In George We Trust
A(nother) blog about the most storied franchise in sports
Friday, February 10, 2006
'Tis the season for projections
Spring training is so close you can practically smell the fresh cut grass and hear the popping of the catcher's mitt. It's that time of year when experts and amateurs alike put their 2 cents in and try to predict the future, so here we go:

SG at Replacement Level Yankees Weblog went through the Diamond Mind Baseball simulation motions and projects the Yankees to post 85 wins in 2006.

Steve Lombardi over at WasWatching.com got the projection bug about a week later, and projected the Yanks to total 87 wins in 2006.

Rob LaBrie of SportsColumn.com fame is thinking 90 wins for 2006.

Tom Singer at MLB.com doesn't give a number, but he sees a second place finish for the Yanks in 2006, with the Blue Jays taking the division.

With that, I thought I'd take a crack at 2006 projections. I'll cut to the chase and give you my projection now, with the explanation of how I got to that number to follow.

IGWT's Yankees 2006 projected win total: 98

That's a better win total than 85 or 87 (if you're a Yankee fan), and here's how I came to that number:
  1. First, using Baseball-Reference.com, I took all the Wins, Runs Scored, and Runs Allowed data for every AL team since the last expansion in 1998 and tabulated it to get each team's run differential (that's 112 data points total).
  2. I plotted each team's number of wins versus it's run differential, and got this fancy little chart (sorry if can't read it, Blogger isn't very friendly when it comes to inserting tables into posts): The 2 high points are the 1998 Yankees and the 2001 Mariners, while the 2003 Tigers get the joy of having the lowest point on the chart. Obviously, there's a relationship between Run Differential and Wins, a relationship I used in this projection.
  3. I added a trendline to chart in order to get the equation of the line, which describes the Run Differential-Wins relationship (it's about time I put that Civil Engineering degree on working to some good use). Instead of getting one equation and therefore only one relationship, I used 7 trendlines and got 7 different relationships (for the geeks like me out there: I ended up with a linear, exponential, and 2nd through 6th order polynomial equations for the relationship).
  4. With the relationships between Wins and Run Differential at hand, all I needed were projected values for how many runs the Yankees would score as well as give up. Using RotoChamp.com's 2006 fantasy projections as well as BaseballThinkFactory.org's 2006 ZiPS projections, I totaled how many runs they expect the Yankees to score and surrender next year. Here's that data:
    RotoChampZiPS
    Runs Scored

    776

    		818
    Runs Allowed630630
    Run Differential+146+188
    It's definitely fair to say the Runs Scored/Allowed values are low, but they aren't crucial. It's the Run Differential that'll control this project, and those values are very reasonable. It's kind of interesting how both project the Yanks to surrender the same number of runs though. Hmmm...
  5. Using the aforementioned relationships and the projected data above, I calculated how many wins each equation predicts for 2006. Here that is:
    RotoChamp WinsZiPS Wins
    Equation 1

    95.970

    		100.267
    Equation 296.470101.852
    Equation 395.999100.366
    Equation 4

    96.014

    		100.385
    Equation 596.023100.383
    Equation 695.99199.297
    Equation 796.994100.247
  6. Finally, I took the average of those 14 projected totals, rounded it off to a whole number, and walla, 98 wins.

Hopefully I didn't turn too many of you off from the blog with all the geek talk, but hopefully I nailed the prediction and we're headed for a 98 wins season, which should be good enough to take the division. Again.

Written at 8:01 AM by Mike A.
1 comments
1 Comments:
Anonymous Anonymous said...
Looks to me that the regressions are about right. If all are healthy, 975-1000 runs, 98-104 wins, and 27 ring are all in reach.
4:45 AM  
Post a Comment
<< Home
Some rights of this page's plain text stuffs are reserved for the author.
The Template is generated via PsycHo and is Licensed.