Follow @HSFBamerica
We're one week into the postseason here in Southern California and we're now seeing lots of movement in our High School Football America Diamond Rankings developed by SportsMeasures. In order to explain how the rankings are scientifically developed, I've asked Diamond Rankings guru Patrick Fisher to give us a lesson.
Here's what Patrick had to say....
SportsMeasures’ Diamond Rankings are created by using the educational testing theorem Rasch Model. The Rasch Model (aka fundamental measurement, objective measurement, Rasch Measurement - learn more at www.Rasch.org) is used extensively in education (K-12) testing, high-stakes certification testing, health care including rehabilitation, low vision, diagnostic care and, many other fields. SportsMeasures has been pioneering various solutions to sports measurement issues for over 25 years.
The most obvious problem area of measurement in sports is the top 25 ranking of teams in all sports, but particularly in the high profile sports of college football. Traditionally, college football teams have been ranked in polls by the voting of coaches or sports journalists. With the advent of the Bowl Championship Series, computer rankings were made part of the ranking process. Unfortunately, that process is bereft with problems, not the least of which is that their methodology is wholly unscientific.
That brings us to the current state of affairs late in the 2013 football season and the quantum leap ahead offered by our Diamond Rankings.
Why measures and standard error?
Good question!
Objective measurement is the only way to produce a scientifically reliable measure that will stand up to scrutiny. It can be trusted and relied on in every instance. Think of this as a 12-inch ruler that measures ability in football (or any sport). The ruler created by SportsMeasures for High School Football America to evaluate the California Interscholastic Federation’s Southern and Los Angeles Sections measures every 11-man football team in those sections. These measures are on a scale of 0-100, but the very best may exceed 100 and the worst may go below 0. We will not artificially restrict the scale for the purposes of it looking pretty. We want the general scale to be 0-100 for the purposes of it being easily consumed by the public.
Using objective measurement brings an heretofore unseen level of fairness and accuracy to sports. Fairness because teams are frequently ranked according to their winning percentage, despite the acknowledgment that strength of schedule and conference strength play important factors in determining the true ability level of a team. When the season wraps up and it is time to select the teams for the playoffs, these factors are critical to 5-4 teams in tough conferences.
The 8-1 team in the "Cupcake Conference" doesn’t want to hear about SoS (Strength of Schedule) and conference strength, but when their one loss is to a 6-3 team by a decisive margin from a mediocre conference, it seems likely that the 5-4 team is quite likely better than the "Cupcakes". So the question is - is it fair to the 5-4 team which plays a tougher schedule and did OK against that schedule to not make the playoffs than for an 8-1 "Cupcake" to get in to the playoff just because they have a glamorous record earned against a cupcake schedule?
Fairness and accuracy go hand-in-hand. The fairness situation described above is possible only because of the accuracy of the measures and that accuracy is due to the objective nature of the measurement methodology.
There is
another definition of accuracy that must be discussed too. This definition of
accuracy is neither discussed nor defined by ANY ranking source because they
can’t calculate one, given their methodology. Standard error of measurement is a
given in any scientific methodology. No social science method is 100% accurate,
which is why error is needed. The acknowledgment and calculation of standard
error is essential to having rigorous measures. Error also gives you a frame of
reference for simply predicting which team is likely to win any given game.
Here's an example - a team has a measure of 90.00 with an error of 5.00; their opponent’s
measure is 86.00 with an error of 4.5. When adding/subtracting their errors we
see that their measures now “overlap” as 90 - 5 = 85 and 86 + 4.5 = 90.5. When
measures overlap after factoring in error, that means that there is a 50/50
probability of either team winning.
Enjoy the rankings....Patrick
Now, back to what has transpired since our previous rankings...
With #4 Long Beach Poly's Pac-5 opening round playoff win over former #7 Orange Lutheran the Lancers fall to #9 overall, which allowed Rancho Cucagmonga to move-up two spots from last week to #7. Corona Centennial pops back into the Top 10 at #8 after its early season departure following back-to-back losses.
One of the biggest mover this week is Servite, which climbs from #28 to #21 after defeating former #15 Westlake.
Here's the High School Football America Diamond Rankings through November 16, 2013.
| Team | Record | Measure | Error |
| 1. St. John Bosco | 11-0 | 104.83 | 6.07 |
| 2. Mission Viejo | 11-0 | 99.81 | 6.64 |
| 3. Mater Dei | 9-2 | 97.69 | 5.53 |
| 4. Long Beach Poly | 10-1 | 96.80 | 5.80 |
| 5. Serra | 11-0 | 94.95 | 6.06 |
| 6. Vista Murrieta | 10-1 | 93.68 | 5.21 |
| 7. Rancho Cucamonga | 10-1 | 92.68 | 5.49 |
| 8. Corona Centennial | 9-2 | 92.46 | 5.41 |
| 9. Orange Lutheran | 7-4 | 92.13 | 5.46 |
| 10. St. Bonaventure | 8-3 | 92.04 | 5.49 |
| 11. Alemany | 9-2 | 91.48 | 5.87 |
| 12. Oaks Christian | 8-3 | 90.75 | 5.48 |
| 13. Venice | 9-2 | 90.74 | 5.11 |
| 14. Narbonne | 9-2 | 87.91 | 5.53 |
| 15. Tesoro | 9-2 | 87.37 | 6.26 |
| 16. Hart | 9-2 | 86.81 | 5.14 |
| 17. JSerra | 7-3 | 86.61 | 6.36 |
| 18. Notre Dame (Sherman Oaks) | 7-4 | 86.06 | 5.79 |
| 19. Norco | 8-3 | 85.69 | 5.71 |
| 20. Chaminade | 9-2 | 85.54 | 5.28 |
| 21. Servite | 6-5 | 85.42 | 6.06 |
| 22. Palmdale | 9-2 | 85.03 | 5.10 |
| 23. Westlake | 8-3 | 84.07 | 6.10 |
| 24. Crenshaw | 7-4 | 83.75 | 5.42 |
| 25. San Pedro | 8-3 | 82.92 | 5.20 |
| 26. Upland | 8-3 | 82.40 | 6.12 |
| 27. Bishop Amat | 6-4 | 82.75 | 5.99 |
| 28. Valencia (Valencia) | 9-2 | 82.13 | 5.59 |
| 29. Arroyo Grande | 8-3 | 81.84 | 4.71 |
| 30. San Fernando | 11-0 | 81.79 | 5.00 |
| 31. West Torrance | 8-3 | 81.39 | 5.20 |
| 32. Canyon Country Canyon | 8-3 | 81.33 | 5.43 |
| 33. Lompoc | 10-1 | 81.28 | 4.69 |
| 34. Birmingham | 8-3 | 81.04 | 5.13 |
| 35. Moorpark | 7-3 | 80.65 | 6.05 |
| 36. Palos Verdes | 9-2 | 80.47 | 4.71 |
| 37. Camarillo | 10-1 | 80.05 | 5.01 |
| 38. St. Francis | 9-2 | 79.59 | 5.09 |
| 39. Loyola | 4-6 | 79.17 | 6.13 |
| 40. Charter Oak | 7-4 | 79.13 | 5.74 |
| 41. Chaparral | 8-3 | 78.92 | 6.03 |
| 42. Chino Hills | 7-4 | 78.80 | 5.68 |
| 43. Garfield | 8-3 | 78.11 | 5.73 |
| 44. Rancho Verde | 9-2 | 77.76 | 4.62 |
| 45. Corona Santiago | 6-5 | 77.62 | 5.88 |
| 46. Lakewood | 5-6 | 77.29 | 5.74 |
| 47. Newberry Park | 5-5 | 77.29 | 6.06 |
| 48. Mira Costa | 8-3 | 77.03 | 5.55 |
| 49. Thousand Oaks | 6-4 | 77.02 | 6.22 |
| 50. Edison | 8-3 | 76.93 | 5.06 |
