Messin' Around with Power Ratings

[LAZ]

Using my numbers, the average FLHS football team has a Mean (Average) Power Ranking of 118.37.

The Median (mid-point) Power Ranking for the entire field is 118.99

The MAD (mean absolute deviation) for the entire field is 16.23

Breaking it down by Classification.

Mean Power Rating

6A	128.55
5A	125.43
4A	124.99
7A	124.90
2A	122.63
3A	120.97
1A	118.94
Rural	110.28
SSAA	101.52
Ind	100.27

Median Power Rating

 

6A	128.65
5A	125.59
4A	125.46
7A	124.99
2A	122.48
3A	122.23
1A	119.30
Rural	110.76
SSAA	101.62
Ind	100.39

MAD - I use this to determine classification parity. A lower number indicates more parity.

SSAA	12.08
6A	12.19
7A	12.80
Ind	13.00
2A	13.65
5A	13.88
4A	14.66
Rural	15.42
3A	15.59
1A	17.17

[LAZ]

Now in comparison take a look at the MAD (Parity) in college football. It's much more competitive. Thus, the reason systems can predict high school games in the 80% range but in college you're doing great if you get in the low to mid 70%. Nearly every week some of the best forecasting systems known to man can't pick 70% FBS winners "straight up".    

FBS-G5	6.71
FBS-P4	7.30
FCS	8.04
D2	9.29
NAIA	11.51
D3	12.98

[Dr. D]

Thank you (again) for the very interesting data.  I would like to offer some statistical context, with two caveats: 1) I don’t know how the “Power” rankings are calculated, and 2) analyzing with a “mean of means” is tricky.

Nevertheless, basic statistical analysis indicates that the Rural, SSAA, and Independent values are statistically significant outliers from the 1A-7A values (which have no statistically different variation among them).  In other words, the 1A-7A values are essentially the same, while the Rural, SSAA, and Independent values are “weaker” than those 7 classes.

This finding suggests that classes 1A-7A have comparable numbers of “strong”, “average”, and “weak” teams, and are essentially “equal” in overall strength when using “Power” rankings.  It also indicates that school enrollment is not correlated with overall strength.  Additionally, there may be statistical validation for the existence of a Rural class.  Once again, competitive equity would be best served by sorting the “strong” teams into their own classes(es), the “average” teams into their own class(es), and the “weak” teams into their own class(es) and foregoing classification based on school enrollment.   

[LAZ]

Agreed. It really supports the argument for a relegation system.