Announcement

Collapse
No announcement yet.

Oakland A's Performance Shows That 'Moneyball' Doesn't Always Pay Off

Collapse
X
 
  • Filter
  • Time
  • Show
Clear All
new posts

  • Oakland A's Performance Shows That 'Moneyball' Doesn't Always Pay Off

    Oakland A's performance shows that 'moneyball' doesn't always pay off
    The use of statistical analysis to identify underappreciated players may lead to periods of success, but the concept has never been a panacea for the problems of a low-budget team.

    Michael Hiltzik

    September 10, 2009

    If anybody out there happens to be interested, the Oakland Athletics are on track to turn in their worst performance on the baseball field in more than 10 years.

    Normally, the travails of a losing ballclub wouldn't be of much note outside its hometown. But the A's aren't just any faltering ballclub.

    Over the last decade or so, the team and its general manager, Billy Beane, have been the leading exponents of a style of major league management known as "moneyball," after the title of Michael Lewis’ 2003 book about the franchise.

    Moneyball, in Lewis' formulation, meant the use of sophisticated statistical analysis to identify underappreciated -- and therefore underpriced -- players and assemble them into a successful team.

    This is a powerful concept in a sport where have-nots face the challenge of competing, on the field and in business, with permanent haves such as the New York Yankees. The A's, not exactly a small-market but certainly a low-budget organization, employed it to stay competitive for years at a time, reaching the playoffs five times since 1999 despite a payroll consistently ranking among the lowest in the game.

    This won't be one of those years. The team's won-lost mark of 62-77is tied for fourth worst in the American League.

    The A's performance presents an interesting question. Does it show the limitations of moneyball -- do the immutable economics of Major League Baseball, in which the rich get richer and leave the other teams in the dust, trump even the best statistical analysis?

    Or to put it another way, does it actually confirm the utility of moneyball -- are the A's behind the eight ball in part because every other team in the league is now using the same methods?

    Certainly the latter phenomenon is a real one.

    "Every team today has a guy doing this sort of analysis," Kevin Goldstein, a writer for the website Baseball Prospectus, told me recently. "Oakland originally was getting more bang for the buck, but players today are more properly valued as a whole. So it's harder to find diamonds in the rough."

    The Boston Red Sox went to the extent of adding Bill James, baseball's godfather of statistical analysis (whose annual 1982-88 Baseball Abstracts still occupy an honored place on my bookshelf), to their front office in 2003.

    He helped the Red Sox build the team that won the World Series in 2004, their first championship in 86 years, and again in 2007. The statistics-based insights popularized by James and his followers and absorbed by other teams included the importance of on-base percentage, the costliness of unsuccessful stolen-base attempts and the need to assess defensive skills in new ways.

    These concepts contributed, directly or indirectly, to the success of poverty-row teams like the Tampa Bay Rays, a gang of unheralded youngsters that snagged the American League pennant last year.

    The Rays' opening-day 2008 payroll, $43.8 million, ranked 29th among the 30 major league clubs and came to about 21% of the top spending club, the Yankees, who didn't make it into the playoffs.

    Many teams have gotten better at using statistical analysis to counterbalance the prevailing economic realities of the sport.

    Consider that the most important inflection point in the relationship between a young player and his club comes three years after his rookie year, when he becomes eligible for salary arbitration and his price can skyrocket -- as happened when an arbitrator jacked up the salary of Philadelphia Phillies slugger Ryan Howard to $10 million from $900,000 in 2008. (A player can become a free agent three years after that.)

    That means that if you can identify a promising talent ahead of your rivals and lock him into a long-term contract that moves the arbitration deadline off by a year or two, you can develop young players and keep your payroll under control for a long time.

    It's no mystery why there's been a surge in rich contracts for untested prospects -- Exhibit A being the four-year, $15-million contract the Washington Nationals gave their No. 1 draft pick this year, a 21-year-old fireballer from San Diego State named Stephen Strasburg.

    The deal will still give Strasburg two years of eligibility for salary arbitration, but the Nationals get four years to make him the centerpiece of their pitching rotation before an arbitrator gets a crack at him.

    "He's still a bargain if he's as good as they say," Goldstein says.

    Making such deals with amateurs puts a premium on high-grade analysis, though it has to be integrated with more traditional skills.

    "Moneyball works in giving you another set of tools," says Maury Brown, a veteran baseball analyst and founder of the Biz of Baseball website. "But you still need good scouting, and luck -- you have to draft well and get everyone hitting at the same time."

    Moneyball has never been a panacea for all the problems of a low-budget team. Its limitations were evident in Lewis' book, which made clear that although one can analyze a player's past performance, the very act of measuring, followed by laying the expectation on a player that he will continue to live up to or even surpass that measurement, affects his performance in unpredictable ways.

    The best example of that was Jeremy Brown, an A's first-round draft choice in 2002, who was so emblematic of Beane's quest for underappreciated talent that he got his own chapter in "Moneyball." Brown's major league career ultimately spanned all of five games in 2006. Shipped back to the minors and possibly undone by all the attention he'd received, he retired in 2008.

    The truth is that in baseball, as in any field with an infinite number of variables, sometimes it seems like you don't know nothin'. This year the Yankees, whose $201-million payroll outdistanced the second-highest spender by about 35%, rest securely atop the American League East. But the runner-up, the New York Mets (with a $149-million payroll), are 18 games out of first place in the National League East. Their playoff chances are effectively zero.

    The Chicago Cubs, ranked third in payroll, aren't a postseason factor this year. The Dodgers (No. 9 in payroll) are in first place in the National League West, but are feeling the hot breath of the surging Colorado Rockies (18th). In the American League West, the Angels (sixth in overall payroll), are struggling to hold off the Texas Rangers (22nd).

    As for the A's, whose $62-million opening-day payroll was fifth-stingiest in the league this year, no one expects them to remain doormats permanently, or even for much longer. The skill of Beane's staff at identifying young talent and drafting wisely has yielded a widely admired farm system, which could make the major league club a contender again very soon. On the other side of the coin, not even the Cubs' lavish payroll in recent years -- ranked ninth or better every year since 2004 -- has enabled them to win their first World Series since 1908 or first pennant since 1945.

    The performance of such small-market teams as the Rays, Rangers, Minnesota Twins and Florida Marlins, which are all making the most of limited resources this year, validates the premise of moneyball as a sound one. There's a lesson here that has had to be relearned periodically ever since David beat Goliath, which is that the key to winning a test of strength in which you're hopelessly outweighed is to be brainier than the other guy.

    But there may be a less uplifting lesson in the long-term success, whether on the field or at the turnstiles, of the dominant teams in baseball. In a contest of resources, the rich sometimes outlast the poor.

    The fate of teams dependent on moneyball may be limited to at-best-brief spells of success interspersed with periods of drought; teams with the determination to spend vast wads of cash have a much better chance of staying in the running pretty much all the time, as the Yankees, Red Sox and Angels prove. (Even the Mets, the free-spending sad sacks of 2009, placed first or second in their division in five of the previous 10 seasons.)

    "Without a high-priced payroll, you can't be expected to win year in, year out," baseball analyst Brown says.

    Or as Goldstein put it when I asked him what it would take for a small-market team to build a dynasty that would stand for years: "You better be stocked with geniuses."

    Michael Hiltzik's column appears Mondays and Thursdays. Reach him at michael.hiltzik@latimes.com, read previous columns at www.latimes.com/hiltzik, and follow @latimeshiltzik on Twitter.

    Copyright © 2009, The Los Angeles Times
    http://www.latimes.com/business/la-f...5,print.column

  • #2
    Sabermetrics does work, but maybe Beane isn't the best at scouting talent. He did say Fielder and Greinke weren't good picks in the 2002 draft.

    Comment


    • #3
      He said Fielder was really fat and that high school pitchers have a quarter of the chance that college hitters do to make the bigs.

      Read the book.

      Comment


      • #4
        Originally posted by Mainge View Post
        He said Fielder was really fat and that high school pitchers have a quarter of the chance that college hitters do to make the bigs.

        Read the book.
        You also look at the past drafts and Beane's best players have really just been Swisher and Street.

        Comment


        • #5
          ShakeN'Bake?

          Comment


          • #6
            He liked Youkilis but couldn't get him.
            This post was brought to you by: Dat SEC Speed

            Comment


            • #7
              As soon as team realized that the best players in baseball were those that made outs the most infrequently, this "under-appreciated" group of players ceased to exist. Other than that, it's simply paying for intangibles that inconsistently contribute to wins.

              IMO, the next revolution coincides with team completely disregarding the stolen base, and a stat head finding a way to translate stolen bases into true slugging percentage. Yes, it requires a threshold for successful SB% and a punishment for bad SB% but if SLG is just average based reached, and if a walk is good as a single, soon there will be some way to express that a single and a SB is just as good as a double, especially to lead off an inning. I have no idea how to begin to compute it, but as soon as teams stop paying for speed, this will certainly start to gain traction in the bang-for-the-buck community.

              Comment


              • #8
                Moneyball was completely about finding inequalities in the economic system in baseball. If college hitters are undervalued relative to high school pitchers, it makes more sense for a team that has limited resources to try and get the undervalued guy. But the individual parts that are most often attributed to beane and moneyball (defense is suck, speed is suck, yay walks, etc) aren't set in stone. If on base guys become overvalued and you can efficiently build a team around guys of a different sort, that's what you do.

                In theory, anyways. Obviously the most efficient players are ones that don't make outs, but they aren't being undervalued anymore. The thing to watch is how beane tries to find the next great undervalued guys. If he can, he'll continue to be successful.
                poop

                Comment


                • #9
                  A walk is not as good as a single.

                  A single is worth .5 runs. A walk is worth .34 runs.

                  A single and a SB (.18) is worth .68 runs, a double is worth .72 runs. And a CS is worth -.32 runs.

                  I don't think what you're saying really matters at all. There is already weights for things in the game. Adding SB to SLG% doesn't really do anything. At best it gives you what weights already tell you.

                  Comment


                  • #10
                    Originally posted by Bobbob1313 View Post
                    Moneyball was completely about finding inequalities in the economic system in baseball. If college hitters are undervalued relative to high school pitchers, it makes more sense for a team that has limited resources to try and get the undervalued guy. But the individual parts that are most often attributed to beane and moneyball (defense is suck, speed is suck, yay walks, etc) aren't set in stone. If on base guys become overvalued and you can efficiently build a team around guys of a different sort, that's what you do.

                    In theory, anyways. Obviously the most efficient players are ones that don't make outs, but they aren't being undervalued anymore. The thing to watch is how beane tries to find the next great undervalued guys. If he can, he'll continue to be successful.
                    He had some of the best defensive teams for several years running. So one could assume he had discovered the next inefficiency in the market and used it to get to the ALCS in 2006. The success wasn't as great as the Big Three days. He's never has been able to repeat the Big Three success (Harden was chronically hurt) and Eric Chavez fell off the face of the Earth.
                    Last edited by Party; 09-25-2009, 07:00 AM.

                    Comment


                    • #11
                      Originally posted by nny View Post
                      A walk is not as good as a single.

                      A single is worth .5 runs. A walk is worth .34 runs.


                      A single and a SB (.18) is worth .68 runs, a double is worth .72 runs. And a CS is worth -.32 runs.

                      I don't think what you're saying really matters at all. There is already weights for things in the game. Adding SB to SLG% doesn't really do anything. At best it gives you what weights already tell you.
                      This may be true in the aggregate, but I think that's a poor way of assigning value to any individual outcome. A walk and a single are equally valuable when you have no runners on base, for example. I think any system that simply multiples singles by some average value of runs is diluting the information too much. It's more work, but you have to take into consideration the in-game situation where the single or walk occurred.

                      Moreover, if you assign value based upon Win Percentage Added rather than mere average runs added, the situation becomes even more context dependent. A caught stealing may cost -.32 runs on average, but that says nothing about the particular instances when the player decided to steal and got caught. A CS by a speedster looking to up his SB totals when his team is down 7 and the opposing team isn't too concerned about a single additional run is no way as damaging as the guy who tests out the catcher's arm in a tie game, 1 out, in the bottom of the 8th and fails. Since we have the data to make these calculations, we might as well use it to give us a more nuanced and complete picture of the player's base-running ability (which is a function of their ability to read in-game situations and make the right decision).
                      Last edited by CrimsonCane; 09-26-2009, 09:56 AM.

                      Comment


                      • #12
                        Originally posted by CrimsonCane View Post
                        This is be true in the aggregate, but I think that's a poor way of assigning value to any individual outcome. A walk and a single are equally valuable when you have no runners on base, for example. I think any system that simply multiples singles by some average value of runs is diluting the information too much. It's more work, but you have to take into consideration the in-game situation where the single or walk occurred.

                        Moreover, if you assign value based upon Win Percentage Added rather than mere average runs added, the situation becomes even more context dependent. A caught stealing may cost -.32 runs on average, but that says nothing about the particular instances when the player decided to steal and got caught. A CS by a speedster looking to up his SB totals when his team is down 7 and the opposing team isn't too concerned about a single additional run is no way as damaging as the guy who tests out the catcher's arm in a tie game, 1 out, in the bottom of the 8th and fails. Since we have the data to make these calculations, we might as well use it to give us a more nuanced and complete picture of the player's base-running ability (which is a function of their ability to read in-game situations and make the right decision).
                        The values I listed weren't Pin Percentage Added but rather for individual players. For them you can't really deem it situational because the batter has no control over the situation, they only have control over if it's a single or a walk.

                        I certainly imagine WPA does it differently based on the situation but I never have looked into WPA.

                        Comment


                        • #13
                          Originally posted by nny View Post
                          The values I listed weren't Pin Percentage Added but rather for individual players. For them you can't really deem it situational because the batter has no control over the situation, they only have control over if it's a single or a walk.

                          I certainly imagine WPA does it differently based on the situation but I never have looked into WPA.
                          By that reasoning, you invalidate the basis of the original stats you posted. You can't develop an entirely context-independent statistic because there's no such thing as theoretical baseball. You're starting point always has to be real game data. A batter may not have control over his situation, but you can't simply say that, based on that, you can create a context-independent value based on an aggregation of situational events.

                          Comment


                          • #14
                            Originally posted by CrimsonCane View Post
                            By that reasoning, you invalidate the basis of the original stats you posted. You can't develop an entirely context-independent statistic because there's no such thing as theoretical baseball. You're starting point always has to be real game data. A batter may not have control over his situation, but you can't simply say that, based on that, you can create a context-independent value based on an aggregation of situational events.
                            all run stats are theoretical, so I'm not really sure what you're getting at. Both for hitting (RC, XR, so on) and pitching (FIP, tRA, so on). It's stripping things and trying to look at what only a player is responsible for. Are they perfect? No. There are little things like you mentioned. But they are just that, little things.

                            And swift's proposed stat is no different
                            Last edited by nny; 09-25-2009, 11:34 AM.

                            Comment


                            • #15
                              They are not theoretical. A single is worth X runs because an aggregation of hitting stats makes it so. Unless, I'm thinking of different values (I'm relying on Expected Runs Matrices).

                              To contrast, you can use math to actually define the value of different spaces in Monopoly. You don't actually need to have a large sample size of people playing Monopoly to do this either. The probability dictated by the rules of the game (2 6-sided dice) allows you to determine that number with no need for actual game data.

                              Baseball needs game data as a starting point for any statistic. Players are not dices with perfect probabilities.

                              Furthermore, your stats are also taking factors out of the players control into consideration. Albeit it standardizes for everyone. The value of a single or double in terms of runs is dependent on what happens, on average, after that players at-bat. If you don't take that in to consideration, you can never arrive at the stat. That's the point I was trying to get at.

                              Comment

                              Working...
                              X