Thursday, June 25, 2009

Analyzing Dice-K

Two days ago, Bob Ryan wrote an article entitled It Just Doesn't Add Up for the Boston Globe. The article is about the fact that Daisuke Matsuzaka is struggling this year, and isn't worth the money the Red Sox paid to sign him. In it Ryan briefly touches on the topics of interest, but never does any sort of actual analysis either of baseball's salary scale or of the reasons for Daisuke's ineffectiveness this year. So I will now attempt to pick up the ball he has dropped. Because with how awful the Red Sox media is, Lord knows someone needs to fill the void by competently discussing the team (These guys do a better job than I will ever manage, by the way).

Last year Daisuke went 18-3, a phenomenal Win-Loss record. However, he was nowhere near as good a pitcher as that suggests. Before I get into why, let's lay down some basic parameters here. The goal of pitching is to prevent runs, plain and simple. There are different ways of approaching this, the most effective of which is striking people out. Ground balls are another extremely effective method, as no amount of steroids Raul Ibanez injects before breakfast can cause a ground ball to go over the outfield fence. However, pitching is not the only factor here. Team defense accounts for approximately 1/3 of run prevention, leaving the pitcher responsible for the other 2/3. We'll come back to this point. And then we come to luck. It is really amazing how much dumb luck influences pitching. To paraphrase Bill James (a.k.a. Sabermetric Jesus), it is entirely possible that what I am going to refer to here as luck is not actually that, but rather just something that we do not yet know how to measure. And so it may be, but until we devise that measurement, luck it shall be. Things that are primarily luck include Batting Average on Balls In Play (Hereafter BABIP), Home Run per Fly Ball %, and Strand Rate. For more complete rundowns of these numbers, I recommend Fangraphs, The Hardball Times, or (If you'd rather pay money for something you can get for free) Baseball Prospectus. BABIP measures the amount of balls hit into the field of play that result in hits. Now, there are obviously things that affect this slightly, but by and large every pitcher's BABIP should be about .300. Numbers lower or higher than that are simply good luck or bad luck, and are completely unsustainable for extended periods of time for reasons beyond a pitcher's control. HR/FB for all pitchers should be right around 11%. The home park a pitcher spends his time in can affect this, but the only way to be certain to prevent home runs is to lower the rate at which you allow fly balls, as 11% of them are going to go over the fence. And finally, strand rate is simply the number of players who reach base against a pitcher who do not score. Simple enough. This should normalize around 72% for an average pitcher, with better pitchers being able to post slightly better numbers than that (For obvious reasons). Okay. I can't believe I haven't run down that stuff before, but that's done now. Since there was quite a bit of it, let's start fresh with a new paragraph.

As I mentioned above, Dice-K was 18-3 last year, with an excellent ERA of 2.90 (Earned Runs Allowed per 9 Innings, for those of you who are baseball neophytes. You've all stopped reading by now, haven't you?). But in truth, he was nowhere near that good. If we compare his 2008 statistics to his 2007 statistics, a year in which he was merely a solid pitcher rather than an ace, we find some rather startling results. Let's do this in chart-ish form.

K/9 (Strikeouts per 9 innings)
2007: 8.84
2008: 8.27

BB/9 (Walks per 9 innings)
2007: 3.52
2008: 5.05

2007: .306
2008: .267

2007: 11.7
2008: 7.5

Strand Rate:
2007: 73.9%
2008: 80.6%

Notice anything? In K/9 and BB/9, two luck-independent outcomes which are largely in the control of the pitcher, Matsuzaka actually performed worse in 2008 than in 2007, even though he improved his ERA by a full run and a half per game. This improvement is because of the last three numbers listed, luck-related statistics in which Daisuke performed at a rate far above the sustainable league average. In fact, if we normalize his luck-dependent stats to determine what his ERA should have been in each of those years (xFIP, if you're wondering), we find that in 2007 he pitched with an ERA of 4.42, while in 2008 he checked in with a worse showing of 4.82, a progression in the wrong direction. So he actually got worse between the two seasons, which still does not entirely explain this season's disaster. Ignoring the components to jump straight to the conclusion, what is his xFIP so far this year? 4.93, almost identical to last year. So what's the problem? Primarily bad luck. His K/9 and BB/9 have actually both improved from last year, and while there's still a long ways to go before his BB/9 hits a number that we can call 'good' (I'd settle for 'acceptable' at this point), it's legit improvement. But his BABIP has gone through the roof, checking in at an obscene .441, and his fly balls allowed are clearing the fence 17.5% of the time. These are the primary reasons for his decreased performance, and are things that should work themselves out over time. But is it entirely bad luck? No.

The thing about normalizing BABIP to exactly .300 for each pitcher is that this assumes that pitcher has a league-average defense taking the field behind him. This is almost never the case, meaning small variations in team BABIP should be expected, with good defensive teams posting a lower number, while poor defensive teams post a higher one. Last year's Red Sox were an excellent defensive team, posting the fourth-best UZR (Ultimate Zone Rating, available on Fangraphs. Why UZR? Because fielding percentage means less than nothing) in all of baseball. This year's version comes in 29th, only better than the beyond-inept Washington Nationals. The primary culprits are Mike Lowell, Julio Lugo and Jacoby Ellsbury, all of whom have fallen off a cliff (Since Lugo was fringe-average in the first place, this is not at all a good thing for him). The fact that Jason Bay has posted a worse showing in both years than Manny Ramirez did over the first half of last year is not really helping things, either. So while we should expect Daisuke's BABIP to rise above league average, we're still talking about a range closer to .310, not the stratosphere in which it currently resides. Given the chance, he will resume pitching at the level we are accustomed to seeing him at, even if we do not always understand what level that actually is.

Now let's go back to our good friend Bobby Boy. His article is not necessarily about how Daisuke is finished, but rather about the fact that Bob believes he is overpaid. Now that we've finished with a performance analysis, let's move on to salary matters. To argue his thesis statement in the article in question, Ryan provides us with the following pieces of evidence:

Daisuke's salary, presented in a misleading fashion
That is all

Besides making the reader question how Ryan still has a job, this does not actually accomplish much. Let's go through money matters now. Daisuke was a free agent in 2006. This is important because, in MLB, the rookie salary scale artificially depresses the salaries of young players for the first six years they are in the majors, years which are often their best. This makes baseball salaries often seem different than they actually are. Player salaries need to be divided up between cost-controlled youngsters and free agents to get an actual idea of what players are paid. And on the free agent market of the winter of 2006, free agents were paid about $4,000,000 per win above replacement player that they provided for their team the previous year. This amount increased linearly with each additional win, meaking a 2-win player worth $8 million, a 3-win player worth $12 million, and so on (This can be easily determined by merely dividing the total dollars all free agents were given that offseason by the WAR they collectively provided the previous year). Ryan does not once mention this. Now, I am not saying the best way to run a team is to go out and spend money on free agents. But if you do so, this is what it will cost you. This is the same free agent market where Carlos Silva is worth $12 million annually for four years (The Silva deal was signed one year later, when free agent value had risen to $4.5 million per win. The point stands). Providing this context makes Daisuke's deal look very reasonable, possibly even exactly where it should be, which is probably why Ryan did not put it in his article (Showing blatant disrespect for his reading audience, as he is presuming they will not know this. Either that or he does not know it himself, and is unwilling to research a subject before he writes about it, which calls into question why he has his job). In 2007, Daisuke was worth 3.9 Wins Above Replacement (WAR), or $15.6 million, based on the year he signed his contract. In 2008, 3.3 WAR, or $13.2 million. Per the terms of his contract, in 2007 Daisuke was paid $6 million by the Red Sox. In 2008 he made $8 million (By the way, lest you think I shouldn't be riding Ryan about the research, these numbers took me 17 seconds to find using Google. I timed myself). So why is he overpaid again?

Well, this is because Ryan combines the posting fee paid to his former team in Japan as part of Daisuke's salary. It is not. Daisuke was a professional athlete under obligation to remain in Japan. To release him from that deal, his team offered negotiation rights with him to the highest bidder, who happened to be the Red Sox. Essentially, the Red Sox spent $51 million to buy out his deal with his old team. Whether that was money well-spent or not, it is certainly not a part of his contract. And the fact that this money bought exclusive negotiation rights is almost certainly what enabled the Red Sox to sign Matsuzaka to what is actually a below-market deal. As you would not know if you read the Ryan article, Daisuke signed a contract for 6 years, for a total of $52 million dollars. Per the going rate of the free agent market, of which Daisuke was a member, this valued him at almost exactly the same worth as Odalis Perez. Why are you complaining, Bob? If the Red Sox get no value out of Daisuke the rest of the season, over the first three years of his contract they will have made a profit on the wins received for the money they've paid him so far, making the signing a good one for the team. And I hate watching him pitch. There is nothing more frustrating than watching a pitcher who seemingly cannot throw strikes. But it is still a good contract.

Ryan also seems to be objecting to one more thing, which is that Daisuke was billed as the Next Big Thing, and he has not delivered on that. Fair enough. But who billed him as that? Was it the man himself? Certainly not. It was the media, of which Bob Ryan is a large part. After Daisuke's seventh career start, Bob Ryan announced to the world that "The Daisuke Era has officially begun." Is there a single mention of the fact that what Ryan primarily objects to about Daisuke in his article is something that he himself helped create? No, there is not. Because writing that would involve both honesty and accountability, two qualities you are not likely to find in an article written by Bob Ryan. If you want to learn about baseball, the best thing you can do is avoid reading the Boston newspapers. All they will do is keep you mis- and uninformed.


sox19_wheels said...

There's a reason Bob Ryan didn't include this stuff: it would bore any reader. I honestly couldn't complete reading this because my head hurt from all these statistics I didn't understand. I'm sure this is well thought out and accurate, but a pain in any ass to read.

By the way, I know we're not talking fielding percentage, but Ellsbury committed his first ever error last week. Join that with the fact that he has to cover for J.D. Drew's slow ass and a 420' right center field, he is doing fine.

HOLLA(R) said...

We're still within the small-sample-size-fluke range w/r/t this year's UZR, but Drew rates substantially better than Ellsbury thus far. Ellsbury was better than Drew last year, though. In terms of guys he has to cover for, let's point some fingers at Bay. His metrics are abysmal. Any statistics I can help clarify for you?

HOLLA(R) said...

Also, the fact that a players performance can be more accurately predicted using solely component stats rather than the actual final numbers doesn't make you feel a tingly sensation in your pants? Lord know it has that effect on me.

That was too much information, wasn't it?

UncleRick said...

Seriously, Mike...get a job! There are lies, damnable lies and then there are statistics. Watch the game, have fun, see the Red Sox win again. Could your statistics have predicted the outcome of last night's game? The Heartbeat is all that matters! Stop watching the game for the stats!

HOLLA(R) said...

Hey now. I watch the Red Sox almost every day. Rather than watching the games for the stats, I use the stats to better understand what I am seeing, as I prefer having a conceptual knowledge of the events happening on the field, rather than trusting my eyes to take them in at face value. Part of what I don't understand about what has been (badly) dubbed 'The Statistical Revolution' is the underlying assumption of the old-school baseball guard that these new people want to do away with all of them and replace them with computers, which will reduce the game to a playerless binary code. No. The computers are used to help understand the game itself, which is the point of the whole enterprise. Scouts are a necessity as well. The numbers are simply a way to improve the picture we all enjoy looking at.

UncleRick said...

Have you been watching Jason Bay in left field recently?

HOLLA(R) said...

Yeah. He looks good, I'll give him that. But there's a good consensus among the fielding metrics that he's well below average, both with Boston and before that with Pittsburgh. And with fielding, I trust the numbers over my eyes. Eyes are how Nate McLouth, worst CF in the game, won a Gold Glove. As fundamentally sound as someone looks, one thing my eyes can't really take into effect is range, and UZR handles that very well.

UncleRick said...

I guess what I'm saying about Bay in left field is I would rather see him make a running, sliding, game-ending catch than talk about how he isn't a good outfielder. Your statistic cannot predict that outcome and do not enhance my viewing pleasure.