Welcome back. This is the second part in a series on answering the question "Which teams are deserving of a playoff invitation?" In it I'll outline a model I'll refer to as "Extended Standings" and over the course of the series I'll provide the exact details so anyone interested can independently verify the results.
Although the focus of the series is not necessarily a critique of the status quo, to help us establish a baseline this article will take a deeper look at the two existing models the GHSA uses to determine which teams are deserving of a playoff invitation: region standings and the Class A power ratings.
We'll look at each through the five characteristics any playoff invitation model should have as outlined in Part I of the series – Transparency, Objectivity, Economy, Equitability, and Rigorousness.
The GHSA region standings are fairly straightforward, although each region has some latitude to make various relatively minor adjustments, mostly regarding tie-breaking procedures. Region standings are certainly transparent and objective, and perhaps even reasonably equitable, particularly within the region. This equitability suffers somewhat when comparing teams across regions because of the dubious assumption the fourth team in any region is more deserving of a playoff invitation than the fifth team in any region. Still, while this method may not result in the optimal seeding of the participants, it’d be difficult to argue a 32-team playoff has ever overlooked a realistic contender for the state championship altogether.
Obviously the region standings do not possess or even claim any serious degree of rigor, so I won’t spend any time on that. However, I think it’s worth examining the more subtle weakness of region standings – they are not economical.
Because region standings are so simple, traditionally a hallmark of being economical, that statement may have thrown a few of you, so let's do just a quick refresher from Part I of the series. Being economical ensures the fundamental data required to perform meaningful comparisons between teams is used. There are two sides of being economical: too much data and too little data. Too much data requires us to discard data of marginal value to avoid unnecessary complexity. Too little data requires us to incorporate data of fundamental value to avoid oversimplification.
Although not necessarily obvious, region standings amazingly exclude roughly 40% of the fundamental data available in the regular season.
How did I reach this preposterous conclusion?
Well, this season, GHSA teams will play approximately 2,100 regular season games. Of those, only about 1,300, or 60%, will be reflected in region standings. Since the region standings are used to determine playoff invitations, the other 40% of the data from regular season games is completely discarded.
Let's use this year’s Corky Kell Classic as an example. Although the Classic hosted some of the top matchups of the regular season, not a single outcome made any difference as to if any of the participants will make the playoffs or where they will be seeded.
So, from the perspective of playoff invitations, those great matchups were just meaningless exercises organized around financial incentives and bragging rights.
To be clear, organizing games around financial incentives and bragging rights is perfectly fine, but the point here is that region standings discard those non-region matchups in spite of them containing fundamental data in regards to performing meaningful comparisons between teams.
Imagine if the results of the Corky Kell Classic, for example the game between Colquitt County and North Gwinnett, factored into playoff invitations. Granted both will probably make the playoffs under the region standings anyway, but conceivably the top seed in Class AAAAAA could have been at stake even at this early stage in the season.
In short, an economical model would count all games in determining which teams are deserving of a playoff invitation. In fact, when combined with a model that has rigor, this can drive significant scheduling changes. Suddenly it would be crucial to schedule quality opponents at every possible opportunity, pushing the state’s top teams toward each other in an effort to improve their standing.
The Extended Standings, which will debut tomorrow, takes all GHSA games into account over the course of the season. The Buford-McEachern matchup in week 3 is a great example of a cross-classification game where the outcome will factor into the Extended Standings but be ignored by the traditional region standings.
Now let's take a closer look at the Class A power ratings, which are presumably the front runner for the Big 44 power ratings. The exact details of the power ratings can be found here .
Just like region standings, this model is transparent and objective. Concerning economical, it includes region and non-region games and even factors in out-of-state opponents, but it doesn’t quite meet the threshold of being economical for reasons I’ll cover in a future article.
Although certainly more difficult to calculate than region standings, the Class A power ratings still cannot be considered rigorous. It does go a step further than region standings in that it examines opponents’ records, but rigor requires a thorough analysis of the entirety of the data. Again, I’ll cover this idea further in a future article.
For now, let's focus on whether the Class A power ratings are equitable. That is, do the results seem fair and just?
Let’s look at a couple of scenarios to see.
Suppose a Class A team, Roosevelt High, plays a Class AA team, Kennedy High. In a fairly close game, Kennedy prevails 23-16. Under the Class A power ratings, Roosevelt would get 0 points for the loss with a 0.5 bonus since the loss came against a Class AA team. Suppose Kennedy finished the regular season 10-0, then Roosevelt would get an additional 10 points for a total of 10.5 points.
In this scenario, Roosevelt is obviously a competitive Class A team, losing by only a touchdown to Kennedy, an undefeated Class AA team.
Now let’s see if the Class A power ratings treat Roosevelt equitably by comparing it to another example. Suppose a different Class A team, McKinley High, opted to play an average Class A team, Taylor High, and won 23 to 20. McKinley would be awarded 10 points for the victory. If Taylor went on to win 5 games that year, as an average Class A team could reasonably be expected to do, then McKinley would be awarded 5 more points, for a total of 15 points.
The end result is that Roosevelt, who lost 23-16 to an undefeated Class AA team, would be rated lower than McKinley, a 23-20 victor over a 5-5 Class A team.
This was the exact scenario in 2014 when Eagle's Landing Christian lost 23-16 to undefeated Greater Atlanta Christian, the eventual state runner up in Class AA, while Prince Avenue Christian defeated a 5-5 Hebron Christian team by 23-20. The Class A power ratings favored Prince Avenue Christian as more deserving of a playoff invitation.
That’s off the basis of one game, but let’s look at ELCA’s 2014 season as a whole and calculate each game’s contribution to their Class A power rating:
Total points = 114.93 / 10 games played = 11.49 points
[***Special thanks to CheerTigerDad for catching the erroneous inclusion of Crawford County, Cross Keys, and North Cobb Christian in Mount Vernon Presbyterian's record. Those opponents played non-region schedules and should not have been counted in the record.***]
Clearly winning is important under the Class A power ratings, which seems equitable in and of itself, but it seems the system is out of balance when a victory over a 2-8 Class A team can help more than a close game against an undefeated opponent in a higher class. In fact, a 7-6 victory over a 2-8 Class A team is the same as a 43-42 shootout loss to an undefeated Class AAAAA team, each worth 12 points. If the Class AAAAA team ends up 9-1 instead of 10-0, then that game is worth only 11 points, so defeating the 2-8 Class A team would have been better.
For comparison, let’s quickly look at Prince Avenue Christian’s 2014 season as well:
Total points = 125.86 / 10 games played = 12.59 points
[***Special thanks to CheerTigerDad for catching the misidentification of Oglethorpe County as Class A and the erroneous inclusion of Glascock County in Athens Academy's record. Glascock County played a non-region schedule and should not have been counted in the record.***]
Here’s some quick notes between the two:
- Not counting games against ELCA, their opponent’s went a total of 64-26 on the season (71.1%) while Prince Avenue Christian’s opponents went 37-51 (42.0%)
- ELCA played two undefeated teams, one even from Class AA, while Prince Avenue Christian played one
- Eight of ELCA’s opponents had winning regular season records while three of Prince Avenue Christian's opponents did
- ELCA played four opponents from a higher classification while Prince Avenue Christian played one
[***Although not included in the original article, I wanted to highlight a great point brought up by CheerTigerDad concerning an inherent inequality between teams in regions of different size: "ELCA only has 4 other teams to play against in their Region 5. That means ELCA can go out and get any other 6 schools they want (or could find that wanted to play them in their case last year). PACS has 8 region teams that they are forced to play against, which only leaves 2 games available to pick and choose. The teams that PACS has to play would most likely have worse records since they all play against each other (for each win, there has to be a loss)."; see comments for more details***]
Returning briefly to the four principles of equitability from the Part I of this series, it would seem the Class A power ratings are in violation of the last principle -- Stronger opponents should be valued more than weaker opponents. In fact, if we assume ELCA's 23-16 loss to Greater Atlanta Christian to be their best game of the season, Prince Avenue Christian has seven games that earned more points. Those games featured seven Class A opponents with a combined 28-40 record.
Yes, there is a bonus for playing a higher classification, but in the end, the model trades equitability for ease of computation and is just too far out of balance, rewarding teams for easy wins over weak opponents more than close games against strong opponents.
Coming up I'll show the Extended Standings rewards teams for scheduling tougher opposition.
Well, I’ll close for now. Again, if you have any thoughts I’d love to hear them!
About the Author
