Sunday, February 12, 2006

2006 predictions

Now that most of the major free agents have signed, it's time to turn the prediction-maker on and look at the MLB in 2006. I'll start out by telling you how I go about doing this:

  • The statistical metric I use is Win Shares. It's a Bill James stat that takes a team's wins (100, say) multiplies by 3, and distributes those Win Shares among the players on the team. 1 win = 3 Win Shares (WS, for short).
  • Then I make out a tentative lineup for a team for next season. This lineup includes the 8 starters (9 in the AL with the DH), 5 starting pitchers, and closer. I make an educated guess (the best I can) how many Win Shares the player will produce in the coming year, taking into account as many factors as possible. I determine how many Win Shares these players "will" produce in 2006 and compare it to the number of WS the team got from the same positions in 2005. For example, the 2005 Braves got 199 Win Shares from their 8 starters, 5 starting pitchers, and closer. By my prediction, the 2006 Braves will get 203 Win Shares from these same positions. This means that next year's Braves will have 4 extra Win Shares, or roughly 1 extra win. So my rough prediction for the 2006 Braves is 91 wins, 1 more than their 90 last season. Taking into account all the other variables, I create a 5-win range for the Braves. So while 91 wins is my official prediction, there's little chance they will settle on exactly that many wins. What I will predict with greater accuracy is that the Braves will win between 89-93 games. This takes into account many other unforeseen factors. This, then, is the core of my predictions.

Now I'll tell you everything that my predicting system does take into account:

  • Difference in league or ballpark. Win Shares adjust a player's reward for the park in which their games are played. So, unlike batting average or slugging percentage, you're not going to have everything thrown off by Coors Field on one end and Petco Park (San Diego) on the other. A good example is Luis Castillo. Castillo is a second baseman who spent last year with the Marlins, playing one of the more pitcher-friendly parks in baseball, Dolphins Stadium. This year, he's been traded to the Twins. So not only is he playing in a more hitter-friendly league, he's playing in a more hitter-friendly ballpark, even in relation to the league. So we would expect Castillo's raw batting stats (AVG, SLG, etc.) to go up drastically next year, even if his rough level of talent and production is the same. WS, however, is not fooled. Luis Castillo is who he is, wherever he plays.
  • Luck (to a certain degree). My system factors out some of the luck inherent in baseball. I may, in my predictions, call for another 28 Win Shares for Brian Roberts, since he's only 28 years old and not inclined to suffer a serious decline in quality this year. But the realist in me says that, considering Roberts' otherwise unspectacular career, that he was amazingly lucky in 2005. So my prediction will be more in line with his career numbers, although somewhat higher to reflect the remote possibility that he did just suddenly get better at his age (it really never happens, except, of course, when it does). I've also looked at some more sophisticated stats in The Hardball Times Baseball Annual which try to predict which players got lucky (and unlucky) in 2005. Mark Ellis, of the Athletics, had an excellent 2005, hitting 384/477/316 with good defense, producing 21 WS. But the Annual cites Ellis as, according to their PrOPS stat, the luckiest hitter in baseball last year. This makes perfect sense because Ellis, like Roberts, has never been this good before, and likely just had a career year last season. The Annual also constructs a similar table for entire teams.
  • Variations in single statistics. WS are nice, because they help us keep sight of a player's ultimate contribution, placed in one bottom line number. So we aren't distracted by Andruw Jones' career-high 51 HR in 2005, since we know that those HR did not make him significantly more valuable, nor did they represent anything but what is likely a career year. Andruw did not likely just wake up one day with extra home run power, and history suggests that his power will disappear just as suddenly as it appeared, unless Andruw is a once-in-a-lifetime player or taking steroids (both of which I doubt).

And now that things that my system does not take into account:

  • Luck (in every other sense). One of the drawbacks of Win Shares is also its strength; it inexorably ties a team's performance to the players'. The sum of the parts is always equal to the whole in Win Shares. In reality, this isn't always so. If a team gets exceptionally lucky one season, Win Shares isn't going to notice. It will simply divide the luck in 3 and evenly distribute it to every player. Not only is luck never evenly distributed, but it's also just luck, so it's useless when predicting the future. So when I'm determining a team's 2006 record, I'm assuming that the team that won 90 games last year really was good enough to win 90 games, and not just a lucky team. Because a 90-win team with a lot of luck can turn into an 80-win team without it, making an unwitting predicter look stupid.
  • Regression to the mean (on the team level). Regression to the mean is a fairly complex name for a simple phenomenon; performances to any extreme (good or bad) tend to trend back toward the average over time. This makes sense; a player who hits .380 over the first half of the season is more likely to get closer to the average as the season goes on. It's easier to hit 50 HR in one season than it is to hit 500 in your career. So a player's performance to any extreme (good or bad) needs to be adjusted back to the average due to regression to the mean. I've adjusted for this on a player-by-player basis, but I haven't on the team level.
  • Bullpens. I mentioned before that I only use the numbers of the 8 positions players, 5 starting pitchers, and a closer. This assumes that the bullpen doesn't change from year to year, which it can and will. A team's bench players make a relatively small contribution; and the difference between the best and worst pinch-hitter is truly marginal. The same cannot be said of middle relievers. Middle relievers play an increasingly important part in the game, and their performance can add (or subtract) several wins to a team's total. So why don't I measure them? Why don't I include middle relievers? 2 reasons: 1) I simply don't have the knowledge and 2) neither does anyone. I pride myself on being fairly familiar with every regular in the major leagues. And while I can recognize the names of most middle relievers in baseball, I'm not nearly qualified to tell you what they did last season, let alone predict what they'll do next season. On the second point, relief pitchers are the most volatile "stocks" in baseball; they go from good to bad fairly frequently, moreso than other players, and no one really knows why. So if Bill James can't accurately predict what Julian Tavarez will do next year, I sure as hell ain't going to try.
  • Intangibles. I often use this word derisively, but there are simply many things that affect a team's W-L record that my system doesn't take into account. For example, the Reds recently changed owners, General Managers, and had their front office effectively gutted. This will have a potentially large impact on the team's record as soon as this year, but no one can predict what.
  • Platoons/Changes in roster. My system assumes that one person is the sole inhabitant of a position, which never happens. I also don't/can't take into account how the team will change over the course of the season. Trades will be made, some with a lot of impact, and there are a couple free agents still out there.
  • The basic answer is that while I'm not great at predicting the future, neither is anyone in baseball. It's why they play the games.

I've spent so much time qualifying my predictions that the final ones seem anticlimactic. Before I make my final calls, I'll share with you the rough answers I got when I did the math on Win Shares. These aren't my "final" answers, but this is what my numbers told me (with a brief explanation):

Baltimore: 74 wins (72-76 win range)
2005 wins: 74
The Oriole lineup hasn't gotten significantly better, and their pitching only marginally so. This is a bit tough on them, but they're in a tough division.
Boston: 96 wins (94-98 win range)
2005 wins: 95
The Red Sox are losing a good deal but also gaining a good deal. The Red Sox have several volatile players, players who could be either good or bad, and their performance will be telling. This was also true last year, when the awfulness and injuries to Schilling and Foulke torpedoed a potential World Champion.
New York: 95 wins (93-97 win range)
2005 wins: 95
Yeah, they've still got a good lineup, but they're sporting the exact same pitching staff they trotted out a year ago, except that Randy Johnson is now 41 (forty-one) years old. The difference between the Red Sox and Yankees is marginal; any number of unforeseen factors could determine the AL East race; Theo Epstein could get pneumonia and lose the division to the Yankees.
Tampa Bay: 67 wins (65-69 win range)
2005 wins: 67
I guess it's a bit of a cop-out to predict the same number of wins as last year. But the D-Rays haven't gotten much better. The x-factor that I don't take into account is a fairly large group of young hitting prospects. Will they get to the majors in time to make big impact this year? Who knows?
Toronto: 82 wins (80-84 win range)
2005 wins: 80
Again, you see how conservative I am in my predictions. Realistically, I think the Blue Jays will be better than even 83 wins. But they're not the contenders everyone else thinks they are; they've still got a truly unimpressive outfield, a hole at second base, an unpredictable rotation, and despite all the trades and shenanigans, only marginally better production at 1B-3B-DH. The Jays' hopes rely on the health of ace Roy Halladay and volatile (and expensive) reliever B.J. Ryan.

1. Boston – 96 (94-98)
2. New York* – 95 (93-97)
3. Toronto – 82 (80-84)
4. Baltimore – 74 (72-76)
5. Tampa Bay – 67 (65-69)
* --
represents Wild Card

Chicago: 93 wins (91-95 win range)
2005 wins: 99
The White Sox were quite the lucky team in 2005, as I don't think they were really as good as 99 wins, their World Series sweep notwithstanding. GM Kenny Williams has taken good steps to make a good team, rather than rolling the dice on the same unimpressive 2005 lineup. They're a better team in a very competitive AL Central.
Cleveland: 91 wins (89-93 win range)
2005 wins: 93
The Indians were relatively unlucky in 2005, so one would expect a rebound in 2006. But along with Regression to the Mean (R2M in the future), the Indians had several good performances that may not be reproduced in 2006. They've also lost starting pitcher Kevin Millwood, who had an excellent year for them in 2005. They replaced him with Paul Byrd, who is good, but represents a rotation that is, on the whole, unreliable. Give me Chicago's pitching with Cleveland's hitting, and you've got a World Champion.
Detroit: 77 wins (75-79 win range)
2005 wins: 71
R2M is mainly why the Tigers are getting better. Remember the movie Weekend at Bernie's, where two guys tried to pass of a dead guy as alive so they could have a party? I often think of the Tigers as baseball's equivalent. They're not obviously
dead and festering, like the Royals. But they're an essentially lifeless and dying organization with a pretty poor future. The Placido Polanco trade is the only faint pulse that Bernie might not be quite dead.
Kansas City: 60 wins (58-62 win range)
2005 wins: 56
The aforementioned festering corpse. They moved up to 60 wins because it's just as hard to lose 106 games 2 years in a row as it is to win
Minnesota: 92 (90-94 win range)
2005 wins: 83
The Twins were a terrible-hitting team in 2005. Several players (such as Justin Morneau) simply must get better, and the trade for Luis Castillo also helps fill a gaping hole. The AL Central should be an exciting 3-way race, as my win range would indicate.

1. Chicago – 93 (91-95)
2. Minnesota – 92 (90-94)
3. Cleveland – 91 (89-93)
4. Detroit – 77 (75-79)
5. Kansas City – 60 (58-62)

Los Angeles Angels of Anaheim in the USA in the Western Hemisphere of Earth in the Milky Way: 88 (90-92 win range)
2005 wins: 95
The Angels have been surprisingly silent this off-season, losing key pitchers as well as catcher Bengie Molina to free agency. They've got a stacked deck of young hitting prospects, but how many of them are major-league ready? And who will pitch? It's doubtful they'll be anything better than 2nd-best to the rebuilt Oakland A's.
Oakland: 96 (94-98 win range)
2005 wins: 87
The A's also dealt with some bad luck in 2004, but have done a good job of correcting it. They're a better team all-around and, by my win range, the best team in baseball going into the season.
Seattle: 79 (77-81 win range)
2005 wins: 69
The Mariners should rebound due to R2M and the addition a some free agents. They're not an awful team, but they're still the 4th-best team in a 4-team division.
Texas: 79 (77-81 win range)
2005 wins: 79
Intuitively, I think the Rangers are a better team than this. But the stats are informative; their outfield is better, yes, but their infield is worse, and their pitching staff is as unpredictable as ever. They could put it all together and make a run at 90 wins, but only if the pitching gods smile on them. And the pitching gods have never smiled in Arlington.

1. Oakland – 96 (94-98)
2. Los Angeles – 88 (86-90)
3. Texas – 79 (77-81)
4. Seattle – 79 (77-81)

Atlanta: 91 (89-93 win range)
2005 wins: 90
If ever a team deserved the benefit of the doubt, it's the Braves. They shouldn't have won the division about 5 times the past few years, but they always did. The loss of Leo Mazzone, however, could potentially be a big crack in the foundation. But the Braves can also take comfort in the fact that they had a lot of great rookies in 2005; rookies who are only likely to get better in 2006 (but Jeff Francouer is not
the next Dale Murphy).
Florida: 59 wins (57-61 win range)
2005 wins: 83
The only regulars returning from the 83-win 2005 team are outfielder/potential MVP Miguel Cabrera and potential Cy Young-winner Dontrelle Willis. Together, they account for about 25 wins or so. That leaves a cast of unproven rookies and humpties to fill out the lineup and play minor league ball in the highly-competitive NL East.
New York: 88 wins (86-90 win range)
2005 wins: 83
You could certainly argue that I've given the Mets short shrift here. I do think they've improved by more than 5 wins this off-season, but not a whole lot more. They've still got a lot of problems. Their pitching staff is held together by Pedro Martinez, Tom Glavine, some scotch tape, and a lot of hope. Still, they could very well be the team to dethrone the Braves in 2006.
Philadelphia: 84 wins (82-86 win range)
2005 wins: 89
The Phillies got left out of the off-season free agent love-in, making them a worse team than they were last year. They too have a potent offense, but a pitching staff worthy of Major League.
Washington: 81 wins (79-83 win range)
2005 wins: 81
It took a lot of energy for the Nationals to end up back where they started. That's the story of GM Jim Bowden. He does a lot and makes a lot of headlines, but it's uncertain whether he actually, you know, helps his team win more games.

1. Atlanta – 91 (89-93)
2. New York* – 88 (86-90)
3. Philadelphia – 84 (82-86)
4. Washington – 81 (79-83)
5. Florida – 59 (57 – 61)

Chicago: 73 wins (71-75 win range)
2005 wins: 79
The Cubs did very little this off-season, and they're back in the familiar position of pinning all their hopes upon the fragile arms of Mark Prior and Kerry Wood. The Cubs will get better production with Juan Pierre in center, and might possibly get better production from Jacque Jones in right (Jones is replacing Jeromy Burnitz, who is a better hitter, if older). What dooms the Cubs is the fact that there's no way Derrek Lee will reproduce his memorable uber-career year in 2005, where he produced 37 Win Shares. 23 Win Shares would be a good guess for Lee, meaning the Cubs are losing about 5 wins right off the top.
Cincinnati: 74 wins (72-76 win range)
2005 wins: 73
The Reds have good hitting, but their pitching has to get a lot better for them to even be remotely competitive. Boy, if I wrote the same season preview for the Reds every year, would anyone notice?
Houston: 79 wins (77-81 win range)
2005 wins: 89
The Astros were another exceptionally lucky team in 2005. Their offense is awful, and despite the addition of Preston Wilson, it's the same bunch of guys this year, only older. Oh, and they lost their best pitcher. Even if Roger Clemens does come back, it will only be for a few months, and nowhere near as good as he was last year.
Milwaukee: 81 wins (79-83 win range)
2005 wins: 81
The Brewers' hitting prospects should only get better in 2006, but there's still a big question mark surrounding their pitching staff, especially concerning the health of ace Ben Sheets. Even considering this, I think the Brewers are, at this point, the second-best team in the division.
Pittsburgh: 76 wins (74-78 win range)
2005 wins: 67
Even now, staring at the numbers, it's impossible for me to believe that the Pirates will win more than 70 games next year.
St. Louis Cardinals: 90 wins (88-92 win range)
2005 wins: 100
The Cardinals are a significantly worse team than they were last year. They've lost a good chunk of production from Reggie Sanders and Larry Walker, replacing them with bargain-bin finds like Larry Bigbie and Juan Encarnacion. They should get better production from 3B Scott Rolen, but they lost a good 2B in Mark Grudzielanek and will deal with David Eckstein's inevitable decline.

1. St. Louis – 90 (88-92)
2. Milwaukee – 81 (79-83)
3. Houston – 79 (77-81)
4. Pittsburgh – 76 (74-78)
5. Cincinnati – 74 (72-76)
6. Chicago** – 73 (71-75)
** -- keep in mind that these are "rough" predictions. I can't really believe that the Cubs are any worse than the Pirates or Reds.

Arizona: 75 wins (73-77 win range)
2005 wins: 77
The Diamondbacks were the luckiest team by far in baseball in 2005. So while I don't think they're significantly worse than they were last year (as my prediction suggests) they are, realistically, even worse than that. They've got a lot of prospects, but few have so much as made an appearance in the majors.
Colorado: 71 wins (69-73 win range)
2005 wins: 67
It's too bad the Rockies aren't near a large body of water, because boy, are they treading it.
Los Angeles: 86 wins (84-88 win range)
2005 wins: 71
The Dodgers had so many injury problems in 2005 that they can't help but get better. They're not a great team, but I think they're easily good enough to win the West, in spite of their management.
San Diego: 79 wins (77-81 win range)
2005 wins: 82
The Padres spent most of their off-season energies re-signing their free agents. So after all that effort, they're the same mediocre team they were last year. But whereas the rest of the division has gotten better, they have not. They decided that one of baseball's best second baseman, Mark Loretta (in terms of bang for your buck), wasn't as valuable as Boston's backup catcher, Doug Mirabelli, so they traded for him. Kevin Towers, you surround your brilliance with stupidity in a curious way.
San Francisco: 74 wins (72-76 win range)
2005 wins: 75
If you thought the Giants were an old, beat-up team last year, well they're even older this year. This is what the Giants are without Barry Bonds and his 50 Win Shares. And even if the Giants get Bonds back (for a MAXIMUM of 140 games), the odds that it will be the same, Ruthian Barry are remote. He's 41 years old now, and several knee surgeries removed from his last MVP season. Even steroids can't save him now.

You'll notice that I didn't term these my "final" predictions. I'm saving those for some more time for reflection, as well as a chance to conider the "bullshit" factor. This is a sense of realism that needs to be applied to all statistics to avoid following mistaken statistics into a stupid decision. It's not an excuse to discard your logic for your "gut instinct," but a chance to combine the two for, hopefully, a more reasonable conclusion.

More to come later . . .

No comments: