Last night the College Football Playoff Committee released its first rankings of the 2017 season and we are now able to expose the flaws that could lead to the wrong teams playing for the National Championship. Based upon the initial rankings, we can easily dissect their thinking and identify why it is wrong.
The first principle that drove the initial rankings appears to be strength of schedule, or more accurately, a glamourous dance card. Clemson has played six top fifty teams and defeated five of them. Notre Dame has played five top fifty teams and defeated four of them. As a result, Clemson is No. 4 and Notre Dame is No. 3 according to the committee. Notre Dame is probably ahead of Clemson because its loss was by a single point to No. 1 Georgia whereas Clemson threw up on itself at Syracuse. While the RPG system recognizes that Clemson has played the hardest schedule among ranked teams, and Notre Dame has faced the third most difficult competition (USC’s schedule is second toughest), we also recognize that all wins are not equal. The committee seams to assume that if a team won, it played well, and that all teams play equally well in their victories.
This is a major flaw in committee (and football insider) thinking. A team can play poorly and still win if it plays relatively better than its opponent on game day. This weekend Wisconsin turned in a dismal performance against Illinois (74.75, a ‘C’ at best) but still won because Illinois played even worse. Likewise, a team can play well and yet lose if its opponent plays marginally better. That’s what happened at the Horseshoe as Penn State turned in a better losing performance than Wisconsin’s winning performance (84.59) in a narrow loss to Ohio State (which received a grade of 94.93).
Victories over comparable competition must be differentiated by playing performance, just as passing grades on school tests are differentiated by numerical grades, but not once in the broadcast reveal of the rankings did any expert mention playing performance. If they had, they would have to admit that although Clemson played very well against Georgia Tech (100.09), it was their first grade of more than 90 in their string of lackluster–but winning!–performances. If we are to choose the four best teams–not records–for the playoff, we must know how well each team plays the game. The committee seems to consider record and SOS while we consider playing performance and SOS. Wins and losses are simply the result of comparing two test scores and not very useful in determining which teams are better than others. For the record, Notre Dame is No. 6 and Clemson is No. 8 in our rankings.
A quick word about SOS: the committee seems dazzled by top 25 and top 50 teams and dismisses the lower 80 teams in the FBS subdivision but the RPG system knows precisely how much more difficult it is to play a top 50 team or a top 25 team or a top 10 team. Using team No. 51 as the average team with an SOS rating of 100, we increment proportionately the playing grades against higher ranked teams and proportionately decrement the grades earned against lower ranked teams. This is not left up to fallible human judgment. It is based upon ten years of statistical collection and categorization.
The second flaw in the committee’s thinking has to do with head-to-head competition. Like little girls neatly lining their dolls up on a shelf, the committee plugged Oklahoma into the No. 5 slot, one position ahead of Ohio State, whom they defeated, and two positions ahead of Penn State which lost to Ohio State. The symmetry is satisfying. Unfortunately, it is also wrong. We have Penn State No. 3 and Ohio State at No. 5 because the Nittany Lions’ one loss was on the road to a top 10 team by a single point while Ohio State’s loss was at home to a top 10 team by a wider margin. The names of the teams have no intrinsic meaning. More importantly, Penn State played better in its other victories than did Ohio State. Using the committee’s faulty logic, Iowa State should be ranked ahead of both Oklahoma and TCU.
The committee seems to find cosmic and everlasting meaning in beating a team head-to-head but we merely treat each game as another grade to average with all other grades. In fact, we would encourage the committee to mask the names of the teams whose resumes they are comparing and reveal the names only after the teams have been evaluated. That would eliminate some human bias.
As for Oklahoma, the Sooners have turned in a string of solid ‘B’ performances since their win in Columbus. Not bad, but not the playing performance of the fifth best team in America. We have Oklahoma at No. 12 with a high ‘B’ grade for the season.
The committee did get two things right: we agree that Georgia is now No. 1 and Alabama is No. 2 by the slimmest of margins (101.42 to 101.34) and we agree that the polls had Wisconsin egregiously overrated (at No. 4). Georgia graded out at an excellent 105.55 against the dumpster fire formally known as the Florida Gators while Alabama did nothing wrong with its bye week. The Tides’ problem is that the opponents it previously defeated continue to prove how weak they are. Vanderbilt, Arkansas, Tennessee and Ole Miss have all fallen out of the top 75 and that has hurt Alabama’s seasonal grade despite terrific playing performance grades against those teams. The committee may believe it severely punished unbeaten Wisconsin with its No. 9 ranking but we were even harder on the Badgers, leaving them at No. 13, one spot behind those Sooners. Wisconsin’s schedule has been the easiest among our top 16 and yet the Badgers have often played down to their competition.
You can find our top 16 rankings, including our surprise choice for No. 4, at http://nemosnumbers.com/football-rankings/. This week we replaced Michigan with USC, which played its best game of the year. Next week we intend to add Iowa State and possibly Virginia Tech, assuming it can defeat Miami.