Using Sims to Win the ALCS



The following is an article posted to  that goes over the chances and probabilities of the Tampa Bay Rays moving onto their FIRST ever World Series.


Last week, we ran our Diamond Mind simulation and predicted the Tampa Bay Rays would easily defeat the Chicago White Sox, most likely in three or four games, and the Los Angeles Angels would upend the defending world champion Boston Red Sox … although that prediction came with the stipulation that the odds would swing in favor of Boston if it pulled off a Game 1 victory. In fact, the simulations showed the Red Sox’s odds of winning the series increased from 42.6 percent to 67 percent if they won the opening game — and that Jon Lester would be a factor in whether they did. Sure enough, the Rays did polish off the White Sox in short order, and the Red Sox rode Lester’s stellar pitching to a Game 1 victory and went on to take the series from the Angels.


About the simulation

The simulations were done using the Diamond Mind Baseball software, which was developed by renowned baseball statistics expert Tom Tippett. Diamond Mind is widely regarded as the most sophisticated and realistic baseball simulation software.

Diamond Mind is owned by Imagine Sports, a Silicon Valley-based Internet company that develops multiplayer online sports games, including Diamond Mind Online, an online baseball management game licensed and promoted by MLB Advanced Media. 


So what does our Diamond Mind simulation project for the American League Championship Series? Before revealing our projected results for the ALCS, here is a brief recap of our methodology:


    • We updated our projections and ratings for each player based on his 2008 regular-season performance.

      • Although playoff rosters were not yet finalized, we used our best judgment about whom would be selected, the starting rotations, the batting orders and the bullpen and bench roles.

      • We made judgments about players carrying injuries into the postseason. Beyond deciding, for purposes of setting roles, whether or not a player would play, we did not attempt to simulate how such an injury might hamper the player, except to the extent already reflected in his regular-season performance.

      • We then simulated the series 2,000 times.


    As unpredictable as the outcome of the regular season can be, luck can play an even greater role in a short series. Nevertheless, the results of our Diamond Mind software showed Tampa Bay winning 1,107 of the 2,000 simulations, a 55.4 percent probability that the Ray will be the new AL champions. This series looks to be higher scoring than we’ve projected for the National League Championship Series — the Rays averaged 4.7 runs per game in our simulations, while the Red Sox averaged 4.3 runs per game. The table below shows the number of times out of the 2,000 simulations that each team could win the series.

    ..TR> ..TR>


    Interestingly, the most likely scenario is that the series will go six games, not seven, regardless of which team wins.


    The difference? The Rays look like they’ll have a slight edge on the mound. While Boston’s bullpen generally outperformed Tampa’s in our simulations, the Rays’ young starters did much better than the Red Sox’s rotation. Tampa projected to get overall strong performances from Scott Kazmir, James Shields, and Matt Garza. With the exception of Lester, however, the Red Sox starters struggled: Josh Beckett, recently hampered by injury, posted an ERA of 4.50 for the simulation runs, and it got worse from there. Daisuke Matsuzaka had an ERA of 5.06, and he’s the starting pitcher for the critical first game.


    As for the hitters, Boston actually out-hit Tampa, posting a slightly better batting average and hitting significantly more home runs. Tampa’s edge, however, came from getting more baserunners — the Rays walked more, and on-base percentage is a big key to scoring runs. Tampa’s lineup also projected to share the load more evenly. Upton and Longoria led the Rays in getting on base (each averaged a .346 OBP in the simulations) and scored the most runs, while Cliff Floyd provided the most power (.500 slugging average) and led the team in RBIs along with Carlos Pena, who provided a high number of clutch hits.


    Boston was paced by Dustin Pedroia and Youkilis, who posted the best overall numbers of either team. Pedroia, who had been slumping before a big hit in Game 4 of the ALDS, led both teams in hits while posting an OPS of .846, showing he’s ready to break out of the slump. Youkilis, who also had a rough ALDS, led Boston in RBIs and runs scored and posted an OPS of .860, best of either team. Drew also chipped in with an OPS of .789 and some clutch home runs in some of the simulations. Other than some decent but unspectacular numbers from Bay, however, the rest of the Boston lineup did not contribute much, although Ortiz projected to contribute a bit more than he did in the ALDS.

    Looking for a surprise star … another David Eckstein, perhaps? The simulations say you might want to keep an eye on the catchers. Boston’s Jason Varitek, who mustered only three singles over the course of the ALDS, hit a surprising number of big home runs in the simulations, and Dioner Navarro stayed hot and led the Tampa starters in batting average and had a fair number of clutch hits in the 2,000 simulation runs, usually doubles. And if you really want a rabbit pulled out of the hat, the simulations showed surprising contributions from Alex Cora for Boston and Gabe Gross for Tampa, both coming off the bench.


    However, before you go putting your money on the Rays, there’s one huge caveat …


    As we noted in our division series previews, the odds can swing dramatically in favor of the winner of Game 1 of a short series. In fact, the winner of the first game went on to win every series in the 2007 playoffs, and the same is true so far in 2008. If the Red Sox win Game 1, watch out — the odds swing heavily in their favor, as was true of their series with the Angels. When the Red Sox won Game 1 in the simulations, they went on to win the series 1,220 times, or 61 percent. And this points to a key leverage point — if the Boston starting pitchers, particularly Matsuzaka and Beckett, perform better than the simulations projected, Boston is likely to take Game 1 and the series. That’s more likely to happen if the Red Sox can keep the Tampa hitters off the base paths. So key indicators of this series will be Tampa’s team on-base percentage and the control of the Boston pitchers.

    Another key leverage point you might want to watch are the home runs; our simulations show Boston might need to out-homer the Rays to win. That said, we still see the upstart Rays continuing their Cinderella season and advancing to their first World Series … but they’d better win that first game.





    TAM in 4 TAM in 5 TAM in 6 TAM in 7 BOS in 4 BOS in 5 BOS in 6 BOS in 7 Overall
    145 263 371 328 282 291 226 94 1,107-893 TAM


    LETS GO RAYS!!!!!!!!!!!! I’ll be rooting for you! Im tired of seeing the Red Sox all the time. I want a Tampa VS LA World Series. keep on truckin’!

    — Jenna (Thunder)

    I honestly think the next 2 games will tell alot about our chances. We have the knuckleballer tomorrow night, then we get Dice-K for the second time.
    It has been a wild time already in this ALCS series.
    I have a friend who plays for your cubbies..Chad Gaudin he loves it there because of the fans and the great food in the city.
    I hated to see the Cubbies out so early, but L A did a number on your pitching staff. Good thing is you have most of that great team back next season and can make another run at it all.

    Rays Renegade

    Leave a Reply

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

    You are commenting using your 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: