March 25, 2003 (973) 596-8392 September 29, 2003 -- Major League Baseball Standings National League East American League East Team Won Lost Pct. GB Team Won Lost Pct. GB Phillies 92 70 .568 - YANKEES 110 52 .679 - Braves 86 76 .531 6 Red Sox 106 56 .654 4 METS 84 78 .519 8 Blue Jays 81 81 .500 29 Marlins 76 86 .469 16 Orioles 65 97 .401 45 Expos 76 86 .469 16 Tampa Bay 47 115 .290 63 National League Central American League Central Team Won Lost Pct. GB Team Won Lost Pct. GB Astros 95 67 .586 - White Sox 90 72 .556 - Cards 88 74 .543 7 Twins 86 76 .531 4 Cubs 81 81 .500 14 Indians 70 92 .432 20 Pirates 73 89 .451 22 Royals 63 99 .389 27 Reds 71 91 .438 24 Tigers 58 104 .358 42 Brewers 59 103 .364 46 National League West American League West Team Won Lost Pct. GB Team Won Lost Pct. GB D'backs 102 60 .630 - Mariners 106 56 .654 - Giants 92 70 .568 10 Athletics 88 74 .543 18 Dodgers 84 78 .519 18 Angels 86 76 .531 20 Rockies 77 85 .475 25 Rangers 75 87 .463 21 Padres 64 98 .395 38 Mathematical Analysis Of The 2003 Baseball Season For the sixth year, Prof. Bruce Bukiet of the New Jersey Institute of Technology has used a mathematical model to predict the number of games that each Major League Baseball team can be expected to win this season. In the past, the method has performed as well as or better than the usual magazines that post their predictions each Spring. Bukiet originally designed the mathematical model to determine optimal batting orders, to see how much difference lineup makes in the number of wins a team can expect to obtain. The method also can be used to compute the influence that specific injuries or trades should have on the total number of wins. The Philadelphia Phillies, the Houston Astros and the Arizona D'backs should win their divisions in the National League East, Central and West respectively, with the San Francisco Giants repeating as the NL Wild Card team. In the American League, the Yankees, White Sox and Mariners should win their division titles while the Red Sox win the AL Wild Card. ``Once again, the Red Sox appear great on paper yet they somehow always seem to underperform. I haven't found a way yet to model the Red Sox' propensity to self-destruct late in the season'', says Bukiet. ``This is the sixth time we've run the models to predict an entire season,'' said Bukiet. ``The most important assumptions are that a player's performance can be based on his performance over the past several years and that the prospective players on each team will not change during the course of the season. We cannot predict injuries or trades.'' The "Markov Chain" method relies on the fact that there are a finite number of possible situations in a baseball game. A Markov process is a particular type of random analysis in which only the present state of the process is useful for predicting the future. Past history and the manner in which the process has evolved to the current state are both irrelevant. ``We designed two lineups for each team because of interleague play. One lineup features a designated hitter, and one lineup has the pitcher batting'' says Bukiet. Lineups were made by using lists of likely starters and lineups from sports magazines. ``We could have used the method to produce optimal lineups, but since managers don't set up their lineups this way, we used lineups similar to those used by most managers.'' As in the past, Bukiet considered how many games each team plays against each opponent and whether the game is at home or away. League home team advantages were based on last year's results. Then the probability of each team winning each particular matchup is computed. (Average team pitching was used since one cannot say who will pitch which game during the season). Adding up the probabilities for each of its games gives the expected number of wins. Even if the probabilities computed are exactly right, there should be some variation in the results, just as when flipping a coin 50 times, one will almost certainly not get exactly 25 heads and 25 tails. Bukiet began developing his baseball model in 1987 as a hobby. His findings have been published in the paper, "A Markov Chain Approach to Baseball," in the February 1997 issue of the journal Operations Research. His work has been featured on CNN Headline News, National Post (Canada), Star Ledger (New Jersey) and on radio. The original study found that the optimal spot for a pitcher in a batting order is seventh or eighth, putting him as far away as possible from the team's best hitter, traditionally used in the clean-up spot. The best hitter, by the way, should bat second, according to the computer model. Bukiet, has a doctorate in mathematics from the Courant Institute at New York University. NJIT is a public research university enrolling over 8,800 students. The university offers bachelor's, master's and doctoral degrees in 80 programs throughout its five colleges: Newark College of Engineering, New Jersey School of Architecture, College of Science and Liberal Arts, the School of Management and the Albert Dorman Honors College. Research initiatives include manufacturing, microelectronics, multimedia, transportation, computer science, solar astrophysics, environmental engineering and science, and architecture and building science. According to Yahoo! Internet Life magazine rankings, NJIT has been America's most wired public university for three consecutive years.