D

Rob’s opinion is one of a national guy who doesn’t watch this team on a regular basis. Millwood has been great this year, and luck doesn’t have anything to do with it. Right on, Evan.

Ahem! Cough, cough…since you brought up Nellie, let’s look at some Fangraphs numbers, shall we?

Cruz is currently 3rd in the AL in both of the following: 9.6 UZR 15.4 UZR/150

That is higher than Ichy, Upton, Podsednik, or Granderson. He is exactly twice as good in the OF as Crawford.

Defensive metrics are far from perfect at this point, but that is the best we have to work with, and shows Cruz as an excellent fielder.

@Evan-
1) Agree his argument on Millwood is silly and unfair. He’s been a horse and a true stopper.
2) Agree that he must have never seen Cruz stand frozen on a Texas leaguer, take the wrong angle and watch the ball fly over his head or start a slide at the chalk and end up with the ball dropping between him and the wall (he does have an arm to be reckoned with, however).
3) Another unquantifiable factor might be that Millwood’s increased pace and reduced nibbling has helped the fielders, too. Keeping them in the game better.

Evan-

I usually love your stuff, but your analysis of Neyer’s points is a bit naive.

You say, “Stats analysts don’t believe in luck.” Wrong; they just don’t call it luck. They call it low-probability outcomes. For every bell curve of expected outcomes, there is a tail, and Neyer is right that Millwood’s stats place him very squarely within that tail.

They further believe that as sample size increases, outcomes hew more closely to the mean. In other words, as Millwood pitches more this season, the number of runners that score will become closer to normal. This isn’t, as you argue, a criticism of Millwood for stranding runners; it’s a recognition that there’s a low probability he will continue to do so, and, if he doesn’t improve the other parts of his game to compensate, more runs will score against him.

While I’m not totally with the stats guys that there’s no such thing as clutch, Neyer makes a solid point here. Perhaps it might be worth your while to pick up a copy of Moneyball (which you recently noted you’d never read). It’s quite an entertaining read and it might give you better context to understand what guys like Neyer are saying.

Right on, Evan. Some of the comments after the ESPN article are incredible. One even told all us Milly supporters; “You want Millwood in the [All Star] game? Go online and vote for him a dozen times, then make a fake email address and go vote a dozen more. Or, convince people he’s from Japan.” Brilliant!
If Milly pitched for any other team I’ll bet chalking his success this year up to luck doesn’t even happen. Obviously if the Rangers have any success at all they are just lucky. I really want to hear what TAG has to say. I’m hoping he comes unhinged!

The whole problem with Neyer’s point is that (in addition to the strand rate), he points to the very low batting average on balls in play (BABiP) and that Millwood’s is the lowest in the AL this season (about 100pts lower than last year and 80 lower than 07). His argument relies on the assumption that all BABiP will eventually trend to the same # and that since Millwood’s is lower this year that means he lucky, as hit balls are going for outs and not falling in.

Here is the HUGE problem with that analysis – it assumes that all balls hit off the bat are the same and that simply luck dictates what happens. It DOESN’T take into account a pitcher inducing WEAK ground balls, pop ups and lazy fly balls due to changing speads, moving the ball around and hitting corners – all to keep a hitter off balance (and all things that Kevin has done tremendously well this, in addition to working quickly to keep hitters off-balance and hit fielders even MORE read).

Greg Maddox made a career out of that. Was he lucky for 20 years? Yes, Millwood doesn’t have tremendous stuff. But that doesn’t mean he can’t pitch tremendously well. The 2 don’t alway correlate (see Mad Dawg Madrigal).

You have to watch the games to see what is HAPPENING on those balls put in play. I don’t remember leaving a single Millwood start this year thinking: “boy he got lucky tonight, they were hitting missiles at people.” Besides, there is no way that luck would hold up over 15+ starts.

I pointed this article out to Evan for 2 reasons:

1. I can’t stand when national media tries to offer an ill-informed opinion on a local team without having ANY first hand knowledge (or frankly when ANYONE tries to offer opinions when they aren’t informed); and
2. While stats have importance, improper analysis and over-dependence on them can be foolish.

Enjoy the game tonight – I’ll be at the Porch watching the first part of the game and then at a movie with Mrs. Jack Daddy.

@JackDaddy — I’m not sure why you think Neyer or Anderson are offering ill-informed opinions. The criticism you offer of them seems to boil down to, they haven’t watched Millwood and thus don’t know what they are talking about.

You suggest that Millwood’s BABIP is low because he’s eliciting a lot of weakly hit balls, and cite Greg Maddux as an example of someone who has done that for 20 years. Maddux, though, has a career BABIP of .289, and his ERA is only slightly lower than what his peripherals suggest it should be.

Research has indicated that, for the most part, pitchers have very little control over whether or not balls in play go for hits…that is mostly defense and chance.

Either Millwood has developed an entirely new way of pitching that no one, up to this point, has ever done before, or else he’s an outlier who will regress to the mean in 2009.

And I feel like this is deja vu all over again, primarily because I had this same argument in regards to Armando Galarraga last year.

Evan,

Good job. Millwood is certainly a better pitcher this year than he was in 2007 or 2008.You should realize by now that the stat freaks think all numbers can be considered in a vacuum and anything that appears to violate that point of view is “luck” or ” random variation”.

As to Nellie’s defensive ability It is always amusing to hear the stat guys say defensive stats are not “perfect”. Most of the newer such stats are very subjective and should be viewed with suspicion. Any stat which tries to predict what should happen instead of recording what did happen has a built in problem.

@Auditor —

I don’t understand your criticism of the pitch count article.

[...] regards to the very animated conversation about Kevin Millwood’s success this season, Adam Morris takes me to task on Lonestarball with lots of thoughtful reasoning. And [...]

[...] regards to the very animated conversation about Kevin Millwood’s success this season, Adam Morris takes me to task on Lonestarball with lots of thoughtful reasoning. And [...]

This is all very interesting, in a stats-drive-players (rather than a players-create-the-stats) sort of way. I hardly feel quantitatively skilled enough to enter Neyer’s little numerical world, so I won’t.

I have read Moneyball. It’s a great read. But what stumps me is Billy Beane’s lack of success with this approach at Oakland. How many championships have the A’s won under him? How many AL pennants? Division championships?

@AJM,

What exactly do you not understand when I said he did not make a point? He was rambling about his own opinion without and meaningful insight. If you are going to be critical, at least make an argument with some statistical input. The best way to describe his overall attitute is with this quote from his June 29th blog:

“You know, before we get all mushy and dew-eyed about the Nolan Ryan Approach to Developing Young Pitchers, let’s see how this approach works.”

Sorry, I agree with Joe and Bill on this one.

Evan-

All fair points. Sorry for my confusion on your having read Moneyball. I think I was confusing you with Jamey Newberg on that issue– you are the only folks I regularly look to for Rangers analysis, and you both do a great job.

@ JE

It’s not that they are too lazy to learn, they just can’t understand. Can’t understand the concept that players don’t have a unique ability to control every outcome in baseball and a lot of it is luck. That BABIP trends toward a certain level for EVERY pitcher, maybe not exactly the same level but certainly not .266.

@ JackDaddy

Maddux made a living on not giving up homeruns because of his high GB% and by walking 1.8 per 9 innings.
Please NEVER EVER AGAIN compare Maddux and Millwood. That is just awful. You should apologize to Greg Maddux (I feel dirty for saying how great Maddux is, as I am a Mets fan, but geez, Maddux and Millwood, all they have in common is they are right handed and their last names start with M’s).

I don’t think you understand luck or the role of luck in statistical analysis.

There is skill. A skill is someting that a player — or perhaps a team has — that is can reproduce consistently. A pitcher’s ability to strike people out is a skills. A batter’s ability to draw walks is a skill.

We know that they are skills because they tend to correlate from one year to the next. Given large enough sample sizes, we see people who are good at during one period tend to be good at the next period.

ERA is not as much a skill. There is much more variation from year to year. Batting average on balls in play is somewhat a skill a players — with a fair amount of variation, though — and not much a skill for pitchers.

Skill is the predictable part, or at least the theoretically predictable part. Statistical analysis attempts to do a better job of predicting that stuff. Skill is the stuff that players can get get better at through their early years, and the basis for development and the basis for most projections.

Luck is the rest. There’s a LOT of luck in baseball. Luck is why we need large sample sizes. People can be lucky or unlucky in the short term, but in the longer term it tends balance out overall. Call it being hot, or call it being lucky. It’s the unpredictable abberations that are not reproducable.

Actually stats analysts care A LOT about luck. They try to figure out which part is luck, and which part is skill. Projectsions are about trying to take out the luck part, and project the skill part in or to make predictions. Take out the unlucky part, take out the lucky part, and what’s you’ve got left is skill.

*************

Now, are you saying that Millwood as developed a skill that he didn’t used to have? I don’t think you are. You admit that he won’t be this good after the break. That means that you think there’s been a element of luck.

It could be that you think he’s been luck AND good — as much champions have to be. The difference between you and Neyer is that you think it’s been more skill than luck, and he thinks its been more luck than skill. But you both agree that there is luck AND skill involved.

This post almost seems to be a joke.

(1) Evan recognizes that Millwood has gotten lucky to strand as many base runners as he has, yet rips on Neyer for pointing it out.

(2) Grant talks about the effectiveness of good defense, yet why does Texas have the 10th best pitching in the AL (5th worst). Does the defense only play well for Millwood?

(3) Assuming you don’t believe in “clutch” (which I don’t), wouldn’t it be obvious that having a lower batting average against with runners on, and an even lower batting average against with runners in scoring position seem to indicate a little luck was involved?

(4) Let’s just look at Home runs. Millwood has given up 11 home runs in 266 at bats with noone on. Yet he has only given up 3 home runs in 185 at bats with runners on base. This means that of Millwood’s 14 Home runs, only three have been multiple run homers. That sounds like luck to me.

By the way, here is an interesting stat. Millwood is tied for 9th in the AL in the number of Home runs given up. That would be a scary stat to me if I were a Rangers fan. A couple grand slams will certainly make Millwood’s ERA go higher.

Josh: If I remember correctly, Millwood was a teammate of Maddux. That is one other thing the two have in common.

@Jack Daddy:

RE: Your comment to Josh….

What you said would be fine if it had any semblance of being true. However, it doesn’t. Maddux’s career BABIP was .286. In the last 5 years of his career, Maddux did not have a single year where his BABIP was below that number. However, Millwood’s BABIP this year is .261 with a lifetime .299 BABIP.

The thing that made Maddux effective late in his career was a lack of walks. Millwood knows nothing about that subject.

We’ll be able to settle the issue of an increase in hard hit or softly hit ball rates when Hit F/X accumulates enough information. Sadly, based on what I’m reading here, the people who are making the anti-statistics arguments (and are buying into the illusion of small sample sizes) wouldn’t even bother learning about it.

They’ll just plug up their ears, assume the fetal position, and chant “Stats suck, stats suck, stats suck” while the rest of the world leaves them behind. Sad but true.

This article would have seemed ridiculously backwards 15 years ago to the cutting edge. The fact that there are still sportswriters out there today producing this kind of half-witted, ill-informed drivel (really, strand rate is a controllable skill that a pitcher can learn at age 34? DO GO ON!) is pathetic.

Evan, Adam Morris over at lonestarball.com has posted a response to your post. He addresses many of your points in response to Neyer.

http://www.lonestarball.com/2009/7/3/937068/kevin-millwood-rob-neyer-evan

C’mon man, you know that Rob was arguing about the # of baserunners, not the number of baserunners stranded. Millwood is allowing too many runners to reach base to sustain that ERA. I’m happy for the guy (have also owned him in a keeper league for years) but you know he can’t keep this up. I don’t know what the record is for # of baserunners allowed with a lower than 3.00 era, but it has to be close to his #’s this year.

A lot of people are complaining about these “moneyball statistics” taking away from the game (i.e. that watching a prospect play is the only way to truly evaluate them). What would you say if I told you that Wal-Mart would decide what items to offer in their store not by aggregated market research showing purchasing trends across their customer base, but instead on what a few “scouts” saw selling well at competing outlets in their region of the country? I hope you would think they were crazy because each “scout” would likely have a different value set of what was important in a burgeoning product and, furthermore, would probably not be able to stand at the store 24/7/365 to watch every customer, every day. Hopefully, you would immediately sell stock in any company with a management team that foolish.

Yet for some reason baseball people think it’s just as well that their team spends millions of dollars on a prospect or free agent just because he’s a (least favorite scouting term ever) “5-tool player” or he has “intangibles” (read: short, white and unathletic) regardless of his measurables up to that point. What these people usually fail to consider is that a scout watches this player a set number of times and can only observe a set number of things, whereas accurately compiled statistics clue us in to trends that the naked eye might miss because humans cannot see everything.

The hands on, “scouting approach” may produce superior results from time to time, but over time (and given similar resources) the team that relies on hard data and measurables will out-produce their intuitive competition because they will be making decisions based not on emotion, but on unbiased facts. This is true whether you’re playing poker, running a corporation, or judging baseball prospects.

Thus, I would believe that Kevin Milwood has some special gift to induce weak contact to his infielders if he could produce this result consistently over the course of even 2-3 seasons, but I will not believe that at his current age he’s suddenly discovered something no one else in baseball has ever been able to produce. It is far more likely that the man has enjoyed a combination of improved defense and, mostly, luck.

By all means, however, please continue believing that investments in overpriced veterans and “intangibles” will win World Series titles. God knows it’s worked pretty well for the Yankees in the past decade.

Evan wrote, “The stats community argues that Millwood can’t control his defense, so the impact the improved defense has had on the pitching staff must be chalked-up to luck or “random variation.” But then why is 2007 used as a baseline comparison? Can’t 2007’s high ERA be somewhat blamed on an overall awful defensive team that committed an AL-worst 124 errors?”

Good thing we’re not looking at his ERA or anything reliant on defense. His 2007 FIP, which strips out fielding and other luck based factors, was 4.55. His 2008 FIP was 4.02. His 2009 FIP is 4.48.

Or in other words, you can knock down strawmen all you want, but you’re just making yourself look silly at this point.

@Matt

Just another case of people who don’t accept sabermetrics trying to flip them back on us without understanding how they actually work.

In 2005, Millwood had a 2.86 ERA with 79% LOB. Guess that was luck for 192.1 innings. I think what all of the purists are trying to say is I can sit there and pin-point the stats according the point I want to make, but for those who do not watch the game, stats just water down the enjoyment. I am not a stathead – I put a whole lot less emphasis on W-L because there are factors for that, as there are factors for each stat that you just cannot see unless you watch the game!

@Brett
And yet I doubt Evan writes a column about how he was horribly wrong and Neyer was right. If he did, then much respect. But somehow, I don’t see it happening…..

What do you think of Kevin Millwood’s regression to the mean now?