Numbers out the ying-yang
After reviewing the Week 10 Top 24, a loyal reader had the following question:
Allow me to shed some light on the subject (hopefully). Let's first review how the formula works. Basically, I take the record, total points, yards, and turnovers for each team in every game played so far. Then I calculate two subscores using the difference between a team's and their opponents' numbers and a ratio. I average these two values together to get a final subscore. To determine the SOS for a team, I take 67% of each opponent's subscore and 33% of each opponent's opponent's subscore. Going back to the subscores, the record gets a 50% weighting, then 35% goes to the statisical subscores (each stat is weighted roughly the same), and then 15% goes to SOS. This yields a ratio score and a difference score. I average these two and quantize between 1 and 0 to get the final score you see posted on the site. So to figure out why a team has a higher score than another team, you have to look at the individual components, both for difference and ratio. The "per game" numbers aren't really used in the calculations, but provide a clearer barometer of a team's performance than totals when viewing the numbers prior to season's end. The efficiencies shown below (also not directly used in any calculations) are based on scoring 70 points for every 80 yards of offense. It's an arbitrary adjustment, I realize, but a telling statistic nonetheless.
Confused yet? Okay, let's start with Penn State and Texas Tech. The difference components are roughly the same, but Penn State leads in every ratio component, and has a higher SOS value. I know, I know, the Big 12 is "better" than the Big Te(leve)n, but look at who each team has played through week 10. I have no doubt that TTU's final SOS will be higher than PSU, but up until that point, PSU does have the slight edge. More importantly, look at the efficiencies: Penn State scores more often with less yardage and allows fewer points and fewer yards, and they almost never turned the ball over.
As for TCU over Alabama, the Horned Frogs have superior numbers in every offensive and defensive category. Alabama's got SOS cornered and of course has a better record, but it's the massive discrepancy in the statistical categories that was keeping TCU just ahead of an undefeated 'Bama team.
"What is making Penn State appear higher than Texas Tech, and a one-loss TCU higher than undefeated Alabama?"
Allow me to shed some light on the subject (hopefully). Let's first review how the formula works. Basically, I take the record, total points, yards, and turnovers for each team in every game played so far. Then I calculate two subscores using the difference between a team's and their opponents' numbers and a ratio. I average these two values together to get a final subscore. To determine the SOS for a team, I take 67% of each opponent's subscore and 33% of each opponent's opponent's subscore. Going back to the subscores, the record gets a 50% weighting, then 35% goes to the statisical subscores (each stat is weighted roughly the same), and then 15% goes to SOS. This yields a ratio score and a difference score. I average these two and quantize between 1 and 0 to get the final score you see posted on the site. So to figure out why a team has a higher score than another team, you have to look at the individual components, both for difference and ratio. The "per game" numbers aren't really used in the calculations, but provide a clearer barometer of a team's performance than totals when viewing the numbers prior to season's end. The efficiencies shown below (also not directly used in any calculations) are based on scoring 70 points for every 80 yards of offense. It's an arbitrary adjustment, I realize, but a telling statistic nonetheless.
Points | 376 | 423 | 366 | 290 |
Rank | 7 | 3 | 9 | 26 |
PPG | 41.77 | 47 | 36.6 | 32.22 |
Rank | 8 | 3 | 17 | 33 |
Yards | 4138 | 5034 | 4085 | 3321 |
Rank | 9 | 2 | 13 | 49 |
YPG | 459.77 | 559.33 | 408.5 | 369 |
Rank | 12 | 2 | 29 | 61 |
Tos Committed | 8 | 12 | 11 | 12 |
Rank | 5 | 32 | 24 | 25 |
TO/G | 0.88 | 1.33 | 1.1 | 1.33 |
Rank | 4 | 28 | 10 | 22 |
Points | 100 | 202 | 108 | 110 |
Rank | 4 | 53 | 5 | 7 |
PPG | 11.11 | 22.44 | 10.8 | 12.22 |
Rank | 4 | 47 | 3 | 6 |
Yards | 2393 | 3146 | 2145 | 2263 |
Rank | 12 | 66 | 4 | 7 |
YPG | 265.88 | 349.55 | 214.5 | 251.44 |
Rank | 6 | 56 | 2 | 4 |
Tos Forced | 18 | 22 | 26 | 16 |
Rank | 43 | 13 | 2 | 59 |
TO/G | 2 | 2.44 | 2.6 | 1.77 |
Rank | 47 | 20 | 9 | 67 |
Points | 276 | 221 | 258 | 180 |
PPG | 30.66 | 24.56 | 25.8 | 20 |
Yards | 1745 | 1888 | 1940 | 1058 |
YPG | 193.89 | 209.78 | 194 | 117.56 |
TO Margin | 10 | 10 | 15 | 4 |
TO/G | 1.12 | 1.11 | 1.5 | 0.44 |
Points | 3.76 | 2.094 | 3.389 | 2.636 |
PPG | 3.757 | 2.094 | 3.389 | 2.637 |
Yards | 1.729 | 1.600 | 1.904 | 1.468 |
YPG | 1.729 | 1.600 | 1.904 | 1.468 |
TO Margin | 2.25 | 1.833 | 2.364 | 1.333 |
TO/G | 2.273 | 1.835 | 2.364 | 1.331 |
SOS Score | 0.3407 | 0.2832 | 0.2581 | 0.4657 |
Rank | 88 | 92 | 101 | 56 |
Off. Effic. | 103.84% | 96.03% | 102.39% | 99.79% |
Rank | 7 | 22 | 11 | 17 |
Def. Effic. | 47.75% | 73.38% | 57.54% | 55.55% |
Rank | 4 | 38 | 11 | 8 |
Confused yet? Okay, let's start with Penn State and Texas Tech. The difference components are roughly the same, but Penn State leads in every ratio component, and has a higher SOS value. I know, I know, the Big 12 is "better" than the Big Te(leve)n, but look at who each team has played through week 10. I have no doubt that TTU's final SOS will be higher than PSU, but up until that point, PSU does have the slight edge. More importantly, look at the efficiencies: Penn State scores more often with less yardage and allows fewer points and fewer yards, and they almost never turned the ball over.
As for TCU over Alabama, the Horned Frogs have superior numbers in every offensive and defensive category. Alabama's got SOS cornered and of course has a better record, but it's the massive discrepancy in the statistical categories that was keeping TCU just ahead of an undefeated 'Bama team.
0 Comments:
Post a Comment
<< Home