Epilogue To The Statistical Case For Selecting Alabama To Be In The College Football Playoffs

The CFP selection committee released their second to last ranking on November 29. We had hoped that ranking would place Alabama #5 in the poll. In our November 28 article entitled “The Statistical Case For Selecting Alabama To Be In The College Football Playoffs”, based on its strength of schedule and cumulative scoring, we made a compelling case with the numbers that the Crimson Tide should actually be ranked #4. That article can be accessed here. Alas, the CFP chose to ignore our detailed analysis and put Alabama #6, with Ohio State given the much sought after #5 slot. That puts Ohio State in the best position if there is any shakeup among the top four resulting from their conference championship games this weekend.

While all four are playing, defeats would likely knock out only two teams from the playoffs. Georgia is rated #1 and is favored by 17.0 points over LSU, while Michigan is rated #2 and favored over Purdue by 16.5 points. Either would need blowout losses by a wide margin to even be at a risk of being knocked out. But if they lost and stayed in, it could still have some effect based on the dynamic process that should occur from the re-calibration of rankings based on how opponents perform. For instance, if LSU were to beat Georgia by a lot more than the 1 point difference LSU beat Alabama in overtime, that should in fact improve the relative standing of the Crimson Tide. Similarly but in a different direction, if Purdue were to beat Michigan, that should lead to a reduction in Ohio State’s relative standing as the team that just soundly defeated them isn’t as strong as previously thought.

The two teams in the most vulnerable position to be knocked out are #3 TCU and #4 USC. TCU is a 2.5 point favorite over #10 Kansas State and USC is a 3.0 point favorite over #11 Utah. Most believe that a close loss by undefeated TCU gives it a shot at staying in the playoffs, but that’s likely not the case with USC. Meaningful losses by both of those teams would likely have the effect of lifting both Ohio State and Alabama into the playoffs. As is clear, however, Ohio State is in a more favorable position going into this last weekend. That is why the #5 position in the most recent ranking was as coveted as it was. While there is still a narrow path for the Crimson Tide to make it into the playoffs, for the most part that requires both TCU and USC to lose this weekend, with the loss at TCU being formidable.

However, while that narrow path is Alabama’s most likely way into the top four, it isn’t the only way. There is the slimmest of possibilities that even if just one of those teams is knocked out they can still get in if Ohio State and Alabama in effect switch relative positions. We don’t know how close those teams were in the rankings that just came out Tuesday, but if they were very close its conceivable that what happens in the Georgia or Michigan games can lead to a change in their relative positions. While neither play, as noted above if LSU is viewed as stronger or Michigan is viewed as weaker it could tilt the assessment of Alabama upward or Ohio State downward. This would however require the CFP selection committee to go deeper into analytically ranking the strength of schedules of the various teams, something that they have appeared to be disinclined to do so far.

While the CFP committee has been quoted referring to schedules, it tends to always be anecdotal references to various games and not the sort of systematic evaluation of all games that is now available from credible data that quantifies strength of schedule. That’s unfortunate and seems to be inconsistent with the CFP’s mission when it was established in 2014 with the singular goal of matching the four best teams in an end of year playoff. It’s rankings only start midyear and weren’t meant to duplicate the other polls. Indeed, given their goal, they were supposed to be a ranking that dug deeper than just the win-loss record and avoided the regional bias often seen in polls. With its members consisting of athletic directors and retired college coaches and players, they had the knowledge without the inherent bias of current coaches and this was expected to have them take into account factors the polls didn’t and be different. After all, if it wasn’t expected to be different, why not just say that the CFP ranking will be the same as one of the popular polls or an average of a group of them. In short, the CFP should be less about narrative and more about analysis, with the latter giving them confidence to be different from the polls.

The latest CFP ranking is almost exactly the same as the latest AP ranking as well as the latest USA Today Coaches ranking. That suggests a fair amount of multicollinearity in the data that isn’t a good sign. The first six positions are exactly the same with some minor inversions in the remaining four positions. In fact, there is only one team in each of the polls in the ninth position that isn’t in the CFP ranking. Of course, the only spots that really matter in the CFP ranking are the first four. A case can be made to start with the top ten at midseason to chart the runners up, but why do they presently rank the top 25 like the polls do? Who cares? The CFP isn’t supposed to be like the polls, yet that is what it looks like now. It comes out Tuesday night well after the polls have been released. How much is it influenced by those polls? Is there a conscious desire to make the CFP look like the polls?

Despite the narrative and the anecdotal stories of individual games, it’s clear that the simple win-loss record remains the overriding metric used by the CFP selection committee. That is evident in its most recent ranking where the first three spots are the undefeated teams, the next two are one loss teams, the next four are two loss teams and the top ten are rounded out by a three loss team. The preference for that metric was just confirmed by committee chairman Boo Corrigan when he said this week, “So obviously winning these games does matter. It’s something we look at, it’s something we talk about more than anything else.”

I’m certainly not saying that win-loss record isn’t an extremely important metric. But when you start with the top teams and you’re focused on getting the best performing ones into a systematic playoff, there is a broader range of factual information that should be taken into account. Factual data such as total points scored versus points given up, the ratio of those two and the strength of schedule in terms of opponents are all quantifiable statistics, just like the win-loss record is. While they all tend to broadly correlate in some way with win-loss records, they augment and in many ways present a more refined measure of that key statistic. When selecting teams for a playoff, if you want the best teams and the most exciting playoffs, it is imperative that this broader range of information be considered.

In particular, the strength of schedule data provides a degree of difficulty factor, much like the multiple used in assessing the sport of diving, that should be layered on top of other measures. It seems to me that it should specifically be part of a detailed analysis and not just part of a narrative with anecdotal references. The TeamRankings strength of schedule factors for college football are updated each week and at any given time are meant to measure the entire schedule, so they are comparable to other cumulative measures such as win-loss and scoring. The latest factor for Alabama is 13.7, again putting it at the top of all 131 FBS college football teams. That factor is 21% above Georgia, 29% above Michigan, 34% above TCU, 163% above USC and 18% above Ohio State. That is worth reflecting on, particular as it relates to TCU and USC. Both of those teams had scoring differences versus opponents more than Alabama did, despite the fact that they had quantifiably easier schedules, most particularly in the case of USC. Those facts are meaningful asterisks on their win-loss records that should make that singular metric mean less.

The following table includes an updated version of the information in the previous article taking into account the latest games and rankings. It also shows Alabama as ranked #4 based on the average ranking across all four factual metrics utilized. The bottom half of this table goes a step farther than the ordinal ranking based on the average of four rankings in the top half. It takes the two metrics based on cumulative scores, both the average difference per game and the ratio of points scored to points given up, and multiplies both by the strength of schedule factor to come up with a measure directly adjusting those cumulative scores based on the precise schedule difficulty. By definition, that takes into account the schedule difficulty involved in all games and not just ones that fit into an anecdotal narrative. If there are better specific metrics on strength of schedule than the ones issued by TeamRankings, those could be used. However, its difficult to argue with the analytical logic of taking full season performance measured by cumulative scores and adjusting by schedule difficulty to come up with a more apples to apples cumulative score comparison. Those adjusted score metrics directly incorporate Alabama’s clear strength of schedule superiority and that shows up in its ordinal ranking in both. The win-loss ratio doesn’t lend itself as much to such adjustment, so the ordinal rankings from it are included to come up with an adjusted ranking based on the average of three rankings. From there, a new overall ranking with Georgia #1, Michigan #2, Alabama #3 and Ohio State #4 is developed as shown in the table below.

I believe such a ranking as of today more appropriately takes into account Alabama’s extraordinary strength of schedule and above average point spread per game. I also suspect that Alabama’s two losses are apparently viewed by some who view the win-loss record as paramount as disqualifying it for the playoffs because there has never been a two loss team in the top four. Alabama lost its two games by a cumulative of 4 points in less than that number of seconds left in those games. As I noted in my previous article, they were certainly both losses, but ones that have large asterisks on them based on the specific facts and circumstances. The 1 point loss to LSU in overtime is particularly galling with a new video showing that an Alabama linebacker was blocked in the back, impeding his ability to stop LSU’s game winning touchdown. Had officials called that illegal block, it would likely have thwarted that touchdown and the successful 2 point conversion afterwards to win the game by 1 point. I could also point to a controversial call in the Tennessee game that nullified an interception by Alabama that would have likely sealed the game for the Crimson Tide, but that would seem to be just making excuses.

We are where we are today, and the only way the Crimson Tide will make it into the playoffs is with some unexpected outcomes in the conference games this weekend. My hope is that this analysis makes it’s way to folks on the selection committee and that in the event of some shakeups, they give some consideration to all the points made here. I do know that if events work out where Alabama is selected, that it will give the country some great games in the playoff. The talent in the players and the coaching staff and time to prepare almost guarantee that outcome. Alabama has a 9–4 win-loss record in the CFP. While a 9–4 regular season record wouldn’t get a team into the present CFP top ten, that is quite an impressive record when you know its in the playoffs matched against the best. It’s all about context, and Alabama’s regular season strength of schedule doesn’t seem to have been given the consideration it deserves. With 9 wins in playoff games, Alabama has 50% more wins in those games than Georgia and Ohio State combined. Performance when it matters most is the mark of champions.

Notice that I said if the Crimson Tide is selected, I referred to playoff games in the plural. That is because I’m confident that if that occurs, Alabama will win both the semi-final and final games to emerge as college football national champions for the 19th time. That’s what real champions do. Roll Tide!

Jock McCown is a native of Alabama who now lives in Pound Ridge, which is an hour north of New York City. He is an internationally recognized expert on container shipping and maritime commerce. When not engaged in matters related to his vocational field, as a biking enthusiast he will frequently be practicing his favorite transport avocation riding among the bucolic hills of northern Westchester County.