World Series Champs Highlight Conflict Over Sabermetrics

Hey, there folks. I’m the new writer around these parts. You may know me as one of the writers on Off the Facade or as one of the extinct columnists from MVN’s podcasting site 360 The Pitch. I’ll be writing regularly here now as well, providing the Yankee Ying to David’s Red Sox Yang when as the rivalry heats up. I’ll be looking at Hot-Stove moves and anything else that may come down the pike. It’s a bit of a tough time for the stats folks these days as two key players in the sabermetric revolution won’t be returning to their jobs, and a supposed smallball team just won the World Series. So without further ado….
The White Sox are a sabermetricianís worst nightmare.
They just finished in improbably season capped by an astounding 11-1 run through the postseason with an awful offense. The team had an on-base percentage of .322, good for 11th in the American League, six points below the league average. Yet, they also had a slugging percentage of .425 and an ISO of .158. Both of those numbers were one percentage point above the league average.
So what then does this tell us about the White Sox offense? First, itís clear that this smallball/Ozzie-ball talk is unfounded. This was a team that didnít rely on getting on base and manufacturing runs because they werenít that good at getting on base. They were average to hitting for power, but somehow, they won.
The first clues to their success come in the way manager Ozzie Guillen constructed the lineup. He adhered, knowingly or not, to a fairly decent lineup based upon each playerís on-base percentage. Here is the World Series Game 1 lineup with OBP.

Podsednik .315
Iguchi .342
Dye .333
Konerko .375
Everett .311
Rowand .329
Pierzynski .308
Crede .303
Uribe .301


This lineup is designed to maximize runs from the start. It leads with the top four guys on the team in OBP. While the fact that the top two OBP leaders on the World Series champions were .375 and .351 is sad, Ozzie was able to give clean-up hitter Paul Konerko the best opportunity to come up with runners on base. After that, Everett remains in the fifth spot because of his slugging while the bottom third of the lineup were awful at getting on. (Tell me: Does this lineup miss Frank Thomas and his career .427 OBP or what?)
Meanwhile, as David pointed out a few days ago, the White Sox fluke victory came about through their miraculous pitching.
But as a sabermetric backlash embraces baseball with the dismissal of Paul DePodesta and resignation of Theo Epstein, itís important to look at how a decidedly non-sabermetric-minded team still adheres to a basic principles of baseball statistical analysis: Put the guys who get on base first in the lineup to maximize your teamís run production. If even zany managers like Ozzie Guillen understand this concept, why is baseball so afraid of sabermetrics?

Advertisements

16 Responses to World Series Champs Highlight Conflict Over Sabermetrics

  1. Rob Bonter says:

    All the Red Sox have to do to fill the Theo-sabermetrics gap is promote Bill James to GM. (Yes, I’m kidding.) By the way, if Theo is/was such a hot-shot sabermetrician, why did he hire two career LOW WALK-RATE guys as his hitting coaches for Boston and Pawtucket? How can a slasher/hacker teach strike zone discipline when he didn’t practice it during his active career?

  2. Johnson says:

    I think you are placing too much importance on OBP here… the regular season numbers have the sox within .012 of every team except NYY and BOS. They were basically mid-range… not so extreme.
    And the notion of putting the batters who get on base witht he highest frequency early in the order is hardly a discovery of sabermetrics… kind of common sense.

  3. Here’s what I posted on Baseball Perspectives in the comments section.
    I think you bring up some valid points, but not every manager constructs lineups with their top OBP guys leading off. Take the Mets. Jose Reyes had 675 lead-off at-bats. Yet, his OBP was .299 out of the one hole (and .300 on the season). Off the Mets regulars, his was the lowest. If everyone knows that you’re supposed to bat guys who get on base in the one hole, why is Reyes batting first? The same can be said of the Marlins too. It’s not as wide-spread an accepted belief as it should be.
    On another note: I’ll have to explore the relationship between OBP and runs scored a little closer, but a 10 percent difference in OBP could have a significant impact on a team’s success especially in turning close games (the White Sox lucky speciality) into not-close games.

  4. Doug says:

    I think the point to the White Sox was that their opponents’ OBP was awful — because of the pitching and the defense — and that gap is what wins games. Baseball is not a contest to pile up counting numbers, it’s a contest to score more runs than the opposition.
    Sabermetrics isn’t invalid, but it is infested with over-generalizations that need to be re-examined. For example, because we don’t have a good system for rating fielding doesn’t mean we can pretend it doesn’t exist. Another item: the White Sox set out this season to reduce the standard deviation in their run scoring; does the fact they never scored 19 runs against the 12th man in somebody’s bullpen actually mean anything? Do the imbalances in the schedule now warp statistics beyond easy rectification?
    We can go on all night.
    BTW, calling Chicago’s victory a “fluke” is insane. They led their division from Opening Day, won by six games, would have won 100 games if Bruce Froemming and Hunter Wendelestedt hadn’t been out to show up Ozzie Guillen in a series in Oakland in May, and then they scored twice as many runs as they allowed in the playoffs. The fluke would have been if they didn’t win.
    When the hairsplitting got crazy was in, say, BP’s analysis of a division series. 30 points of OBP isn’t meaningful in 5 games, it represents four or five baserunners all else being equal.

  5. Doug says:

    BTW, an OBP six points below league average is one baserunner A WEEK. Six points is six thousandths. We need to get over this incredible, ridiculous hairsplitting. There’s a huge difference between a .400 OBP and a .300. .322 versus .328 is simply not all that important.
    The Chicago offense was not “awful”, it was “pedestrian”..

  6. Benjamin Kabak says:

    I think you can point to a 35-19 record in 1-run games as something of a fluke. They had great pitching. That’s why they won, but don’t expect them to repeat it next year.
    Pitching wins. Limited your oponnent win games. But eventually teams start to lose one-run games also. Just look at the Nationals.

  7. Johnson says:

    My response to the comments on my site:
    Yes, sometimes managers donít lead-off the best OBP, but this is part of my point…
    There are other factors that come into play… Reyes, for example, had 60 stolen bases and 41 extra-base hits on the season (17 triples, 9 with none on/out). This may not quite balance out his lower OBP, but it certainly puts him in scoring position more often than the mass of guys who posted .320ish on-base with stolen bases in the few to none range and a handful of extra-base hits.
    Plus, speed like that of Reyes gives you a little more control over the game when he does get on base…
    I am not going to tell you that the Mets lineup was well crafted, it obviously wasn’t (The clear #3 hitter totaled a whopping 15 at-bats in that spot). But there is an argument to be made that Reyes was their best lead-off option.
    OBP tells us a lot about the quality lead-off hitters, but it doesn’t necessarily define the spot.
    Every manageródiscounting anyone who might be brain-deadóis looking for ways to get runners into scoring position early… OBP, and sabermetrics in general, provides a gauge for success helps in figuring new ways to accomplish it, but we have to be careful not to generalize the goal out of existence.
    ______________
    oh, and Doug, nobody wants to say that Sabermetrics is invalid… just that its limits are sometimes, er… overlooked.

  8. David Gassko says:

    . For example, because we donít have a good system for rating fielding doesnít mean we can pretend it doesnít exist.
    Um, sabermetricians don’t. I publish probably the best publicly available fielding stat out there, Range. Ultimate Zone Rating is even better, though no longer fully published. Zone rating is good, as are Baseball Prospectus’s stats. Fieldings is the next great frontier in sabermetrics. The most statistically-inclined team in the MLB are using fielding metrics to improve. Look at Oakland for a great example of that.

  9. Doug says:

    Fieldings is the next great frontier in sabermetrics. The most statistically-inclined team in the MLB are using fielding metrics to improve.
    No doubt, but I am not sure we understand defense any better today than we did twenty years ago. Personally I love DER — and I think McCracken’s work has indicated it is more important than we ever suspected — but individual fielding performance is still murky.
    I suspect the real reason for that is what gets counted (PO, A, E) is still stuck in the Victorian era. ZR was an attempt to clean that up, and the professional metrics are pretty good. But FRAA is, as best as I can tell, barely improved from Pete Palmer’s fielding LWTS from 20 years ago.

  10. Doug says:

    I think you can point to a 35-19 record in 1-run games as something of a fluke. They had great pitching. Thatís why they won, but donít expect them to repeat it next year.
    After making the disclaimer that yours truly is a lifelong Chicago White Sox fan, I need to say two things. No, I don’t expect them to repeat next year. Nobody has virtually everything go right like that twice.
    But to dismiss their record in 1-run games as a fluke may be premature. Having watched them carefully, I saw a recurring pattern that was just too common to be some random fluke. Chisox one-run games tended to follow the same script — seizing a small lead and hanging onto it for dear life. I suspect that the one-run strategies they employed mostly tended to keep lead expansion down and made a lot of games 1-run games that wouldn’t have otherwise been, which they got away with because of the run prevention. Further, their one-run strategies, at least in the first 80 games of the season, were remarkably effective at getting them that one run when they needed it. This may not repeat, not because of “luck”, but because something will happen to upset the balance that let it work. It’s worth watching for another year before writing it down as a fluke. When a team sets out to be good at something, says they are setting out to be good at something, then actually is good at something, you have to assume they might have some idea what they are doing… until they fail, that is!
    Two more random observations:
    I think the obvious trend toward reductionism of all baseball analysis to counting just walks, homers, and strikeouts — which is where we are headed with Three True Outcomes and DIPS and the like — is wrongheaded because it assumes the rest of the game can be generalized away. Don’t tell me it isn’t happening; it’s been obvious for a decade now.
    And, it may be blasphemy, but I think OBP is slightly less meaningful now than it was 20 years ago because now players work to improve it in artificial circumstances, taking meaningless walks down by 9 in the 7th inning, because they know the arbitrators know about OBP now. Metrics are always more useful when the people being measured think they’re doing something else.

  11. David Gassko says:

    But FRAA is, as best as I can tell, barely improved from Pete Palmerís fielding LWTS from 20 years ago.
    Sorry, but that’s just wrong. What I do is much, much better than Fielding LWTS were 20 years ago, or, frankly, are now. So are BP’s DFTs, Michael Humphreys’ DRA, ZR, and Charlie Saeger’s CAD. UZR, of course, is on a whole other planet.

  12. Doug says:

    What I do is much, much better than Fielding LWTS were 20 years ago.
    Granted… But how exactly do you know it’s better? What concerns me about fielding metrics is the inference chain seems kind of long. How do we know that the number that comes out the end actually repesents runs saved or lost?

  13. Doug says:

    It’s slightly off-topic, but as long as I’m here, first, David, I want to compliment you on your site and the level of discussion. My remarks about performance analysis are general bewailing about the state of sabermetrics, now dominated (apparently) by the self-satisfied Ivy League loudmouths at another famous site with an acronym like an oil company, with their interesting but intellectually arrogant and self-serving books, rather than specific comments. (What really annoys me is that I can’t stop myself from giving them eighty bucks a year.)
    That enterprise has reached and passed the dangerous point where they believe their models are more informative than actual baseball, highlighted by their adjusted standings which are three levels of inference away from the game. The White Sox victory, when compared with their essay in this years’ annual, is amazing reading. But then again, despite their name, their projections are just as wacky as anybody else’s.
    I am old enough to remember baseball Before Bill James, when we all just took Leonard Koppett’s word for it that somebody was good. The divergence between what James and Palmer found and conventional wisdom was gigantic. Today, I think the internalization of performance analysis by baseball has eliminated much of that divergence. There’s a lot of meaningless hairsplitting going on, and not just the announcers with their splits. Sabermetrics is here to stay because ballplayers work for money, and money has to be correlated with performance somehow… but so much of what happens is still in The Fog, especially once the ball is hit by the batter.

  14. David Gassko says:

    How do we know that the number that comes out the end actually repesents runs saved or lost?
    Read through the methodology until you get what I’m doing. I give a step-by-step example which should make it easier. I can do another one if you’d like. And when you finally understand the method, both its advantages and its flaws will become obvious. It took me probably 10 times of reading both UZR and DRA articles to fully grasp the system. So you also have to be a bit crazy.
    BTW, what happened to your comments from last night?

  15. PatDrake says:

    It’s easy to look at history and find thread that explains most events. And if you put the 2005 Sox into the mix, you could find a thread that puts them with the norm.
    The bottom line of Sabremetrics is this. When, oh when, is that philosophy going to win anything? Except for Theo and his $120Million payroll team that was largely built by Dan Duquette, this philosophy hasn’t even won a single playoff series. Zip, goosegg.
    It’s amazing that an analysis based upon statistics then attributes the failure of those statistics in post-season to luck. The only difference in the playoffs and regular season, re the game, are a)no need for a 5th starter in the playoffs and b)you have to play with/without the DH more often than in the regular season c)it’s compressed, just like every other sport and it resembles the regular season with 3 and 4 game sets.
    BTW, Podsenik’s OBP was .351 (I’m sure that was a typo).
    Here are my issues with Sabremetrics:
    1)A walk is a positive event, but to equate it with a single is ridiculous.
    2)Re Slg, have you statistically validated that a home run is worth only 33% more than a triple; a triple 50% more valuable than a double and a double 100% more valuable than a single, a home run 400% more valuable than a single, etc?
    3)Why is a hitters’ strikeout no different than any other out, yet a pitchers’ strikeout is a meaningful statistic of pitchers’ effectiveness? The inconsistency is startling.
    3)

  16. David Gassko says:

    Pat, first of all contentions 1 and 2 are kind of false. What most people use in sabermetrics are linear weights, which weight each event based on how much it actually impacts run expectency. A simply LW formula might look like this:
    .47*1B + .78*2B + 1.03*3B + 1.45*HR + .19*SB – .46*CS -.1*Outs = Runs
    That’s very rough; don’t use it for any calculations. Linear weights depend on run enivronment and other factors.
    Anyways, as for #3, the reason is simple. Strikeouts have pretty much the same value as any other out. However, while batters have a great amount of influence over what happens after the ball hits the bat, pitcher influence is very small and difficult to detect. What pitchers have consistent influence on is Ks, BBs, and HRs. That’s why sabermetrics values those three categories so much for pitchers.

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s

%d bloggers like this: