Who should win Major League Baseball's Most Valuable Player Award and Cy Young Award in 2007

Prof. Bruce Bukiet, Mathematical Sciences, NJIT

Kevin Fritz, Hillsborough, NJ

(also check out the 2007 postseason page)

go to Prof. Bukiet's baseball page

go to egrandslam.com

Thanks for visiting my 2007 MVP -- Cy Young Award page. Please scroll down to see a table of the results.

How we define who deserves the MVP and Cy Young Awards

We define the player who deserves these awards as the player in each league who would have taken a team made up of all average players at each position and given that team the most extra wins (above 81-81) if he replaced the average player at his position in the lineup.

Who was considered

We considered all players who placed in the top five in their league in at least one of the following categories:

 How we set things up and perform the computation

A lineup of average players at each position is set up for each league.

NL -- Shortstop, Outfield, Outfield, First Base, Outfield, Third Base, Catcher, Second Base, 9th position (pitchers and pinch hitters)

AL -- Shortstop, Outfield, Outfield, First Base, Designated Hitter, Outfield, Third Base, Catcher, Second Base

The data used is:

    For hitters: at bats, hits, doubles, triples, home runs, walks, stolen bases and times caught attempting a stolen base.

    For pitchers, innings pitched, hits allowed, walks given up and home runs allowed.

For each MVP candidate a similar lineup is set up with the appropriate batter replacing the line with the average performance at his position.

The lineups is adjusted to test the candidate at 2nd, 3rd and 4th in the lineup.

The player being tested has his data averaged (weighted) with the data of the average player at his position based on how many games the player played.

(For example, if a player plays only 81 games then we'd use his performance and that of average players in each amounts).

We then run the code which computes the run distribution expected from each lineup and from that compute the probability of each team winning a game.

The code employs the Markov process model for baseball developed by Prof. Bukiet for his 1997 Operations Research paper.

Multiplying by 162 gives the expected number of wins for this lineup against a team of all average players (not including the player being tested).

To consider pitching performance, we adjust the opposition batters' performance in accordance with how well each pitcher keeps runners off the bases compared to the average pitcher. Of course, the pitcher's performance must be combined in a weighted average with average pitchers taking into account how many innings a given pitcher pitched during the season. (If a pitcher pitched 10% of his team's innings, we take 10% his data plus 90% average data).

The upshot is that an outfielder is compared to other outfielders, DHs compared to DHs etc.

One note: for average position data at each position we used 2006 data

What the Method Does Not Include

The computation does not include various aspects of the game, most of which are quite difficult to quantify in an objective manner, including fielding ability, clutch hitting ability, the fielding ability of the players playing behind a pitcher, leadership ability, heart, who typically bats ahead of or behind the batter being considered.

The Results

For hitters, we present the result at the place in the lineup the batter would have been most valuable (of 2nd, 3rd or 4th).

(Note: the model tends to reward good values of WHIP, innings pitched - so closers might be underrated.)

American League (Hitters in Green, Pitchers in Red -- top several listed)

National League (Hitters in Green, Pitchers in Red -- top several listed)

Thanks for visiting this page. Please feel free to contact me at bukiet@m.njit.edu