February 10, 2005

Prospectus Triple Play

Houston Astros, St. Louis Cardinals, Texas Rangers

by Baseball Prospectus

Printer-friendly

Houston Astros

• Refueled by the Rocket: The question of the offseason was answered in Houston on January 21, when Roger Clemens deigned to return to the Astros rotation for one year. Clemens signed the largest one-year deal in baseball history, one that will pay him the highest salary ever for a pitcher: $18 million.

  Earlier, Clemens had submitted the highest arbitration figure ever filed by a player: $22 million. The Astros had offered $13.5 million--the mid-point was $17.75 million, and once you're splitting the difference, why not kick in the extra to make it an even $18 million, right?

  So, is a pitcher who will turn 43 in August worth eighteen extra-large in 2005? Let's ask our trusty friend, PECOTA, which underestimated Clemens last season:

  
                           G  GS     IP    H  BB   SO  HR   ERA  VORP
  PECOTA 90th percentile  29  27  180.2  156  50  152  14  2.74  55.8
  PECOTA Weighted Mean    28  26  166.2  154  54  153  16  3.69  32.8
  
  Actual                  33  33  214.1  169  79  218  15  2.98  61.3
  

  We run these numbers not to give Nate Silver a hard time, but to make a point that came up a lot at our recent Pizza Feed in Walnut Creek, California. PECOTA operates by comparing players to other players throughout baseball history, and using what those historical players did to project the current player's future. The more comparable players PECOTA can find, and the more similar those comps are, the more accurate we can expect the projection to be.

  That's why someone like Clemens is such a tough slog. How many comparable players are there in the history of the game? The answer: almost none. Clemens' list of 2004 comps was headed by Gaylord Perry in 1980, when Perry was a 41-year-old starter for the Texas Rangers. Other pitchers who show up on Clemens' comps are late-stage Tom Seaver, Phil Niekro, Don Sutton, and Nolan Ryan. But the collective similarity of those comps is very low indeed--Clemens had a similarity index of 11 last year, on a scale where a score of 20 or lower indicates a historically unusual player. In fact, even Barry Bonds (similarity index of 21) was easier to find peers for than Clemens was last year.

  That's the long way of saying that we've never really been down this road, with a power pitcher coming off a Cy Young season in his 40s--it simply hasn't happened. If some guy comes along 20 years from now and does it, we'll have a good comp, but for now, PECOTA gets to wing it just a little. Here's what it sees for Clemens in 2005:

  
                           G  GS     IP    H  BB   SO  HR   ERA  VORP
  PECOTA Weighted Mean    30  30  191.3  173  63  174  21  3.79  36.6
  

  That projection is the 12th best VORP projected for any pitcher in baseball in 2005.

  Is 12th-best really worth $18 million a year? Probably not. Was it worth it to a Houston team with a rapidly-dwinding window to compete? Probably.

St. Louis Cardinals

• Genius?: That Tony LaRussa, he's been around a long time. This year will mark his tenth year at the helm of the Cards, and that's after years managing the White Sox and the A's before arriving by the Mississippi. He's moved up to seventh on the all-time list for managerial wins, and with 80 wins this year, he'll vault all the way to number three, behind only Connie Mack and John McGraw. This offseason, he signed a deal that will keep him in the dugout at Busch Stadium through the 2007 season.

  So, is the guy any good? How could we tell? What about looking at the difference between his team's projected wins and its actual wins? Wouldn't we expect that a great manager's teams would consistenly outperform their Pythagorean projection, owing to the effects of his acumen and wisdom?

  
  Year    Team     W   L   RS   RA   PytW   Delta  Finish
  
  1980    CHA     70  90  587  722    65      5       5
  1981    CHA     54  52  476  423    59     -5     3/6
  1982    CHA     87  75  786  710    89     -2       3
  1983    CHA     99  63  800  650    96      3       1
  1984    CHA     74  88  679  736    75     -1       6
  1985    CHA     85  77  736  720    83      2       3
  1987    OAK     81  81  806  789    83     -2       3
  1988    OAK    104  58  800  620    100     4    AL 1
  1989    OAK     99  63  712  576    97      2    WS 1
  1990    OAK    103  59  733  570    99      4    AL 1
  1991    OAK     84  78  760  773    79      5       4
  1992    OAK     96  66  745  672    89      7       1
  1993    OAK     68  94  715  846    69     -1       7
  1994    OAK     51  63  549  588    53     -2       2
  1995    OAK     67  77  730  761    69     -2       4
  1996    STL     88  74  759  706    86      2       1
  1997    STL     73  89  689  708    79     -6       4
  1998    STL     83  79  810  782    84     -1       3
  1999    STL     75  86  809  838    78     -3       4
  2000    STL     95  67  887  771    91      4       1
  2001    STL     93  69  814  684    94     -1       1
  2002    STL     97  65  787  648    95      2       1
  2003    STL     85  77  876  796    88     -3       3
  2004    STL    105  57  855  659    100     5    NL 1
  

  We're only including full seasons, so 1986, when LaRussa split time between Chicago and Oakland is left out. PytW is the wins predicted by the Pythagorean Theorem, using 1.83 as the exponent.

  That's 10 division titles, four pennants and one world title in 24 full seasons of filling out lineup cards and calling sac bunts. Over the course of all those thousands of games, LaRussa's teams have, in fact, outperformed their Pythagorean projections...by a grand total of 16 wins. Which, frankly, isn't that many.

  Now, before anyone gets all hysterical, we'll admit this is really a toy, and not a way of rigorously analyzing the effect of a manager. But it does illustrate a point: the genius manager might be the most overstated cliché in the game.

  At this point, we know the optimal strategy in many game situations--it's hard to imagine a manager gaining a huge edge there. The idea of the Wizard of Oz pulling levers and besting his counterpart in a battle of wits is simply outdated, Lloyd McClendon aside. This doesn't mean that some managers don't do a better job of game management than others (Dusty Baker, I'm looking at you, dude), and certainly, the managment of a pitching staff is still a big deal (you again, Dusty).

  What statistics can't capture is what is increasingly the real job of a manager: dealing with people. This is the era of manager as soother, facilitator, confidant, ego-stroker and vibe-generator. Those aren't things that you can measure or quantify. Perhaps when it comes to managers, intangibles are what matter.

Texas Rangers

• Picking Pedro: The Rangers have languished for years with an ineffective starting pitching staff, although it has often looked worse than it is due to the park factors in Arlington. Texas fans have pined for a starter to set things right, and the team has delivered, with such contracts as the disaster lavished upon Chan Ho Park and the seemingly-endless number of stints that Kenny Rogers has spent with the club.

  But this year. This year is different. This year the Rangers have brought on a pitcher who will strike fear into the hearts of opposing hitters--Pedro Astacio.

  When we last left Mr. Astacio, he was busy racking up a 10.38 ERA in 8 2/3 innings for the eventual World Champs in Boston last last season. That was his comeback from a June 2003 shoulder surgery.

  PECOTA isn't optimistic, foretelling a 5.34 ERA and a 6.7 VORP for the 35-year-old Astacio. Even the most optmistic Rangers fan would have to be disapointed if this is the big move of the offseason--Pedro hasn't thrown 200 innings in a season since 1999, and even in his last full season, with the Mets in 2002, his 12-11 record was only good for an underwhelming 9.2 VORP, or just a shade under one win over replacement.

  Of course, every inning that Astacio manages to throw is one fewer thrown by Park, so there is an upside.

• No Way, Jose: This past Sunday, the New York Daily News published what it said were details from Jose Canseco's upcoming book, Juiced: Wild Times, Rampant 'Roids, Smash Hits, and How Baseball Got Big. Given the current media climate around the issue of steroids, reaction was typically muted. Canseco, a man who might seem to be a little short on credibility, even forced the President of the United States to issue a denial, stating that, contrary to Canseco's claim, he wasn't aware of any steroid use during his tenure as the owner of the Rangers.

  According to the Daily News, Canseco also claims in the book to have introduced the Rangers to steroids when he arrived in Texas in 1992, including his then-teammates Juan Gonzalez, Rafael Palmeiro, and Ivan Rodriguez. Is there any evidence in their performance?

  
  Palmeiro:
  
  AGE YEAR    BA    OBP    SLG
  26  1991  .322   .389   .532
  27  1992  .268   .352   .434
  28  1993  .295   .371   .554
  29  1994  .319   .392   .550
  
  Rodriguez:
  
  AGE YEAR    BA    OBP    SLG
  19  1991   .264  .276   .354
  20  1992   .260  .300   .360
  21  1993   .273  .315   .412
  22  1994   .298  .360   .488
  
  Gonzalez:
  
  AGE YEAR    BA    OBP    SLG
  21  1991  .264   .321   .479
  22  1992  .260   .304   .529
  23  1993  .310   .368   .632
  24  1994  .275   .330   .472
  

  Eyeballing those numbers, there's nothing that stands out and screams, "He started juicing!" The only year that seems like a quantum leap is Juan Gone in 1993, but there is a steady build to that peak before the dropoff in 1994.

  All three players have denied the reported allegations. We're into Will Carroll territory here, but it seems prudent to reinforce a couple of points. First, players are innocent until proven guilty. Secondly, for good or ill, steroids weren't against the rules of Major League Baseball during Canseco's career. And perhaps most importantly, there isn't any study out there that proves that steroids help players produce more offensively. That doesn't mean that steroids don't help; that means that we don't know that they do. We try to trade in data around here and not conjecture, and we're hoping that people who read Canseco's book will do the same.

--Mark McClusky