The Glaring Flaw and the Impossible


A well-known sportswriter, I’ll call him Mr. Smith, tweeted a quote from SEC commissioner Greg Sankey indicating that he and the other commissioners would rethink the idea of awarding byes in an expanded playoff to conference champions only. As I often do, I asked if any thought was being given to eliminating subjectivity. Another tweeter replied to claim that there is no reasonable way to eliminate subjectivity with a 130+ teams. I disagreed and he and Mr. Smith asked how this might be accomplished. I provided a link to a previous column that spelled out the simple rules for the Power Points System. Mr. Smith almost immediately dismissed the idea claiming it took him 30 seconds to find the system’s “glaring flaw” and that I was “attempting to do the impossible” by eliminating subjectivity. Of course, Twitter is not the best format for offering an informed rebuttal to the arguments made by Mr. Smith that I have heard countless times from others. Therefore, I offer a limited rebuttal here.

The idea that it is impossible to place and advance College Football teams using purely objective means is patently absurd. It is often claimed that subjectivity is necessary and unavoidable because College Football has too many teams and its teams play too few games to fairly compare them by purely objective means. Ideally, if the Power Points System (PPS) were used, all FBS teams would play equal or equal maximum game regular season schedules versus FBS competition only. That said, I have 44 seasons worth of results based on FBS games only. Compared to the various subjective methods used during this time, the PPS agrees on slightly better than three top four teams. Where there is disagreement, the teams favored by the PPS average more wins and games played versus competition ranked by the subjective methods. The average FBS season since 1978 includes better than 620 games between FBS members. If the goal (it is not) were to match subjective methods on all top four teams, I would only need to change the result of one game on average per season to get there. And these are the results without the benefit of teams playing equal or equal maximum game regular season schedules and knowingly competing to according to the PPS’s rules.

I strongly suspect that those who believe too many teams and too few games are a problem for objective methods think objective methods do not work because they will inevitably fail to validate their subjective judgements. They often think the best team should be favored for that reason or better competition as matter of opinion must have greater objective value, but that is not how objective methods work. While objective and subjective values may agree that X is greater than Y a significant percentage of the time, objective values are not designed to match subjective values especially given that voters in subjective methods rarely if ever unanimously agree themselves. Everywhere there are rules, teams place and advance based on which teams best accomplish what the rules in play value without regard to what may be subjectively more impressive to anyone with an opinion. If the rules say X is greater than Y, it is the responsibility of teams to accomplish X rather than Y and being a better team is only a competitive advantage towards posting better results according to the rules in play. If the better team fails to do that, the better team loses under those rules. It is not entitled to place higher simply because it is a better team.

As for the argument that teams play too few games, it is claimed that 12-13 data points is not enough games to compare teams objectively. If so, how do subjective methods get around this problem? How many more games are needed to compare any two teams objectively? And are we really going to pretend that polls and committees do so much more than primarily rank teams by their records? With few exceptions, the CFP selection committee ranks P5 and G5 teams by their records among their own kind. Only when comparing P5 teams to G5 teams does the committee go against this trend and the problem there is that the committee format does nothing to explain why P5 teams get one to four free losses compared to G5 teams before the latter is ranked higher. G5 teams do not know what they need to do, especially schedule wise, to avoid being passed in the rankings by P5 teams with worse records. Actual rules, beyond eliminating the two-class system that has been maintained by biased subjective methods, spell out what G5 teams must do to achieve higher rankings.

Again, Mr. Smith claimed it only took him thirty seconds to identify the PPS’s glaring flaw as he sees it. For starters, he says all conferences are not equal which presumably means they are not equally good. I certainly agree and the PPS makes no claim otherwise. Their differences are a competitive advantage only under any possible objective rules and based on the results of the PPS, the better conferences dominate a high percentage of the time. Their differences are not an argument against playing by rules that apply equally to all teams and leagues without regard to any subjective judgements. That said, his primary argument against the PPS is that it will award more value for lesser wins (as a matter of opinion) when he believes that beating an average P5 team would be more “meaningful” than beating a strong G5 team. Of course, I suspect “meaningful” here is purely subjective and seeks to have value tied to such judgements for the purpose of identifying the best teams which is exactly what objective methods do not do.

Mr. Smith believes the PPS can be gamed by scheduling high value G5 opponents that would presumably be easier games than lesser value but stronger P5 competition. It is important to note that the purpose of competition is to determine a winner. Nothing more and nothing less. Again, where rules apply, teams place and advance based on which teams best accomplish what the rules in play value. There are no objective rules that can guarantee its objective values always reward greater subjective values. There are countless examples throughout sports where this is self-evident. With the PPS, the objective of the competition is to schedule and beat opponents that win. The objective is not to beat better competition as a matter of opinion.

Mr. Smith says he would game the system by scheduling the top two teams from the Sun Belt and Conference USA. This past season that schedule would include Appalachian State, Louisiana, UTSA, and Western Kentucky. That said, this scheduling strategy would be known to all teams. In fact, all teams would be inclined to adopt it just to keep pace. If the strategy is successful, it will likely be successful for several teams. Therefore, who places higher boils down to which teams scheduled higher value G5 competition rather than challenging each other directly. Of course, the strategy could also fail because the “easy” opponent wins the game. Beyond that, how easy would it be for any one team to schedule four such games especially if games are still scheduled years in advance? Also, since this is a scheduling strategy that all teams would likely employ, what will teams offer top G5 teams to get these games scheduled? Would they pay more for an unbalanced deal or agree to a home and home series? Since the PPS assigns much more value to beating top G5 teams than the current beauty contest format, G5 teams would have scheduling leverage they have never had. This is very important when you consider most everyone demands they play absurd non-conference schedules that are unattainable for multiple reasons including the fact that top P5 teams have very limited incentive to play top G5 teams. This can only help lead to a more competitive sport if G5 teams know what the required schedule is to compete for a playoff spot and they have the leverage needed to acquire such schedules compared to a system that requires a perfect storm of luck for a G5 team like Cincinnati to be selected for the CFP.

How weak is the hypothetical schedule Mr. Smith would pursue compared to non-conference schedules under the current format anyway? This past season, Alabama played Miami (FL), Mercer, Southern Miss, and New Mexico State. Only Miami (FL) is arguably better than anyone on the hypothetical schedule and the Hurricanes only beat Appalachian State by two points at home. If you look at the 23 non-conference schedules (Notre Dame and BYU excluded) among teams ranked in the final CFP poll, very few are arguably more difficult than the hypothetical schedule. There were only six non-conference games between teams ranked in the final poll and four included a G5 team and G5 teams finished 3-1 in those games. Even if we pretend it is a simple matter for any single team to acquire that schedule especially years in advance and succeed against that schedule, it is still a better schedule than most non-conference schedules played by CFP top 25 teams. And given that there are only so many of these “easy” high value G5 wins to go around, most teams will not be able to fill up their schedules with such games and the top teams will be inclined to schedule each other after all versus filling in open dates with low value bad teams. I concede that it is possible that a team could get more value for a win versus a lower ranked opponent and that could be the reason it ranks higher than another team. The irony is that such examples happen far more often with polls and committees. Ultimately, the glaring flaw is no flaw and the impossible is not only possible, there is no good reason to continue ranking teams subjectively.

Leave a Reply

Top Candidates: Houston
The Walk-On Redshirt Podcast: Week Three Preview
The Walk-On Redshirt Podcast: Week One Preview