bleedinghuskerred Posted May 14, 2012 Share Posted May 14, 2012 What exactly are you trying to find - the average rankings of a given team based on the current ranking lists people have made? Link to comment
knapplc Posted May 14, 2012 Author Share Posted May 14, 2012 What exactly are you trying to find - the average rankings of a given team based on the current ranking lists people have made? PM sent. Link to comment
corncraze Posted May 14, 2012 Share Posted May 14, 2012 1. LSU 2. U$C 3. Oklahoma 4. Alabama 5. Oregon 6. West Virginia 7. South Carolina 8. Michigan 9. Arkansas 10. Michigan State 11. Kansas State 12. Clemson 13. Georgia 14. Wisconsin 15. Florida State 16. Stanford 17. Nebraska 18. Virginia Tech 19. Boise St. 20. Ohio State 21. Texas 22. Auburn 23. Notre Dame 24. Oklahoma State 25. Georgia Tech 26. TCU 27. Florida 28. Miss. State 29. Washington 30. Penn State Link to comment
jsneb83 Posted May 14, 2012 Share Posted May 14, 2012 If you are trying to find a consensus top 25, the best way is to assign points (ie 30 for 1st, 1 for 30th) and rank the teams by the number of points. As far as averages go, it depends on what confidence level you want. But I would say since most people have the same top 10-12 teams, 15 should be enough. But for the lower ranked teams, where there is more disparity, you will need more, roughly 20-30 to get a good approximation. Link to comment
knapplc Posted May 14, 2012 Author Share Posted May 14, 2012 Do I want to rank by total points or average of points? Or does that get me the same thing? I've got each ranking pointed in reverse order, like you said (30 for first, 29 for second, etc) and my thought was to divide each team's total by the number of entries I get. At the moment there are 12, so Nebraska's total would be 175 with an average of 14.58, while Missouri's total would be 59 (they're ranked eight times) but their average would be 4.92 (59/12). I think that's a better way of doing it than using point totals, but it may get me the same thing when I'm done. I have another suggestion of running the same numbers twice - once with all totals, once removing high/low numbers. I'm thinking now I may not be able to eliminate hi/lo because of teams like Washington, who are only ranked three times, or Pittsburgh who is only ranked once. So I think that idea's out the window. Link to comment
fro daddy Posted May 14, 2012 Share Posted May 14, 2012 you would have to do it by point rank. Thats the best way. I wouldnt eliminate the outliers. Its such a small sample size, that and its also not a true look at the board if you do. People are bound to not feel the same about all teams. Certain teams others always downgrade (i.e. Mizzou) and others are always upgraded (i.e. Texas). Buy dropping those highs and lows you really can change the board poll. You could do a column like whats on pollspeak. Have the top 30 poll. Then show the highest rank and lowest rank given. Link to comment
broganreynik Posted May 14, 2012 Share Posted May 14, 2012 1. Oklahoma 2. Alabama 3. USC 4. Oregon 5. Michigan 6. TCU 7. LSU 8. Georgia 9. Arkansas 10. Virginia Tech 11. South Carolina 12. West Virginia 13. Nebraska 14. Kansas State 15. Wisconsin 16. Washington 17. Mississippi State 18. Boise State 19. Ohio State 20. Southern Miss 21. Oklahoma State 22. Clemson 23. Texas A&M 24. SMU 25. Florida State 26. Iowa 27. Georgia Tech 28. Stanford 29. Arkansas State 30. Michigan State Link to comment
broganreynik Posted May 14, 2012 Share Posted May 14, 2012 Do I want to rank by total points or average of points? Or does that get me the same thing? I've got each ranking pointed in reverse order, like you said (30 for first, 29 for second, etc) and my thought was to divide each team's total by the number of entries I get. At the moment there are 12, so Nebraska's total would be 175 with an average of 14.58, while Missouri's total would be 59 (they're ranked eight times) but their average would be 4.92 (59/12). I think that's a better way of doing it than using point totals, but it may get me the same thing when I'm done. I have another suggestion of running the same numbers twice - once with all totals, once removing high/low numbers. I'm thinking now I may not be able to eliminate hi/lo because of teams like Washington, who are only ranked three times, or Pittsburgh who is only ranked once. So I think that idea's out the window. I think total points is better. If you go by average, then teams near the bottom get a break because the times they aren't ranked don't go against them, inflating their average ranking. Unless you just average it out with all of the lists, but then that just gets you the same as total points, so it'd be fruitless. Link to comment
knapplc Posted May 14, 2012 Author Share Posted May 14, 2012 Do I want to rank by total points or average of points? Or does that get me the same thing? I've got each ranking pointed in reverse order, like you said (30 for first, 29 for second, etc) and my thought was to divide each team's total by the number of entries I get. At the moment there are 12, so Nebraska's total would be 175 with an average of 14.58, while Missouri's total would be 59 (they're ranked eight times) but their average would be 4.92 (59/12). I think that's a better way of doing it than using point totals, but it may get me the same thing when I'm done. I have another suggestion of running the same numbers twice - once with all totals, once removing high/low numbers. I'm thinking now I may not be able to eliminate hi/lo because of teams like Washington, who are only ranked three times, or Pittsburgh who is only ranked once. So I think that idea's out the window. I think total points is better. If you go by average, then teams near the bottom get a break because the times they aren't ranked don't go against them, inflating their average ranking. Unless you just average it out with all of the lists, but then that just gets you the same as total points, so it'd be fruitless. If I divide by the total ballots submitted rather than by the times they're ranked it would. But I think everyone is right - total points is better (and lots easier). Link to comment
jsneb83 Posted May 14, 2012 Share Posted May 14, 2012 Do I want to rank by total points or average of points? Or does that get me the same thing? I've got each ranking pointed in reverse order, like you said (30 for first, 29 for second, etc) and my thought was to divide each team's total by the number of entries I get. At the moment there are 12, so Nebraska's total would be 175 with an average of 14.58, while Missouri's total would be 59 (they're ranked eight times) but their average would be 4.92 (59/12). I think that's a better way of doing it than using point totals, but it may get me the same thing when I'm done. I have another suggestion of running the same numbers twice - once with all totals, once removing high/low numbers. I'm thinking now I may not be able to eliminate hi/lo because of teams like Washington, who are only ranked three times, or Pittsburgh who is only ranked once. So I think that idea's out the window. I think total points is better. If you go by average, then teams near the bottom get a break because the times they aren't ranked don't go against them, inflating their average ranking. Unless you just average it out with all of the lists, but then that just gets you the same as total points, so it'd be fruitless. If I divide by the total ballots submitted rather than by the times they're ranked it would. But I think everyone is right - total points is better (and lots easier). Yeah if you divide every team by the same number of ballots, they would be in the same order. So you are skipping a step by using total points. Link to comment
NUance Posted May 15, 2012 Share Posted May 15, 2012 1 - USC 2 - Oklahoma 3 - Alabama 4 - Ohio St 5 - Florida St 6 - Georgia 7 – Oregon 8 - LSU 8 - West Virginia 10 - Boise State 11 - Kansas St 12 – Clemson 13 - Oklahoma St 14 - Nebraska 15 – Texas 16 - Michigan St 17 - South Carolina 18 - TCU 19 - Virginia Tech 20 - Michigan 21 - Notre Dame 22 - Auburn 23 - Wisconsin 24 - Iowa 25 - Missouri 26 - Florida 27 - Penn State 28 - Stanford 29 - Texas A&M 30 – Nevada Link to comment
Excel Posted May 16, 2012 Share Posted May 16, 2012 There are much easier ways to determine an aggregate top 25 than assigning points and going through all that pain...excel, copy paste to columns, sort, take averages and you're done. If you're handy with excel it should only take a few minutes...I won't ruin it for everyone but ya'll's current Big Ten rankings are; 8. Michigan 13. Ohio State 15. Michigan State 16. Wisconsin 17. Nebraska The only tricky or tiresome parts are copying and pasting all the lists, replacing all the commas, periods dashes and parentheses with blanks, sorting through the fact that you guys used multiple names for the same team and then removing the few outliers you all threw out there. It's important to keep in mind that they're not straight averages of what you all ranked them, there's more in play than just what that single team was ranked, its all the other teams as well. Link to comment
knapplc Posted May 16, 2012 Author Share Posted May 16, 2012 I have them in a spreadsheet. I have a macro to remove all the rank numbers, commas & periods. The multiple names thing is kind of a bear, but expected. The problem with just sorting them by name is, as you mentioned, the outlier teams. I've tossed everything into one column, sorted by name, fixed the different names, and I should have a product by the end of the day (pending any last-minute submissions). Link to comment
Recommended Posts