Saturday, April 04, 2009

Team Predictions

There's a great new site to keep track of MLB depth charts.  Using the data here, and the Lineup Analyzer at Baseball Musings, I estimated the number of runs each team would score based on the player's PECOTA projections from Baseball Prospectus.  Obviously each team won't have this full starting lineup in place every night, but this is just a fun exercise.  Also, I used a regressed version of the output from the Lineup Analyzer.  David Pinto found that it overestimates runs somewhat, so the output was plugged into the regression formula 1.055x -.3231.

Having determined the Runs Scored (RS), I needed a source for Runs Against (RA) so I used the team RAs from Baseball Prospectus, but scaled them down to account for the fact that my cumulative RS was lower than BP's projected RA.

With RA and RS, I used the xlsSports MLB Monte Carlo simulator and ran 1,000 seasons.  I used a standard deviation of .01 in the AL and .02 in the NL, based on the variation in runs scored by various lineups.  Changing a lineup in the AL can make about a 2 win difference, but changing a lineup in the NL can make a 4 win difference. Here are the results, showing the W-L records for the playoff teams, and games back for everyone else.


AL East               AL Central          AL West         
Yankees    101-61     Indians    86-76    Athletics  84-78
Red Sox     97-65     Tigers         5    Angels         4
Rays            9     White Sox      9    Mariners       8
Blue Jays      25     Twins         12    Rangers       14
Orioles        34     Royals        12


NL East               NL Central          NL West         
Mets        87-75     Cubs       97-65    Dodgers    93-69
Braves          1     Brewers    88-74    D-backs        3
Phillies        1     Cardinals     14    Giants        17
Nationals       7     Reds          20    Padres        18
Marlins        17     Astros        25    Rockies       21
                      Pirates       34

Worth mentioning that I docked the Yankees a month's worth of A-Rod, and docked the Twins a month's worth of Mauer.

I was surprised at the Orioles' performance. I thought it would be better than the Blue Jays. I was also surprised at the Angels coming in below .500. In the NL, I'm not sure I'm buying the Nats finishing only 7 games and two games shy of .500. I'm also surprised the Rockies are worse than the Padres and Giants. The current Giants lineup is projected to be (by a long shot) the worst offensive team in baseball.

For the playoffs, this would give us the Red Sox and Brewers as the wild card teams, each of whom is in the same division as the team with the best record.

The Division Series would be Athletics at Yankees, and Red Sox at Indians in the AL. Mets at Cubs, and Brewers at Dodgers in the NL. In 50 simulations of these series (a lower number because I had to do them manually):

Yankees beat the A's 80% of the time, most often in a three game sweep.
Red Sox beat the Indians 70% of the time, most often in four games.
Cubs beat the Mets 60% of the time, most often in four games.
Dodgers beat the Brewers 64% of the time, most often in five games.

The ALCS would play out as Red Sox vs. Yankees, and Dodgers vs. Cubs. In 50 sims of those:

Yankees beat the Red Sox 58% of the time, most often in six games.
Cubs beat the Dodgers 54% of the time, most often in six games.

That gives us a Cubs-Yankees World Series. I ran the sims with the AL having the home field advantage, since it typically wins the All-Star game and is the stronger league right now. In 50 sims, the Yankees destroyed the Cubs, winning the series 74% of the time, most often in a full seven game series.

The sleeper teams are the Braves and Diamondbacks in the NL, and the Rays and Angels in the AL, if you can consider two teams from last year's playoffs "sleepers". For a real sleeper, see the Tigers.

The only teams with absolutely no chance of making the playoffs in the 1,000 season sim were the Orioles and the Pirates.