College Football Playoff Expansion: A Modest Proposal (Eight is Enough)

 Now that the calendar has turned to June it is time to start looking ahead to the upcoming college football season. One topic that has made headlines over the past few weeks is that of a possible expansion of the college football playoffs. It was announced this week that consensus is building to expand the field from the current four-team format to a 12-team tournament.

Additional details of this proposal have begun to emerge. Notably, the top six conference champions would automatically qualify (which may or may not include all five Power Five champions, but which would necessarily include at least one team from the Group of Five conferences). The remaining six at-large teams would be selected by the Selection Committee, assumingly using a similar process to what is used today.

Furthermore, the teams would be seeded from one to 12, with the top four teams receiving a first-round bye and the teams seeded No. 5 to No. 12 playing in an opening round sometime in mid-December, most likely on campus sites or in regional NFL stadiums. It has also been suggested that the committee would prioritize matching teams up by seeds, regardless of potential rematches, even between conference opponents.

I have put a lot of thought into a possible change to the college football playoff and I must respectfully disagree with the proposed 12-team expansion. In fact, I have a counterproposal that I believe is fair, reasonable, and maintains the tradition of the current Bowl System, which I believe still has both value and merit.

How Big Should the Tournament Be?

The fundamental question that needs to be asked first is how big of a tournament should there really be in modern college football. In order to probe the impact of the size of the college football playoffs, I performed a series of calculations with a hypothetical group of 16 teams. Over the years, I have developed a power ranking system that can project point spreads for games between any pair of teams. These point spreads can then be used to calculate the odds that the favored team will win or be upset.

I looked back at the historical data and generated a list of hypothetical teams with an average strength for their historical rank. For this set of teams, the No. 1 ranked team was assigned a power index that is the average of the No. 1 ranked teams over the past 10 years. I assigned a similar average power index to the No. 2 team, the No. 3 team, and so on.

Based on this hypothetical list, I seeded and simulated a four-team, six-team, eight-team, 12-team, and 16-team tournament. I made the assumption that starting with the six team tournament, the champions of the Power Five conferences as well as the top ranked Group of Five conferences would all received automatic bids and any remaining slots would be filled by the top ranked teams still left on the board. I made some assumptions about the rankings of these teams based on the historical averages.

In the four-, eight-, and 16-team brackets, there are no byes and the teams all play the same number of games. In the six-team tournament, the top two seeds get byes, while the 12-team tournament gives the top four seeds byes, consistent with the leading proposal for expansion.

With the teams and structure of each bracket set, I crunched the numbers and calculated the odds for each team to win the National Title in each of the scenarios described above. Figure 1 below gives the result.

As expected, as the size of the tournament grows, the odds for each individual team drops. There are simply more potential winners, which dilutes the chances for each team. However, the odds do not change that much for each individual team as the size of the tournament increases. Based on these assumptions, in all cases the No. 1 team has between a 40-50 percent chance to take home the trophy.

Figure 1 suggests that an expansion to a six-team tournament may be worse than the current four-team format. In this case, the odds for the top two to win increase. The reason for this is the if there is an upset in the first round and one of the lower-ranked teams advances to the semi-finals, the path for the No. 1 or No. 2 is easier.

Once the tournament expands to eight teams and beyond, the odds start to stabilize for each seed. In the eight-team format, the No. 1 seed’s odds get a boost from being matched with the much lower ranked “last team in” (most likely the Group of Five champion). In the 12-team format, the top four seeds get a boost by virtue of getting a bye. But, in all cases, the odds for each seed do not change by more than a few percentage points as the tournament grows beyond eight teams.

In addition, Figure 1 suggests that based on historical data, any team outside of the top six or eight has very long odds to win the tournament. My data suggests that the odds of a team ranked No. 9 or below winning the tournament are about six percent in the 16-team scenario and 2.5 percent in the 12-team scenario.  These odds would be roughly equivalent to a between a No. 7 and No. 9 seed (or lower) winning the NCAA Basketball Tournament in March. It has happened, but it is rare (two occurrences in over 40 years).

For these reasons, I would argue that an eight-team college football playoff is certainly adequate, and most likely ideal.

A Modest Proposal

One of the aspects of the proposed 12-team tournament format that bothers me the most is that is seems to completely do away with the traditional New Year’s Day Bowl match ups and it seems to have the potential to undermine the entire Bowl system. Most importantly, the traditional Rose Bowl match-up of the Big Ten and Pac 12 Champions in Pasadena would mostly likely be completely washed away.

I must admit that I am a bit of a traditionalist, but for me, as a fan, this is essentially unacceptable. Furthermore, it is also completely and utterly unnecessary.

What I would propose instead of a 12-team format is an eight-team tournament that includes the champions of the each of the Power Five conference, the highest ranked Group of Five Champion, and two additional at-large teams.

However, I would also propose an unconventional twist. In setting up the bracket, the Selection Committee should prioritize traditional match ups more (and not less) in the first-round games, regardless of the rankings.

For example, the first round of the Tournament should always include the Big Ten Champ vs. the Pac-12 Champ in the Rose Bowl. The Sugar Bowl should pit the SEC Champ versus the Big 12 Champ, and the Orange Bowl should match the ACC Champion with the one of the at-large teams. The remaining game would be played in one of the other current New Year’s Six Bowls (the Peach Bowl, Fiesta Bowl, or the Cotton Bowl) on a rotating basis.

In this scenario, the Rose Bowl, Sugar Bowl, and Orange Bowl would all be played in New Year’s Day as part of a triple header. The fourth quarterfinal Bowl could be played in New Year’s Eve (or the nearest Saturday) like the current New Year’s Six format. The semifinal and Championship games could all be played in NFL stadiums in subsequent weekends in January. For the two remaining NY6 Bowls, the four “last teams out” would be matched up in these contests, which would also be played on or near New Year’s Eve, just like the current format.

Putting the Proposal to the Test

The best way to compare the various proposals outlined above is to explore how they would shake out based on a real set of data. In this case, it is possible to look back onto the last seven college football seasons (the current four-team playoff era) and see how the tournament would have been structured in each year.

Table 1 below summarizes the final Playoff Rankings from 2014-2020. The teams in bold text won their conference championship, while the teams in italic lost their conference title games. The teams shaded in green would have qualified for the eight-team tournament, while the four teams in yellow are the additional teams that would have qualified for the 12-team tournament. The teams shaded in darker green are the eventual National Champions.

Figure 2 below gives the specific brackets based on the three proposals that I am considering: my proposed eight-team tournament where traditional match-ups are preserved at the top, a more traditional eight-team tournament based strictly on seeds in the middle, and the proposed 12-team tournament at the bottom.

As we can see from Table 1, in an eight-team field, the top six ranked teams and seven of the top eight teams are always included. I have seen concerns raised that, in theory, a team could wind up ranked in the top five, yet still not make an eight-team tournament. In practice, this is virtually impossible and should not be considered a valid counterpoint.

Of the seven teams in the top 8 who would have been excluded in the hypothetical eight-team tournament, four of them (Florida in 2020, Wisconsin in 2019, Auburn in 2017, and Wisconsin in 2016) all lost in their conference championship games, and all but one of the teams wound up losing their Bowl Game. Only the 2016 Wisconsin Badgers won their Bowl, over Western Michigan. Essentially, all seven of those teams either had a chance or later proved themselves to not be worthy of a playoff spot. Based on the historical data, any argument that an eight-team playoff does not allow enough “access” for at large teams does not hold water.

Turning now to Figure 2, we can see how the specific match ups would have occurred in each year using each format. In my proposal, what happens in most years is that the No. 1 seed draws the No. 4 or No. 5 seed in the first round, and then often will draw a lower ranked winner of the non-contract bowl in the semifinals. Conference rematches are easily avoided.

In the other two formats, there are rare occasions of annoying conference rematches in the first round. For example, in 2019 in the 12-team format, No. 6 Oregon might have drawn No. 11 Utah two weeks or so after beating them in the Pac-12 title game. In 2016 in the traditional eight-team format, Michigan would have faced Ohio State in the first round. In 2014, an Egg Bowl rematch of the schools from Mississippi was on the docket.

These types of rematches are easy to avoid, and they absolutely should be avoided. As a fan, I do not want to see a rematch of a regular season game in an early playoff round. Ever. The current trend in the NCAA Basketball Tournament and now the College Football Playoffs to deemphasize this “bracketing principle” is as lazy as it is stupid.

As for the impact of non-traditional seeding and bracketing that I proposed, this can be easily analyzed using the same methods described above in reference to Figure 1. I ran the same calculations for each of the 21 tournaments shown in Figure 2 and compared the odds for each team to be crowned the eventual National Champion.

Instead of showing all seven plots, I show in Figure 3 the results for the 2017 tournaments. This is the year that showed the biggest difference in results between my proposed eight-team tournament with protected Bowl matchups and the proposed 12-team tournament. A glance back at Figure 2 shows why. In this particular year, the top four teams all play each other in the first round. The question is, does this actually impact the final results?

As Figure 3 clearly shows, it does not. The odds for the top four ranked teams do drop slightly, but by less than three percentage points for each team. This change does benefit the No. 5 and No. 6 seeds (Ohio State and Wisconsin) who both avoid a first round game in mid-December and get a relatively softer opponent on New Year’s Day, but at the end of the day, their odds go from three or four percent up to seven or nine percent. While that is clearly advantage, it is also a basically a negligible outlier. The data for the other six years show exactly the same trends and can be viewed here.

The Bottom Line

This analysis provides several new insights.

First, worrying about the exact rankings of the teams is stupid. Figure 3 shows us clearly that whether a team is placed in the position of the No. 3 seed or No. 7 seed or No. 2 seed, the final winner of the Tournament is likely to be unchanged. While this doesn’t mean that they we should not try to rank the teams, goals such as avoiding rematches and preserving traditional bowl matchups should be a much higher priority.

To further emphasize this point, look no further than the initial seed of the eventual National Champions. The No. 1 seed has only won the Playoff twice in seven years. Maybe, this is just because the No. 2 and No. 4 seeds just got lucky. More likely, though, is that the Selection Committee just didn’t know which team was the best. They made a good guess, but in over half the cases they were wrong.

From this point of view, the real purpose of the Selection Committee should not be to exactly rank the teams from No. 1 to No. 8 or No. 12. That, essentially, is a fool’s errand. What the selection committee should focus on, is making sure that the worthiest eight team in the country are included in the tournament. After that the teams will decide, on the field, who is best.

If the tournament field includes the five Power Five Champion, plus the best team from the Group of Five and two other teams, the historical data suggests that this is adequate. There would be complaints from teams that finished ranked No. 7, 8 or 9 in this format, but that is the case for tournaments of any size.

That all said, the powers-that-be have already decided that they can make more money from a 12-team tournament, logic and tradition be damned. Now, don’t get me wrong, I think that a 12-team tournament would be exciting. If nothing else, having No. 9 Georgia play at No 8 Wisconsin or No. 11 LSU at play at No. 6 Ohio State in mid-December would be a lot of fun.

But the question I have is what is to become of the traditional Bowl system? The larger a postseason tournament becomes, the remaining bowls become less and less relevant. What happens to the four teams that lose in the mid-December first round game? It seems likely that their season is over. But doesn’t this then effectively eliminate two high profile New Years Day Bowl Games or at the very least dilute down the participants? Doesn’t this have a negative financial impact?

While there certainly is money to be made from a playoff, it seems like there is also a lot of money that would be lost by downgrading the traditional bowls. If I were in charge, I would make a change to enhance the current Bowl system. The current collection of Bowl contracts often creates matchups which are either lopsided or simply not that interesting.

I would do away with the contracts and give the Selection Committee the authority to assign teams to any available bowl with a simple mandate: create interesting matchups. In my opinion, this would lead to a better overall college football postseason than would an extra round of unnecessary play-offs in mid-December.

At the very least, the Selection Committee should absolutely tweak the rankings a bit to try to create more traditional matchups in the semifinal rounds. In most years, a simple shuffling of the first-round pairing in the hypothetical 12-team tournaments would create matchups very similar to ones that I would propose in my eight-team brackets. Forcing the bracket to follow the exact seeding of the teams is clearly the weakest part of the new proposal.

While I love a good tournament, watching No. 2 Ohio State play No. 7 Baylor or (worse yet) No. 10 Penn State in the Rose Bowl would make me a bit sad. One of the best things about college football is the tradition. If that is thrown by the wayside, the sport is certainly diminished. Based on my analysis, we can have the best of all world. I only hope the people making the final decisions have the courage and intelligence to find a better solution than the current proposal.