Hey everyone,
Now that week seven has come to a close, I feel like my computer poll is valid enough to be worth a look. It's not as solid as it will be after a few more weeks of data, but I feel like it's worth posting at this point. If you'd like a short synopsis of my ranking methodology, please read the quoted text at the bottom. If anyone has questions as to why certain teams are ahead of others, please feel free to ask and I'll dig into the data for an answer.
Before reading the rankings, keep in mind that the ratings following the universities are very important. The difference of 0.02 between ranks 7-11 means the schools are nearly statistically equal. The difference of 0.2 between ECU and Boise State is very significant.
The current G5 access bowl rankings according to my algorithms:
+0 | 1. Marshall 0.364748599
+0 | 2. East Carolina 0.338537535
+2 | 3. Colorado State 0.255491002
+5 | 4. Temple 0.157288922
+2 | 5. Boise State 0.13677057
NR | 6. UAB 0.090422515
- 4 | 7. BYU 0.078858194 (Not eligible for the G5 access bowl slot)
+4 | 8. Georgia Southern 0.071224679
NR | 9. Bowling Green 0.069623009
NR | 10. Utah State 0.061125939
NR | 11. Akron 0.06003316
NR | 12. Arkansas State 0.053997881
The reason Marshall is over ECU is due to the performance of the teams ECU has beat. As of the current AP poll, ECU has wins over 5 unranked teams and a loss to an unranked South Carolina.
Because ECU has a significantly better SoS, if both teams were to win out the rest of their schedules, ECU would be ahead by the end of the year but at this particular moment, Marshall has the edge. Other factors that could boost ECU would be if VT, UNC, or South Carolina climbed back into the rankings.
Something to remember is that these rankings do not predict who will be on top by the end of the year, but rather a snapshot at a point in time. During week 6, the difference between Marshall and ECU was 0.08. This week, ECU gained on Marshall and the difference is less than .03. Expect ECU to overtake Marshall over the next couple weeks if ECU continues to win.
Quote:Note: The formulas that go into the rankings are complex, so I'll only be giving a general overview of how everything is calculated.
The score is broken out into three main categories:
* 15% of the score is from SOS adjusted offensive stats
* 15% of the score is from SOS adjusted defensive stats
* 70% of the score is from win/loss (MoV is not taken into consideration)
For SOS, I start with taking the most recent Massey SOS rankings and add a multiplier to each ranking. This SOS multiplier is used against all three scores (offensive, defensive, and win/loss).
For win/loss, teams are greatly benefited by wins over currently ranked teams and are significantly hurt by losses to unranked teams. This computer poll uses a point based ranking system where each team starts the season with 0 points (thus, no preseason bias).
* A win over a currently ranked AP30 team is +3 points
* A win over a currently unranked AP30 team is +1 point
* A loss to a currently ranked AP30 team is -2 points
* A loss to a currently unranked AP30 team is -3 points
This total score then gets multiplied by the SOS multiplier and makes up 70% of the final score.
The final score has a maximum value of 1. This value is obtainable only with a top defense, top offense, and top record against a good schedule. You may view each team's score as how close each team is to being the "perfect team."
Due to the use of currently ranked/unranked teams in the score calculations, this poll is able to avoid situations like what happened with Miami during the 2013 season. In the regular polls, Miami narrowly beat Florida and shot up in the rankings because Florida was ranked fairly high. However, we noticed that over the next few games, Florida was not actually that good. Unfortunately, due to the nature of human polls, Miami's ranking was never adjusted downwards until they lost. This poll would (and did) adjust them downwards as soon as Florida drops from the top 30.