The official college football season kicked off on Saturday with seven games in "Week Zero" action. Next week, the rest of the college football nation will take the field for the first time in 2023. The wait is officially over.
During the summer, I have been spooling out the results of my annual preseason computer simulation of the entire season. So far, we have reviewed the accuracy of the preseason rankings and reviewed the results from 2022. In addition, I have completed an analysis of the regular season projections for the Michigan State Spartans, the Big Ten, the rest of the Power Five, and the Group of Five.
Today, it is time to assemble all of those pieces in order to project what these simulations and scenarios mean for the college football post season. Which teams will make the college football playoffs? Which teams will fill out the rest of the New Years Six Bowls? We can attempt to answer those questions using the data presented so far.
Review of Conference Results
The results of my 100,000-cycle Monte Carlo simulation and related calculations for each conference race are summarized below in Table 1. Here I highlight the two teams projected to play in each conference championship game on the first weekend of December.
 |
| Table 1: Summary of the predicted Championship Game participants and odds for each conference with the projected winners shown in bold. The right side of the table gives the predicted results based on the "disruptive" scenario including the projected final records and key wins/loses. |
This table is broken into two sections. The left side of the table gives the results of the simulation. It shows the two teams from each conference with the best odds to advance to the conference championship game. The projected winner of that game is shown in bold.
The right side of the table gives the same information based on my "disruptive scenario." This scenario projects the results of each game assuming a historically reasonable number of road upsets. In many cases the conference outcomes are the same. The cells highlighted in yellow show where the results of the two scenarios differ.
In almost all cases, though, the disruptive scenario gives a different final record for each team, and the distribution of win-loss records is more consistent with historical results. For this reason, I use the disruptive scenario to make my "official" picks for the 2023 postseason. The right side of table 1 gives the notable wins and losses for each team in the disruptive scenario. This information is helpful when comparing the teams' theoretical resumes.
Playoff and New Years Six Bowl Projection
If the results of my disruptive simulation were to occur, the following teams would be likely candidates for the college football playoffs and New Years Six Bowls. For the teams not listed in Table 1, the list below gives the notable wins for each team.
- Georgia (12-1), SEC Champion
- Penn State (12-1), Big Ten Champion
- Florida State (11-2), ACC Champion
- Tulane (13-0), AAC Champion
- USC (11-2), Pac-12 Champion
- Texas (11-2), Big 12 Champion
- Michigan (11-1), with a loss to Penn State, and a win over Ohio State
- Notre Dame (11-1), with wins over USC and Ohio State and a loss at Clemson
- Alabama (11-2), SEC runner-up
- Clemson (11-2), ACC runner-uo
- Wisconsin (11-2), Big Ten runner-up
- Oklahoma (11-2), Big 12 runner-up
- Oregon State (11-2), Pac-12 runner-up
- Iowa (10-2), with loses to Penn State and Wisconsin and a win over Minnesota
- Kansas State (10-2), with loses to Texas and Texas Tech and a win over TCU
- Texas Tech (10-2), with loses to Texas and Baylor and wins over Oregon and Kansas State
- Washington (10-2), with loses to USC and Oregon State and wins over Utah and Oregon
In this scenario, there are four Power Five teams with just one loss, 12 Power Five teams with two losses, and a single undefeated Group of Five team.
In this case, the top two seeds in the College Football Playoffs seem clear: Georgia and Penn State as one-loss Power Five conference champions. But the other two seeds are more up for grabs.
Two clear candidates would be the two remaining one-loss teams: Michigan and Notre Dame. Undefeated Tulane would also deserve some consideration. Then, the question would be whether any of the 12 Power Five who finished at either 11-2 or 10-2 are deserving of a top four final ranking.
One could make the argument for Florida State, Texas, or USC as conference champions, but based on strength of schedule and key wins over Texas, LSU, and Tennessee, I believe that Alabama is the only team from the group of two-loss teams that would finish in the top five. Tulane's weak strength of schedule would also eliminate the Green Wave from consideration.
With this in mind, the challenge would be to compare the resumes of Notre Dame, Michigan, and Alabama and to pick the top two teams. A glance back at the strengths of schedule for the three teams reveals that Alabama has the toughest overall schedule at No. 3 nationally. Norte Dame (No. 32) has the second hardest schedule in the group with No. 57 Michigan bringing up the rear.
A look at the key wins and loses tends to support this conclusion. Alabama would have four wins over top 25 teams. Norte Dame and Michigan would both have wins over Ohio State. But, Notre Dame would have a win over USC, while the second-best win on Michigan's schedule would likely be a road win at Minnesota.
Based on this logic, I would predict the college football playoffs to end up as follows.
Sugar Bowl: No. 1 Georgia versus No. 4 Notre Dame
Rose Bowl: No. 2 Penn State versus No. 3 Alabama
As defending champions, Georgia is the natural pick for the No. 1 seed and Alabama's strength of schedule would give the Crimson Tide the nod over Notre Dame. This pairing also eliminates a rematch of the SEC Championship game in the national semifinal, which the selection committee would likely prioritize.
The remaining New Year's Six Bowl games would have to include Florida State, Texas, USC, and Tulane as conference champions. Michigan would also be included. This year's New Year's Six rotation had only the Orange Bowl with any contractual obligations. Florida State (as ACC Champion) and Michigan (as the top ranked SEC or Big Ten team on the board) would be selected to play in Miami.
Next, it would make sense to pair USC and Texas and the Fiesta Bowl seems like a natural geographical fit.
The remaining challenge would be to identify the three two-loss teams that are likely to finish the season ranked the highest in the disruptive scenario. These three teams, along with Tulane, would be placed in the Peach and Cotton Bowls.
In this case, it looks like there would be four prime candidates competing for those three spots: Clemson, Oklahoma, Wisconsin, and Oregon State. All four teams are projected to finish the regular season with one loss, followed by a second loss in their respective conference championship games.
Clemson has a slightly better resume than the other three teams and is starting the season higher in the rankings. Oklahoma, Wisconsin, and Oregon State all would have similar resumes including quite weak strengths of schedule. On balance the Badgers' strength of schedule and preseason ranking is slightly below the other two teams. Wisconsin would also have the worst loss (on the road at Minnesota). Therefore, I would project the Badgers to be the last team out.
The remaining New Years Six Bowls would therefore line up follows, with my projected winners shown in bold.
Orange Bowl: Michigan versus Florida State
Fiesta Bowl: Texas versus USC
Peach Bowl: Clemson versus Oklahoma
Cotton Bowl: Tulane versus Oregon State
As for the playoffs, my calculations project that Georgia and Alabama would both win in the semifinal round, forcing a rematch of SEC Championship Game for all the marble. My computer predicts that Georgia would win that game as well, securing the elusive "three-peat" for the Bulldogs.
With the full analysis of the college football season now complete, there is just one topic left to explore. I have used my simulations to generate a slew of odds for a variety of full season outcomes. It is straightforward to compare these odds to similar values generated using the betting line from our friends in Las Vegas.
Is it possible to use the results of my simulations to perhaps place a few educated wagers? That is the subject of the final installment of this series, coming soon to a screen near you. Stay tuned.
Comments
Post a Comment