Luck Be a Lady
Two quick things before I get to the meat and potatoes:
1) I am very pleased to see Ozzie talking about limiting pitch counts for Sox starters. There's a big debate in the baseball world over whether teams should utilize strict pitch counts or use other measures (such as pitch speed, fatigue signs, etc.) to determine when to take a starter out. I actually think both sides are right. But in this case, with every Sox starter having pitched 200+ innings last year, four of them pitching through October, and two of them pitching in the World Baseball Classic, sticking to some early-season pitch limits isn't the worst idea. I don't doubt that the people who get paid full-time by the Sox to think about these things are already taking this into consideration, but it's still reassuring to hear Ozzie say it.
2) I think it's time to demote Boone Logan to AAA Charlotte, and for a couple reasons. First of all, he's pitched poorly in (very) limited action - despite what his shiny ERA may tell you - giving up three hits and four walks in less than four innings. Secondly, this is the first time in his career he's had to face advanced, major league hitters, and he'd be better served gaining experience while throwing more innings with the Knights. And lastly, if the Sox have a major weakness, it's their bullpen. So far, only McCarthy and Jenks have shown themselves to be dependable. For now, that's been ok, because the starters have been great (especially recently), and the offense has built up large leads. But eventually we're going to need good third, fourth, and fifth options out of the 'pen. Luckily, besides the hope that Thornton, Cotts, and/or Politte will step up, we have a good option at AAA: Javier Lopez. He pitched decently during spring training, and so far this season with Charlotte has surrendured only six hits and two walks in nine innings, while allowing zero runs and striking out eight. I think it's time to call him up, and send Logan down.
On to the main subject of this post: Sox starting pitchers and luck. If I asked you right now who's been the best Sox starter so far, not including today's game (because stats don't include that game yet), you'd either say Mark Buehrle or Jose Contreras, right? Well, yes, on the surface, that's true. However, there is evidence that in fact Javier Vazquez has been the best of the bunch so far, and that Buehrle and Contreras have been quite lucky. And if we considered who's been the worst so far, it would obviously come down to Freddy Garcia and Jon Garland. However, there is evidence that Garland has been extremely unlucky, so that his underlying performance has actually been better than that of Garcia, or at least close.
But what evidence? Well, there are two main categories of luck when it comes to pitchers. I will call them:
Sequential Luck and Locational Luck
The first, Sequential Luck, has to do with the order that hits and walks (and to a lesser degree, sacrifice flys, double plays, etc.) come in. For example, imagine two pitchers, A and B. Both throw a nine inning complete game, give up nine hits (all doubles), walk no batters, throw no wild pitches, and have flawless defense played behind them. And yet Pitcher A has an ERA of o.oo on the day, while Pitcher B has an ERA of 8.00 on the day. Why? Because Pitcher A gave up one double in every inning, and stranded the baserunner in every instance, while Pitcher B gave up all nine doubles in one inning, resulting in eight runs.
Now, you could argue that the difference in ERA is entirely Pitcher B's fault, because he wasn't able to settle down after giving up a double (or nine), as Pitcher A was able to do in every inning. And that's a valid point. (For the same reason I prefer to look at a pitcher's RA, or Runs Against, rather than his ERA; good pitchers pitch through the mistakes of their defense.) But at the same time, a good deal of the difference between Pitchers A and B has to do with luck. What if instead of nine doubles, Pitcher B only gives up two, but they happen to be in the same inning. His ERA for the day (1.00) would still be higher than that of Pitcher A (0.00), but his WHIP would be far lower (0.22 as compared to 1.00). Now let's say those trends continued for an entire season, Pitcher A giving up far more hits than Pitcher B, but Pitcher B's ERA being higher because of the order the hits came in. Any team with half a brain would rather invest in Pitcher B than Pitcher A, because his underlying performance was better, even if the results were not. In short, Pitcher B would be extremely unlucky.
There's a good way to measure Sequential Luck, called Component ERA (usually listed as ERC). A pitcher's ERC is what his ERA should be, given the number of hits and walks he gives up. Basically, ERC is WHIP calibrated to the ERA scale. Then you can compare a pitcher's ERC to his ERA: if the ERC is higher, he was lucky; if the ERA is higher, he was unlucky. It's also very easy to render a new stat - ERC% - which shows just how lucky or unlucky a given pitcher was. An ERC% is the number that you'd have to multiply a pitcher's ERA by to get his ERC. For example, the luckiest starting pitcher so far this season is Detroit's Mike Maroth. His ERA is 0.49, but his ERC is 2.96. His ERC% is therefore 6.04, meaning that his ERA is six times better than it "should" be given his peripheral stats. That's pretty much off the charts in terms of luck; the second luckiest pitcher so far, Brett Myers, only has an ERC% of 1.65. (It should be noted that Maroth's peripheral stats indicate that he should still have a very low ERA - 2.96; a high ERC% doesn't mean you've been a bad pitcher, only a lucky one. That same reasoning also works in reverse.)
On the other end of things, the starting pitcher with the lowest ERC% so far this season is Javier Vazquez. Wait a second - don't we know that guy? Yep, Vazquez's Sequential Luck has been absurdly terrible early this season, with an ERC% of 0.47. While his ERA is still an impressive 3.86, the number of hits and walks he's given up indicate that his ERA "should" be 1.83, more than two full runs lower. In fact, four out of the five Sox starters have been unlucky in this regard so far, with only Contreras (1.11 before today's game) registering an ERC% above the break-even score of 1.00.
Locational Luck should actually be termed Defensive/Locational Luck, because it has two distinct aspects to it, but I'd rather go with the simple name. Locational Luck measures what happens to a ball after it's been hit into play. Many studies have shown that, with the possible exception of knuckle-ballers, pitchers have very little control over what happens to non-home run balls hit into play. In fact, whether a ball hit into play falls in for a hit appears to be mostly a matter of luck, varying greatly from year to year even when all other performance metrics for a pitcher remain the same. And this shouldn't really surprise any observant baseball fan, because we've all seen the screaming line drive hit directly at the left fielder for an out, or the weak ground ball that sneaks past the second baseman for a hit. Baseball truly is a game of inches.
The easiest way to measure this is via a stat called (by Baseball Prospectus) Batting Average on Balls In Play (BABIP), also known (at ESPN) as Balls In Play Average (BIPA). League average for BABIP is usually around .300 (so far this season in the AL it's .303). Therefore, a pitcher with a BABIP of .250 is getting somewhat lucky, whereas a pitcher with a BABIP of .350 is experiencing just the opposite.
However, defense does play a role in BABIP. A team with a good or great defense will tend to lower their pitchers' BABIPs, making an adjustment necessary when comparing pitchers on different teams. However, for today we're just comparing Sox starting pitchers, who each pitch in front of a (nearly) identical defense, so that won't matter. Still, it's worth keeping in mind when considering BABIP; team defense is the second aspect of Locational Luck that I referred to two paragraphs above.
As with ERC, it is possible to translate BABIP into a stat that mirrors ERA; in this case, the new stat is the Defense Independent Pitching Statistic (DIPS), created by the brilliant Voros McCracken. And like ERC and ERC%, DIPS can become DIPS%, giving an easy indication of Locational Luck. Mike Maroth is also the early leader in Locational Luck, with a ridiculous DIPS% of 8.93. And the starting pitcher with the worst Locational Luck is Jon Garland, at 0.49. Wait, that guy sounds familiar too. Garland's BABIP is an atrocious .433 (ESPN has his BIPA at .400; I don't know why there's a difference, as I think it's the exact same stat, but it doesn't really matter for our purposes), meaning he's been giving up a hit more than 40% of the time a batter hits the ball in play against him. As for the DIPS% of the other Sox starters, Garcia and Vazquez are near the league average, but Contreras (2.53) and Buehrle (2.28) are two of the luckiest pitchers in baseball so far.
Here are the full numbers for the five Sox starters, through Saturday's game:
I've highlighted a few of the more out-of-line numbers. As you can see, two of our starters - Vazquez and Garland - have been quite unlucky. Vazquez has probably been our best starter so far. His DIPS score is far better than the others, and his ERC is only a shade behind Buehrle and Contreras. Garland's BABIP is just about guaranteed to drop, which should significantly lower his ERA.
On the other end of things, both Buehrle and Contreras have been quite lucky, and both in terms of BABIP. You'll notice I also listed each pitcher's strikeouts per nine innings. That's because DIPS% is less significant for a strikeout pitcher, and more significant for a contact pitcher. For instance, you'll notice that Vazquez's DIPS% is 0.70, even though his BABIP is a lucky .259; that's because he strikes out 7.71 batters per nine innings, meaning that he remains in control of more batters and does not leave their outcome to luck. I fully expect Contreras' K/9 rate to rise significantly (his career mark is 7.59), meaning that even though his BABIP will rise, it won't affect his DIPS, or his ERA, all that much. On the other hand, Buehrle has the lowest K/9 rate on the team, and his career rate of 5.37 is lower than everyone but Garland's; when his BABIP regresses to the mean, it's going to mean a lot more runs given up. That doesn't mean he's suddenly going to become a bad pitcher. Rather, it just means we can't expect him to finish the season with an ERA under 2.00.
Of course, it's still very early in the season, rendering much of this meaningless. Regardless, I think it's worth introducing these stats now, so that we can revisit them later when more data has piled up.
1) I am very pleased to see Ozzie talking about limiting pitch counts for Sox starters. There's a big debate in the baseball world over whether teams should utilize strict pitch counts or use other measures (such as pitch speed, fatigue signs, etc.) to determine when to take a starter out. I actually think both sides are right. But in this case, with every Sox starter having pitched 200+ innings last year, four of them pitching through October, and two of them pitching in the World Baseball Classic, sticking to some early-season pitch limits isn't the worst idea. I don't doubt that the people who get paid full-time by the Sox to think about these things are already taking this into consideration, but it's still reassuring to hear Ozzie say it.
2) I think it's time to demote Boone Logan to AAA Charlotte, and for a couple reasons. First of all, he's pitched poorly in (very) limited action - despite what his shiny ERA may tell you - giving up three hits and four walks in less than four innings. Secondly, this is the first time in his career he's had to face advanced, major league hitters, and he'd be better served gaining experience while throwing more innings with the Knights. And lastly, if the Sox have a major weakness, it's their bullpen. So far, only McCarthy and Jenks have shown themselves to be dependable. For now, that's been ok, because the starters have been great (especially recently), and the offense has built up large leads. But eventually we're going to need good third, fourth, and fifth options out of the 'pen. Luckily, besides the hope that Thornton, Cotts, and/or Politte will step up, we have a good option at AAA: Javier Lopez. He pitched decently during spring training, and so far this season with Charlotte has surrendured only six hits and two walks in nine innings, while allowing zero runs and striking out eight. I think it's time to call him up, and send Logan down.
On to the main subject of this post: Sox starting pitchers and luck. If I asked you right now who's been the best Sox starter so far, not including today's game (because stats don't include that game yet), you'd either say Mark Buehrle or Jose Contreras, right? Well, yes, on the surface, that's true. However, there is evidence that in fact Javier Vazquez has been the best of the bunch so far, and that Buehrle and Contreras have been quite lucky. And if we considered who's been the worst so far, it would obviously come down to Freddy Garcia and Jon Garland. However, there is evidence that Garland has been extremely unlucky, so that his underlying performance has actually been better than that of Garcia, or at least close.
But what evidence? Well, there are two main categories of luck when it comes to pitchers. I will call them:
Sequential Luck and Locational Luck
The first, Sequential Luck, has to do with the order that hits and walks (and to a lesser degree, sacrifice flys, double plays, etc.) come in. For example, imagine two pitchers, A and B. Both throw a nine inning complete game, give up nine hits (all doubles), walk no batters, throw no wild pitches, and have flawless defense played behind them. And yet Pitcher A has an ERA of o.oo on the day, while Pitcher B has an ERA of 8.00 on the day. Why? Because Pitcher A gave up one double in every inning, and stranded the baserunner in every instance, while Pitcher B gave up all nine doubles in one inning, resulting in eight runs.
Now, you could argue that the difference in ERA is entirely Pitcher B's fault, because he wasn't able to settle down after giving up a double (or nine), as Pitcher A was able to do in every inning. And that's a valid point. (For the same reason I prefer to look at a pitcher's RA, or Runs Against, rather than his ERA; good pitchers pitch through the mistakes of their defense.) But at the same time, a good deal of the difference between Pitchers A and B has to do with luck. What if instead of nine doubles, Pitcher B only gives up two, but they happen to be in the same inning. His ERA for the day (1.00) would still be higher than that of Pitcher A (0.00), but his WHIP would be far lower (0.22 as compared to 1.00). Now let's say those trends continued for an entire season, Pitcher A giving up far more hits than Pitcher B, but Pitcher B's ERA being higher because of the order the hits came in. Any team with half a brain would rather invest in Pitcher B than Pitcher A, because his underlying performance was better, even if the results were not. In short, Pitcher B would be extremely unlucky.
There's a good way to measure Sequential Luck, called Component ERA (usually listed as ERC). A pitcher's ERC is what his ERA should be, given the number of hits and walks he gives up. Basically, ERC is WHIP calibrated to the ERA scale. Then you can compare a pitcher's ERC to his ERA: if the ERC is higher, he was lucky; if the ERA is higher, he was unlucky. It's also very easy to render a new stat - ERC% - which shows just how lucky or unlucky a given pitcher was. An ERC% is the number that you'd have to multiply a pitcher's ERA by to get his ERC. For example, the luckiest starting pitcher so far this season is Detroit's Mike Maroth. His ERA is 0.49, but his ERC is 2.96. His ERC% is therefore 6.04, meaning that his ERA is six times better than it "should" be given his peripheral stats. That's pretty much off the charts in terms of luck; the second luckiest pitcher so far, Brett Myers, only has an ERC% of 1.65. (It should be noted that Maroth's peripheral stats indicate that he should still have a very low ERA - 2.96; a high ERC% doesn't mean you've been a bad pitcher, only a lucky one. That same reasoning also works in reverse.)
On the other end of things, the starting pitcher with the lowest ERC% so far this season is Javier Vazquez. Wait a second - don't we know that guy? Yep, Vazquez's Sequential Luck has been absurdly terrible early this season, with an ERC% of 0.47. While his ERA is still an impressive 3.86, the number of hits and walks he's given up indicate that his ERA "should" be 1.83, more than two full runs lower. In fact, four out of the five Sox starters have been unlucky in this regard so far, with only Contreras (1.11 before today's game) registering an ERC% above the break-even score of 1.00.
Locational Luck should actually be termed Defensive/Locational Luck, because it has two distinct aspects to it, but I'd rather go with the simple name. Locational Luck measures what happens to a ball after it's been hit into play. Many studies have shown that, with the possible exception of knuckle-ballers, pitchers have very little control over what happens to non-home run balls hit into play. In fact, whether a ball hit into play falls in for a hit appears to be mostly a matter of luck, varying greatly from year to year even when all other performance metrics for a pitcher remain the same. And this shouldn't really surprise any observant baseball fan, because we've all seen the screaming line drive hit directly at the left fielder for an out, or the weak ground ball that sneaks past the second baseman for a hit. Baseball truly is a game of inches.
The easiest way to measure this is via a stat called (by Baseball Prospectus) Batting Average on Balls In Play (BABIP), also known (at ESPN) as Balls In Play Average (BIPA). League average for BABIP is usually around .300 (so far this season in the AL it's .303). Therefore, a pitcher with a BABIP of .250 is getting somewhat lucky, whereas a pitcher with a BABIP of .350 is experiencing just the opposite.
However, defense does play a role in BABIP. A team with a good or great defense will tend to lower their pitchers' BABIPs, making an adjustment necessary when comparing pitchers on different teams. However, for today we're just comparing Sox starting pitchers, who each pitch in front of a (nearly) identical defense, so that won't matter. Still, it's worth keeping in mind when considering BABIP; team defense is the second aspect of Locational Luck that I referred to two paragraphs above.
As with ERC, it is possible to translate BABIP into a stat that mirrors ERA; in this case, the new stat is the Defense Independent Pitching Statistic (DIPS), created by the brilliant Voros McCracken. And like ERC and ERC%, DIPS can become DIPS%, giving an easy indication of Locational Luck. Mike Maroth is also the early leader in Locational Luck, with a ridiculous DIPS% of 8.93. And the starting pitcher with the worst Locational Luck is Jon Garland, at 0.49. Wait, that guy sounds familiar too. Garland's BABIP is an atrocious .433 (ESPN has his BIPA at .400; I don't know why there's a difference, as I think it's the exact same stat, but it doesn't really matter for our purposes), meaning he's been giving up a hit more than 40% of the time a batter hits the ball in play against him. As for the DIPS% of the other Sox starters, Garcia and Vazquez are near the league average, but Contreras (2.53) and Buehrle (2.28) are two of the luckiest pitchers in baseball so far.
Here are the full numbers for the five Sox starters, through Saturday's game:
| Name | ERA | ERC | ERC% | DIPS | DIPS% | BABIP | K/9 |
| Buehrle | 1.93 | 1.77 | 0.92 | 4.39 | 2.28 | 0.190 | 3.86 |
| Contreras | 1.61 | 1.79 | 1.11 | 4.07 | 2.53 | 0.200 | 4.71 |
| Garcia | 6.23 | 5.51 | 0.88 | 5.16 | 0.83 | 0.293 | 5.40 |
| Garland | 9.18 | 8.61 | 0.94 | 4.48 | 0.49 | 0.433 | 5.94 |
| Vazquez | 3.86 | 1.83 | 0.47 | 2.71 | 0.70 | 0.259 | 7.71 |
I've highlighted a few of the more out-of-line numbers. As you can see, two of our starters - Vazquez and Garland - have been quite unlucky. Vazquez has probably been our best starter so far. His DIPS score is far better than the others, and his ERC is only a shade behind Buehrle and Contreras. Garland's BABIP is just about guaranteed to drop, which should significantly lower his ERA.
On the other end of things, both Buehrle and Contreras have been quite lucky, and both in terms of BABIP. You'll notice I also listed each pitcher's strikeouts per nine innings. That's because DIPS% is less significant for a strikeout pitcher, and more significant for a contact pitcher. For instance, you'll notice that Vazquez's DIPS% is 0.70, even though his BABIP is a lucky .259; that's because he strikes out 7.71 batters per nine innings, meaning that he remains in control of more batters and does not leave their outcome to luck. I fully expect Contreras' K/9 rate to rise significantly (his career mark is 7.59), meaning that even though his BABIP will rise, it won't affect his DIPS, or his ERA, all that much. On the other hand, Buehrle has the lowest K/9 rate on the team, and his career rate of 5.37 is lower than everyone but Garland's; when his BABIP regresses to the mean, it's going to mean a lot more runs given up. That doesn't mean he's suddenly going to become a bad pitcher. Rather, it just means we can't expect him to finish the season with an ERA under 2.00.
Of course, it's still very early in the season, rendering much of this meaningless. Regardless, I think it's worth introducing these stats now, so that we can revisit them later when more data has piled up.
posted by Jake Berlin | 7:23 PM
0 Comments:
Post a Comment
<< Home