clock menu more-arrow no yes

Filed under:

Ranking Teams In Order of Losses Isn't Going Away

New, 1 comment

At least, not yet.

Kevin Jairaj-USA TODAY Sports

While putting together yesterday's post on how the number of losses a team has determines its place in the rankings, I also wanted to look at how the practice of ranking teams more or less in order of losses has—or hasn't—changed.

After all, we've seen a few instances of high profile repudiations of loss-order ranking in recent seasons. The case of 12-1 Kansas in 2007 is a memorable one, as that team played a truly laughable schedule and finished seventh in the AP Poll behind six two-loss teams. Undefeated 2012 Ohio State finishing third in the AP Poll also comes to mind, but that team's postseason ban is to blame for that result.

The most important one was last year's undefeated Florida State ending up third in the College Football Playoff selection committee's rankings behind one-loss Alabama and Oregon.

There was some amount of controversy during the fall about that undefeated FSU squad falling behind one-loss teams, but it kept that unblemished record for so long in part due to being the luckiest team in the last 17 years, and it was summarily dismissed by Oregon in the Rose Bowl. Is it possible that what we saw last year is a sign that loss-order ranking is on its way out?

To test this, I looked at the same seasons as in the previous post, 2003 and 2006-14. I also again looked only at power conference teams—the 2014 Power 5 and BCS-era auto-bid leagues minus the 2013 AAC, plus Notre Dame—to eliminate issues caused by the discrepancy between how poll voters treat mid-majors versus the power teams.

I went through each year's poll to see how much of the time a power conference team finished immediately behind a team with an equal or lesser number of losses. If there was a mid-major ranked, say, tenth, then I skipped over it and compared No. 11's number of losses to No. 9's. Here is a plot of what percentage of rankings in each season's final AP Poll fit loss-order ranking, plus a three-year moving average to show the trend.

loss order ap

I left out the 2003 season from the graph because the gap between it and the 2006 season would mess up the moving average, but its percentage of rankings following loss order was 89.5%. That puts it well in line with the rest of these seasons.

The trend is going downward, but not quickly. After a few down years, loss-order ranking actually picked back up in 2013 and 2014.

I compared what the AP voters did in this regard to the final F/+ team ratings. Those ratings are the product of mathematical formulas, so they won't have been affected by human psychology.

Here is the result for the same seasons using F/+. I purposefully kept the Y-axis the same on both charts so you can better see the difference.

loss order fplus

The majority of F/+ ratings still follow loss-order ranking, which is fine. Better teams tend to take on fewer losses than worse teams. Still though, loss count alone doesn't tell you a lot thanks to the wildly variable scheduling that goes on in college football plus the impact of good or bad luck on a team's record. The bottom line is that the rate of loss-order following is noticeably lower in F/+ than it is in the AP Poll.

As far as the humans go, I found that the rate of loss-order following really hasn't budged. In every year I looked at but 2010, the number of rankings that broke loss order was either two or three. The percentage dipped in 2009-12 because it was both a run when loss-order violations were high-ish, going 3-4-3-3 in those years, while the number of non-power conference teams was also high-ish, going 5-5-4-4 in that same span. In 2006-07 and 2013-14, no more than three mid-majors made the final AP Poll. Aside from 2010, the numerator of these ratios has remained about the same, but the denominator swelled temporarily before deflating.

The past two seasons saw two and three loss-order violations, respectively, but only two and three mid-majors in the final poll, respectively. TCU and Utah getting called up to major conferences, plus Louisville getting rescued from a return to mid-major status by the ACC and post-MWC BYU falling off, are your culprits there.

As for the allegedly enlightened selection committee, its rankings had only two loss-order discrepancies: the famous FSU at No. 3 choice, and four-loss Auburn at No. 19 ahead of three-loss Louisville at No. 21 (Boise State was between them). The final AP Poll had the advantage of seeing bowl results, and it had only three loss-order discrepancies. Two were basically the same—one-loss FSU in fifth behind two-loss Alabama in fourth, plus four-loss Louisville again behind five-loss Auburn—while the third was national championship game loser Oregon and its two losses ending up ahead of one-loss TCU.

I don't think that loss-order ranking is going the way of the dodo in polls overall. It might be dying in the very limited context of the teams in the running for the top four spots at the end, but that is a continuation of the old BCS-era pattern of voters not caring much about ranking order after the top two. That's how we got results like, for instance, 2011 Oregon being penalized for scheduling then-undefeated No. 1 LSU out of conference and winding up two spots behind a one-loss Stanford team it smoked in the regular season's final AP Poll, with the Ducks' only other loss coming at the hands of the USC team sitting in between them and the Cardinal in the rankings.

We'll have to see how things go with the committee in the future, and its patterns might evolve as members come and go. However, I haven't seen anything yet that suggests that the choice to put 13-0 FSU at No. 3 is a part of a broader pattern of voters going against loss-order ranking.