beorach Posted October 26, 2016 Share Posted October 26, 2016 First, I have to thank Moiraine for contributing to this weekly post. If you're confused about what the z-score is, you can look back on the old post for my long-winded explanation. The short of it is I'm calculating as many z-scores as I can make sense of from within nine statistical categories and then averaging them to develop rankings. A z-score is the difference between a stat and the mean for a stat, expressed in terms of standard deviations. If your team puts up a stat that's exactly one standard deviation higher than the average, the z-score will be 1 for that stat. I average the z-scores within a category to share a composite one. I then average those composites to come up with aggregate composites (for lack of a better term). E.g., I'm taking rushing offense, passing offense, scoring offense, and total offense average z-scores and averaging them to come up with offensive rankings. I think people get it by now so I won't go on... B1G Overall (not weighted) Michigan 1.76 Ohio State 1.48 Nebraska 0.66 Wisconsin 0.43 Iowa 0.41 Minnesota 0.34 Maryland 0.32 Northwestern 0.11 Indiana 0.09 Penn State -0.03 Michigan State -0.36 Illinois -0.57 Purdue -0.76 Rutgers -1.17 B1G Overall (weighted for strength of schedule*) Michigan 1.51 Ohio State 1.18 Nebraska 0.47 Wisconsin 0.42 Iowa 0.27 Minnesota 0.24 Maryland 0.18 Northwestern 0.09 Indiana 0.07 Penn State -0.03 Michigan State -0.29 Illinois -0.40 Purdue -0.45 Rutgers -1.00 B1G Defensively (not weighted) Michigan 2.30 Ohio State 1.79 Wisconsin 1.41 Nebraska 0.90 Minnesota 0.88 Iowa 0.73 Indiana 0.61 Maryland 0.59 Northwestern 0.25 Penn State 0.18 Michigan State -0.31 Illinois -0.60 Purdue -0.78 Rutgers -1.03 B1G Offensively (not weighted) Michigan 1.45 Ohio State 1.28 Nebraska 0.45 Maryland 0.12 Iowa -0.03 Northwestern -0.08 Penn State -0.16 Minnesota -0.33 Indiana -0.36 Michigan State -0.39 Wisconsin -0.40 Purdue -0.56 Illinois -0.83 Rutgers -1.42 * - All I've done to weight per SOS is multiply by the ratio of the team's SOS rating to that of the strongest SOS rating in the FBS. I'm trusting footballperspective.com for the SOS ratings. I used to do some rough calc's of my own but it was very time-consuming and I wasn't sure how valuable the numbers were. Here's the eye candy from Moiraine: Top 25 Statistical Performances Overall (only considering games between FBS teams per the z-score method described previously - not weighted for SOS) Michigan 1.76 Ohio State 1.48 Louisville 1.42 Alabama 1.33 Washington 1.29 Baylor 1.13 Auburn 1.00 Western Michigan 0.98 Florida 0.98 LSU 0.85 Washington State 0.84 Colorado 0.84 San Diego State 0.80 Virginia Tech 0.73 West Virginia 0.69 Nebraska 0.66 Clemson 0.66 Toledo 0.65 Troy 0.58 Houston 0.56 Army 0.55 Memphis 0.50 Boise State 0.49 Wisconsin 0.43 Iowa 0.41 Top 25 Statistical Performances Overall, this time weighted for SOS Michigan 1.51 Ohio State 1.18 Alabama 1.15 Louisville 1.09 Auburn 0.88 Washington 0.84 LSU 0.72 Washington State 0.72 Colorado 0.66 Florida 0.66 Western Michigan 0.62 Baylor 0.60 Clemson 0.57 West Virginia 0.57 Virginia Tech 0.55 Nebraska 0.47 San Diego State 0.44 Wisconsin 0.42 Houston 0.40 Toledo 0.36 Oklahoma 0.36 Boise State 0.35 Troy 0.32 Memphis 0.32 USC 0.28 Top 25 Offensive Statistical Performances to date (not weighted) Louisville 2.17 Washington 1.65 Toledo 1.52 Ohio State 1.51 Michigan 1.41 South Florida 1.29 Alabama 1.29 Baylor 1.26 Texas Tech 1.25 Oklahoma 1.15 Western Michigan 1.08 Louisiana Tech 1.08 Washington State 0.99 Houston 0.99 California 0.93 Texas 0.91 Oregon 0.91 Pittsburgh 0.86 Texas A&M 0.79 Colorado 0.79 Mississippi 0.73 Middle Tennessee 0.72 Memphis 0.69 TCU 0.66 West Virginia 0.65 Nebraska is 29th in this (offensive) category. Top 25 Defensive Statistical Performances to date (not weighted) Michigan 2.18 Florida 1.92 Ohio State 1.80 Alabama 1.65 Army 1.48 LSU 1.47 Wisconsin 1.22 Clemson 1.20 Virginia Tech 1.19 Miami (Florida) 1.12 Auburn 0.95 Baylor 0.93 Minnesota 0.92 Louisville 0.90 Washington 0.88 San Diego State 0.88 Tulane 0.86 Troy 0.86 UCLA 0.85 Western Michigan 0.82 North Carolina State 0.82 Boston College 0.81 Houston 0.81 Colorado 0.80 Washington State 0.77 Nebraska is 26th in this category. Here are Nebraska's average z-scores for all the categories I look at... passing defense 0.81 passing offense 0.12 rushing defense 0.62 rushing offense 0.62 scoring defense 1.37 scoring offense 0.68 total defense 0.81 total offense 0.38 turnover margin 0.50 p.s. - Maybe I should just weight all these? Let me know if that makes more sense to you. 6 Quote Link to comment
NUance Posted October 26, 2016 Share Posted October 26, 2016 Here are beorach's two previous threads with 2016 rankings, for explanation of his methodology: (Were there only two?) Oct. 20 Sept. 30 Quote Link to comment
beorach Posted October 26, 2016 Author Share Posted October 26, 2016 Here are beorach's two previous threads with 2016 rankings, for explanation of his methodology: (Were there only two?) Oct. 20 Sept. 30 Thanks, NUance! I don't think there's a third. My plan had been to just keep one thread going but, as Moiraine pointed out, that probably wasn't the best idea. Quote Link to comment
HuskerExpat Posted October 26, 2016 Share Posted October 26, 2016 First time I've seen your post. Very solid! Thanks for sharing your work!!!! 1 Quote Link to comment
beorach Posted October 26, 2016 Author Share Posted October 26, 2016 First time I've seen your post. Very solid! Thanks for sharing your work!!!! Thanks, HE. It's not much work after the original setup of spreadsheets. I wouldn't even have to copy and paste if I switched everything to Google Docs (as I've been told it would pull in the data from a link automatically?). The board helped me realize I should just be using z-scores, too. I had previously been using these long conditional statements in cells to check where a stat landed on a particular bell curve. I just don't think my basic stats class mentioned z-scores. Quote Link to comment
Recommended Posts
Join the conversation
You can post now and register later. If you have an account, sign in now to post with your account.