with math and stuff that a pitcher has real control over three things: getting strikeouts, preventing walks, and keeping the ball on the ground when contact is made. These three things can be wrapped up in one nice little number: Fielding-Independent Pitching (FIP). It’s on the same scale as ERA, but it only takes into account strikeouts, walks and home runs allowed (if you’re keeping the ball on the ground, it’s not going to go out of the park). In 2008, the Rangers had an FIP of 5.37, good for 28th in the major leagues. This year, the Rangers have had an FIP of 4.58. You would definitely take that number over 5.37, but they’re still 24th in the league. Taking home runs out of the picture since the Ballpark has quite an impact on that, the Rangers ranked 28th last year in strikeout-to-walk ratio and have improved on that as well, but again only to 24th. Those numbers and many others show that the Rangers had one of the three worst pitching staffs in the majors in 2008, and while they have universally improved in pitching categories, it’s still one of the five or so worst staffs in the majors. Those jumps don’t even begin to explain how a team would allow more than a full run less per game.

What does explain the jump, however, is the defense. There are a number of categories one looks at to see if factors a pitcher has no control over are making him look better than he really is. A lot of the time, those factors are the result of luck. When an entire pitching staff all has ‘luck’ in all the same categories, you should probably look for a better explanation.

Two such categories are Batting Average on Balls In Play (BABIP) and Percentage of Runners Left on Base (LOB %). BABIP is just looking at of the times batters make contact, how many times does that contact end up in a hit. When watching baseball, we all see times that a ball is hit hard, certainly hard enough to turn into a hit in most cases, but happens to go right into someone’s glove. The pitcher has made a mistake in those situations, allowing hard contact, but he was not punished for it for reasons outside his own accomplishments. So BABIP is a way of correcting for that. But when you look past one pitcher and look at a whole staff, BABIP becomes a way of approximating the contributions of the fielders: are they just getting to so many balls and making the right throws and such that they can be depended upon to prevent contact from turning into hits on a game-to-game basis? A low BABIP (usually defined as one less than .300) is luck on the part of the pitcher, but it can still be an accomplishment of the fielders behind him.

The other number, LOB%, is similar. Say one pitcher allows three singles in two innings, and they happen to occur in such a way as to allow none of the runners to score, and another pitcher allows three singles in two innings, but they occur in such a way that two runs score. The biggest difference between these two performances is chance. Yes, the first pitcher might have done a better job of pitching from the stretch, but still: the majority of the single factor most responsible for one guy succeeding and another failing has nothing to do with their own performance, it’s chance. LOB% accounts for that. It’s a way of checking if someone has had abnormally good luck at having the runners he allows not coming around to score. Some of this has to do with pitching well: if you allow a runner, but are the sort of dominant guy that can just strike out the next three to strand that runner, bully for you. But the biggest factor in a single pitcher having a low or high LOB% is luck. However, it’s again different if an entire pitching staff has a high LOB%. Then, those runners might be getting stranded because the defense is so good at turning everything into an out. Suddenly it’s hard for runners to come around to score because contact that would normally be good enough to move a runner over is now being caught, or a fly ball that would normally be a sac fly isn’t because a runner respects the outfielder’s arm. These things pile up, and thanks to that, a team’s LOB% can make some commentary on a team’s fielding ability.

A short aside: the performance of just about every pitcher in the starting rotation has been aided by these numbers. Just about everyone is looking better than they’re pitching. But there is one very exciting exception: Derek Holland. Of the seven pitchers who have spent appreciable time in the starting rotation, Holland is one of only two to have a BABIP over .300 (.332). Further, he has a homerun-to-flyball ratio of 14.6%. That’s plain bad luck; he can expect that to fall to around 11% with time. Once it does, he will havea 4.21 FIP. That figure won’t win any Cy Young’s, but it would be the best number in the rotation. Again, for emphasis: Derek Holland is pitching better than anyone in the starting rotation. And he’s 22. If you want reason to hope that the pitching might one day undergo the sort of turnaround that I’m about to illustrate the defense has, there’s your source of hope. And what’s more is that Holland still has clear room to grow: he is currently striking out about seven and a half batters per nine. In his minor league career above AA, he was striking out 10.2 batters per nine, so you can feel a good bit of confidence that he’ll start improving on those K’s pretty soon, though he’s already pretty good at it. Once you get past the unlucky BABIP and HR/FB, you have to be really excited about Holland. We’ve heard he has ace potential. That’s not a lock based on these numbers, but it’s clearly within the realm of possibility.

Both of these numbers, however, give more of a rough outline of defense rather than a clear picture. One number that tries to give such a picture is Ultimate Zone Rating (UZR). Despite the name, it looks at all aspects of a fielder’s defense and puts a number, measured in runs, of how much that fielder’s defense is worth. Sports Illustrated wrote a very nice article about it before the season explaining everything in full.

Another little spice to add to this stew is that of FIP-ERA. That’s just what it looks like: the difference one gets when subtracting the FIP from the ERA. If FIP is a measure of what a pitching staff is doing, and ERA is the actual number of earned runs that occurred, it stands to reason that the difference between the two is a nice mix of the defense’s contribution and pure chance.

Using these four things—BABIP, LD%, UZR, and FIP-ERA—together can give us a real nice idea of just how much the Rangers fielding has improved. In 2008, the Rangers ranked dead last in every single one of those categories. While each of these elements separately can be doubted, when taken together, there can be no doubt this was the worst defense in the major leagues, and it might not have been close. This is probably a separate post, but the 2008 Rangers may well have been historical in how terrible they were. The single most damning thing was a -51.7 UZR. Roughly speaking 10 runs is worth about one win. The Rangers gave up five wins that they would have received had they been merely average as a defensive team. A good many division races are decided by fewer than five wins. Truly remarkable stuff.

2009, however, is equally remarkable for the tremendous improvements made. The Rangers rank 4th in BABIP at .288, 4th in LOB% at 74%, 3rd in FIP-ERA at -0.40, and—here’s the real crown jewel—5th in UZR with 24.7 runs. So with more than 50 games remaining in the season, the Rangers have already gained another seven (!!!) wins going by UZR. What was the worst defense in the MLB is now cemented a place in the top five. That kind of turnaround is going to bring you a lot of success.

This defensive improvement is really pretty magical, and I think that warrants a closer look at some of the specific personnel upgrades the Rangers made. The largest improvement is probably the most obvious: Elvis. I know we all like Michael Young, me included, but UZR does not rate him highly at all as a defensive shortstop. He was five runs below average as a defender in 2008, and it was actually one of his better years. Compare that to Andrus, who has the third highest UZR among shortstops at 11 despite being the youngest regular in the majors. You gotta feel good about him as something to build around defensively.

Many observers have credited Ian Kinsler as improving his gloveworkthis year compared to his first few in the majors. UZR supports this: though he is about seven runs below average for his career, he is fiveand a half runs above average thus far in 2009. The way UZR works, it takes a lot of innings for it to mean much. It’s possible to have a year-long fluke. So his career average is probably a better indicator of what he’s going to do going forward, but that shouldn’t detract from the nice work he’s done so far.

Michael Young’s defensive limitations haven’t really gone away at third base. It’s better than when he was a shortstop, but it’s still 10+ runs below average. The big limiting factor is his range, which makes sense. Range is the hardest thing to determine with the naked eye. He’s really good at handling all the balls he gets to, and we see that and think of him well. Errors are not a problem for him. What you don’t see is him failing to get to a ball that an average fielder would have been able to reach, and that’s what really hurts him. Having said that, he is still an improvement over the mess of terrible the Rangers ran out to third base last year. Four fielders logged more than 200 innings at third for the 2008 Rangers (Blalock, Chris Davis, Ramon Vazquez, and German Duran), and all were as bad as Young or worse. Sometimes much worse. So bad as he is relativeto the rest of the league, he’s an improvement relative to his teammates.

The final notable improvement is probably going to be the most surprising to the Rangers viewer: Nelson Cruz. You might not know this watching him, but UZR rates him as the best right fielder in the majors at 14.5 runs above average. That’s a bit up from his career average of 11.5, but not insanely so. He really is this good. I know, I was shocked, too. His biggest asset is what we talked about for Young’s liability: range. He might look like he takes funny tacks to the ball, and he might look like a puppy in a thunderstorm settling under a pop-up, but the system indicates he does end up getting to more balls than your average right fielder. Baseball isn’t just about the things you can see from watching it; there’s a lot of stuff guys do to help their teams that you just won’t catch withthe naked eye. The difference between an All-Star hitting .300 and another spare dude hitting .275 is something like four hits a month. Nelson, for example, is hitting .272 so far, and if he had 3 more hits in each monthhe’s played so far, he’d be hitting .308. Now think of how many games you watch in a month, don’t you think he could haveslipped three hits past you without noticing? So without a statistical record such as batting average, you wouldn’t know if he was another guy going along at .272 or a star whizzing to a .308 batting average (Not to say Cruz is just going along, but I think we can agree his offensive value is tied more to his power than his average, but whatever, hopefully you get the point). UZR is like that: it tracks things that help a team that you might not be able to pick up from watching games, even if you watch a ton of them. And one of those things is that Nelson Cruz is a damn good right fielder. Further, when you look at Cruz’s work on the basepaths, you start to think he might be kind of sneaky fast for his size. That would certainly help the range. It makes some kind of sense. Sort of. Now, the right fielders in 2008 weren’t so bad, but no one guy had more than 200 innings (about 20 games worth) at the position. So if Nelson hasn’t done anything else, at least he’s stabilized the position, something that still can’t be said of the other two outfield positions that still largely rotate among Byrd, Murphy, Hamilton, and others.

Finally, Byrd is a better defender in center than Hamilton, so if anything nice came out of Josh’s injury, it was getting Byrd more time in center. Josh is another guy that has problems with range. Josh isn’t terrible, an Byrd isn’t much more than average, but Byrd is definitely better, and his extended innings in center are good for the Rangers’ defense.

So when you get right into the changes made in the Rangers defense, it makes some sense. Installing Elvis and shifting Young was an immediate upgrade at two key spots, Ian is still young and might be getting better or might just be a little lucky, and Nelson Cruz is a sleeper candidate for a gold glove if the people who decided such things ever considered stuff like UZR, which they don’t because they’re old and close-minded.

One thing to keep in mind is that stuff like BABIP, LOB%, and FIP-ERA is only partly defensively-related. The other element is luck. There’s always the chance that the team regresses next year and we see wins more in the low 80’s, rather than the 90-win range this team seems headed for. But then again, you can usually say that about 90-win teams. Stuff just has to go your way to win 90 games.

And one final note: the delightful book Baseball Behind The Numbers (doesn’t that just sound like a book I would like) had a chapter determining if there was any “secret sauce” to playoff baseball, if there was any facets of the game that translated particularly well to October baseball. They found that the quality of one’s closer and one’s defense matter a little more in the postseason than everything else. The logical premise supporting the numbers is that everyone who can get to the playoffs can score runs and have decent starting pitchers, that’s a given or they wouldn’t be in the playoffs. The teams that can prevent runs with their gloves have something not everyone else does. And the teams are evenly matched so the games are tighter, thus bringing the closers to the forefront. The Rangers havehad a bit of instability at closer, so they don’t gain points there relative to teams that can throw out a Papelbon or a Rivera every time they havethings close in the ninth, but the Rangers’ ability to defend puts them in the mix. If they can overcome the Angels or grab the wild card, I think the Rangers defensive improvement makes them a threat to do a little more damage than you might be expecting.

-- TC Fleming

PS: The SI article about UZR that is referenced is Here

No comments: