In Part One of this series we expressed our concern about the limited amounts of data that exist for a CFP selection committee who is tasked with comparing teams from different conferences who rarely face one another. To this point, the committee has relied on the precedent of using losses as the primary criteria for selecting playoff participants. Our analysis detailed the small number of games that are played between Power Five (P5) conferences during the regular season. In our opinion, the sample size is too small for the committee to draw meaningful conclusions about the relative strengths of different conferences.
The current wave of conference realignment has ensured a future where leagues will be less balanced than ever before. The lack of P5 vs P5 games being played in modern college football means we need to look further to find a reliable way of comparing teams who often have no common opponents. The committee must find a way to compare these teams without rewarding schools who play in weak conferences and punishing ones who play in deeper leagues like the SEC and Big 10.
In Part Two, we explore the best available alternative data in the absence of a robust number of out-of-conference games. In Part Three, we will propose a metric that effectively utilizes this data.
Part Two - The Data Solution
It is important to contextualize the past 25 years before delving into our Net Yard Per Play analysis. We have data from sixteen BCS title games and nine years of CFP games. Here is a look at the outcomes on both a team and conference basis. It is worth emphasizing that the CFP selects individual teams and not conferences, but when considering win-loss records it becomes evident that teams from the SEC have significantly outperformed the teams from all other leagues.
Between the CFP and BCS systems, 43 games have been played over the last 25 years.
Within this context, the SEC has separated itself from the other conferences, capturing 15 out of the last 25 national championships. This impressive record includes 6 out of 9 CFP titles and 9 out of 16 BCS titles. Despite the focus on Nick Saban's Alabama dynasty, it is noteworthy that six different SEC programs have won titles in the BCS/CFP era. Four of those six programs have captured multiple national championships.
In the CFP-era, the SEC has outperformed all other conferences combined. SEC teams have won sixteen of the 27 total games that have been contested since the College Football Playoff came into existence in 2014. Remember that Alabama and Georgia have faced each other twice in the CFP. Excluding head-to-head matchups between SEC schools, the collective record of SEC teams in the CFP stands at an impressive 14-3.
In contrast, teams from the Big 10 and Big 12 have frequently exited in the semifinal round. In 13 combined playoff appearances, teams from the Big 10 and Big 12 have suffered semifinal losses 10 times.
With the impending inclusion of Oklahoma and Texas, the SEC’s dominance is poised to grow further. With them in the league the SEC will be home to 8 of the 14 programs who have won national titles in the last 25 years. Tennessee’s 1998 national title will fall out of the 25-year window following this season, so barring a surprise national championship run from the Vols, the SEC will have 7 of the 13 teams to win national titles in the last 25 years when this most recent wave of realignment goes into effect in 2024. No other conference will have more than three programs who have won a national title in the last 25 years.
We understand that the figures presented above do not offer groundbreaking insights nor any new information. Consequently, we are hesitant to provide seemingly obvious data, especially when it lacks actionable implications. Historical wins and losses do not readily translate into a metric for future analysis. Nevertheless, these results create a backdrop that prompts an essential question…
"Why do teams win games in the CFP?"
Net Yard Per Play (Net YPP) is a metric that answers that question clearly. The stat combines yards per play and yards allowed per play by deducting the number of yards a team allows per play from the amount it gains on offense.
The formula for winning football comes down to being more efficient with your possessions than you allow your opponent to be. Things tend to go well for a team who is gaining more yards per play on offense than it is allowing on defense. Factors like turnovers, penalties, and Special Teams touchdowns introduce variability within a single game, but they are not reliably repeatable over the course of a season.
Consistently outgaining an opponent on a down-to-down basis remains the most reliable method to secure victory, which is why we emphasize Net Yard Per Play. Why should you trust Net YPP? Here are a few Net YPP statistics from past CFP games…
- In eighteen semi-final games, the team with the higher Net YPP is 17-1.
2022 Michigan, which had a +0.07 Net YPP against TCU is the lone loss
- Excluding the last offensive play in victory formation, TCU had a higher NYPP than Michigan. Thus, it was effectively a tie from a Net YPP perspective.
- In 9 CFP championship games, the team with the higher Net YPP is 7-2.
- There have been 25 CFP games where a team had a +0.15 NYPP advantage. Those teams are 24-1.
The CFP selection process should logically look for teams with the potential to achieve a higher Net YPP than their opponents. So, how should the committee determine those teams?
Our hypothesis is that the teams with the highest NYPP against Power Five (P5) competition prior to the CFP are most likely to perform best in the CFP.
In the 9-year history of the playoff, 20 teams have had an +2.00 Net YPP prior to the CFP… 17 of those teams were selected for the four-team field. The exceptions were 2015 Ohio State, 2016 Louisville, and 2019 Alabama. It should be noted that the 17 teams selected were likely chosen due to factors other than Net YPP.
Additionally, the committee has selected fourteen teams with a Net YPP of 1.00-2.00, and five teams with a Net YPP less than 1.0. Overall, the average team selected for the CFP had a pre-CFP Net YPP of +1.92.
When we examine the performance of teams in the CFP based on their P5 Net YPP, we observe a significant difference. The seventeen teams with a P5 Net YPP of +2.0, they have accounted for seven CFP titles, four runner-up finishes, and six semi-final losses. On the other hand, all five teams with a Net YPP of less than +1.0 lost in the first round by an average margin of 28 points. We should stop here and make this point very clear… Putting a team with a P5 Net YPP of less than 1.00 into the playoff is the most reliable way to produce a blowout. These <1.00 Net YPP teams were 2014 Florida State, 2015 Michigan State, 2017 Clemson, 2020 Notre Dame, and 2021 Cincinnati. Two of the teams with a Net YPP below +1.00 were undefeated prior to the CFP but lost to one-loss opponents with a higher pre-CFP Net YPP.
CFP teams with a Net YPP greater than +2.0 hold a 9-2 record against teams who entered the game a Net YPP of less than 2 yards.
As for teams with a Pre-CFP NYPP between +1.0 and +2.0, their results fall in between the two groups. Out of the fourteen teams in this category, two have won CFP titles (2015 Alabama and 2016 Clemson), five have finished as finalists, and seven have lost in the semi-finals. In other words, this group hasn't produced many champions but generally wins about half of its semi-final games.
Nonetheless, the question we want to answer is how do they perform on a Net YPP basis relative to their pre-CFP results?
The data above supports our hypothesis that teams with a higher Pre-CFP Net YPP (remember this is versus P5 teams only) are likely to produce a higher Net YPP in a CFP game. These teams not only generate a higher Net YPP relative to other teams selected to the CFP, but also maintain a smaller difference from their Pre-CFP Net YPP levels.
29 playoff games have been played by teams with a Pre-CFP Net YPP of +2.0 or higher. In those 29 games, their average Net YPP output was +0.74, with a difference of -1.81 from their Pre-CFP levels. If we exclude the nine games that featured two teams with a Net YPP greater than +2.0, the average Net YPP of the >2.0 teams is +1.93 across 11 games. This shows only a modest decrease from their pre-CFP average, and marks an important threshold.
The poor performances of Lincoln Riley-led Oklahoma teams who entered the playoff with a +2.0 Pre-CFP Net YPP significantly skewed this data. In CFP history, no team with a +2.0 Pre-CFP Net YPP has lost a semifinal <2.0 Pre-CFP Net YPP other than Oklahoma. All other +2.0 teams who have lost in the playoff semifinal did so to other +2.0 teams.
In a twelve-team playoff format, P5 teams that don't win their conferences while still producing a +2.0 Net YPP are likely to secure at-large bids. For example, 2015 Ohio State would have certainly been selected in such a scenario. Both 2019 Alabama and 2016 Louisville were ranked outside the top twelve on selection day, but their Net YPP credentials were strong. 2019 Alabama had an impressive +2.43 NYPP, with their only losses coming by narrow margins to LSU and Auburn.
Despite a +2.3 Net YPP, 2016 Louisville was ranked thirteenth due to its three losses, one of which was a blowout to non-P5 Houston. An FSU team that Louisville defeated 63-20 in a game where they produced a +3.8 Net YPP margin was ranked eleventh on selection day. Such discrepancies should receive more scrutiny in a twelve-team CFP, as the rankings of teams outside the top four selected to the CFP were largely perfunctory in the first nine years.
Since the inception of the CFP, there have been approximately 75 teams with a Net YPP between +1.0 and +2.0. Over the last 9 years, 60 of those teams did not make the CFP (about 7 per year). Simple math suggests that such teams are likely to secure bids in the future. Less than half of them have produced positive results in the current 4-team format, but the dilution of the field should create more matchups between 1.0-2.0 Net YPP teams. The committee's challenge will be to determine which of these teams deserve at-large bids. Superficially, the conferences do not seem to provide a clear answer on a Pre-CFP basis.
As mentioned earlier, the average team selected to the CFP has a Pre-CFP Net YPP of +1.92. While this is slightly below the +2.0 threshold, it is not far off and falls within the range of the P5 conferences. With a range from a high of +2.2 to a low of +1.6, the teams selected from P5 conferences have statistically similar profiles prior to the CFP. If we exclude Michigan State's appearance, which could be considered an outlier, the other seven participants from the Big 10 have averaged a +1.9 Net YPP, aligning with the overall average. However, as shown in the chart below, Notre Dame and Cincinnati have produced steams who are statistically inferior.
If teams have a similar statistical profile prior to the CFP, how can the committee discern which teams are likely to perform well in the CFP?
With 9 years of data, we can see how well teams have matched their pre-CFP performance. Before discussing the data and analysis, it is important to consider the differences between CFP games and those prior to it. The average CFP opponent is likely more challenging than the average P5 opponent faced in regular season or conference championship games. Consequently, we would expect teams' average Net YPP in the CFP to be significantly lower than their pre-CFP levels. The question is, to what extent?
The table below demonstrates that actual CFP performance varies notably by conference. SEC teams, for instance, have a Net YPP in CFP games that is approximately -0.85 lower than their Pre-CFP Net YPP, while Big 12 teams have shown a decline of almost -4 Net YPP in the CFP. Excluding the two games between SEC teams, the SEC's average Net YPP in the CFP is 1.52, which is only 0.7 lower than the conference’s combined Pre-CFP Net YPP. The output of SEC teams is significantly closer to their regular season performance than teams from other conferences.
The most striking data points are the performances of the Big 12 and Notre Dame. These teams outputs have been effectively four yards worse than their pre-CFP Net YPP.
- Producing a Net YPP of -2 to -3 makes it nearly impossible to be competitive, which explains why these teams have a collective record of 1-8 in CFP games.
- The Big 12 hasn't had a team with a positive Net YPP output in any of its five CFP appearances.
- The ACC’s Net YPP performances have been about 2 yards worse than their regular season average, resulting in a CFP Net YPP output that is about 0.
Big 10 teams have performed better than the Big 12 and Notre Dame, but only one of their seven CFP teams had a significantly positive Net YPP. That was 2014 champion Ohio State with a +0.45 CFP Net YPP
- In 6 CFP appearances since Ohio State's 2014 title, the highest Net YPP output of a Big 10 team in a single playoff was 2020 Ohio State’s +0.35.
- Excluding TCU's final play in victory formation would have turned Michigan's +0.07 NYPP advantage into a -0.05 deficit.
- With just two appearances in the CFP, it is difficult to conclude much from the Pac 12’s performances.
Net YPP% tells a similar story
Another way to evaluate a team’s Net YPP is to compare its performance against their opponents relative to their opponents’ average for the year. We refer to this as NYPP%.
The average playoff team should perform better against its CFP opponents than the average of its CFP opponent’s opponents on both offense and defense. That has proven to be true, as the P5 conferences have a positive NYPP% in the College Football Playoff.
A couple of noteworthy data points... First, the SEC is the only conference who had a higher NYPP% against their CFP opponents than they did against their pre-CFP opponents. What does that mean? It means SEC teams are more dominant against their CFP opponents than they are against their regular season foes.
Equally significant, Notre Dame and Cincinnati performed WORSE than the AVERAGE opponents of the CFP teams they faced.
Total head-to-head yardage also supports our analysis. When we examine how conferences perform against each other, it paints a similar picture. The table below illustrates the head-to-head yardage breakdown, read from right to left. For instance, the ACC has been outgained by the Big 10 by 39 yards across three games, but in aggregate, the ACC has outgained all of its opponents by 236 yards. In the 17 games played against teams from other conferences in the CFP, SEC teams have outgained their opponents by a combined 1,948 yards. In doing so the SEC has established a positive yardage differential against every conference. In contrast, the ACC is the only other conference boasts a positive overall yardage differential.
While the total yardage figures above are stark, they are volume related and impacted by the number of games played. Focusing on efficiency or what happens on a per game basis is always more relevant.
- In an average playoff game, teams from the SEC outgain their opponents by roughly 115 yards. That figure includes blowout wins (UGA-TCU, LSU-OU), close wins (2015 Alabama-Clemson, UGA-Ohio St) close losses (2016 Alabama-Clemson) and a single blowout loss (2018 Alabama-Clemson)
- The ACC and Big 10 (read: Clemson and Ohio State) have basically been even against each other in their three contests. Those include one blow out win for each program (2016 & 2020) and a closely contested game in 2019.
- The Big 12 and ND/Cincinnati were significantly outclassed no matter who they faced.
- The Pac 12’s sample size is too small to draw any meaningful conclusions, as it has only had two participants in the CFP and none in the last six years.
What does it all mean?
Modifying the the initial CFP contract that all of the conferences agreed upon prior to its expiration required unanimous support of all ten FBS conferences. In addition to the expansion to twelve teams, there were two structural changes made to obtain the full consensus required.
First, the CFP will provide six automatic bids. Those will go to the six-highest ranked conference champions. This is a departure from the current format where all teams are "at-large" selections.
Second, the top four ranked conference champions will be the first four seeds and receive first round byes. To be clear, these changes technically only apply for the first two years of the expanded format (2024 & 2025). After that, the initial contract expires and this criteria can be revisited when the next CFP contract is negotiated.
The agreed upon changes run counter to how the data above suggests CFP teams should be selected. The net result is that there will be even more pressure to select the six at-large teams correctly. There will almost certainly be incorrect seeding. The field will also include a G5 team that is likely to have a significant talent deficit compared to its CFP opponents.
Correctly selecting the at-large berths requires adopting a different analytical framework than what has been used to date.
In Part Three of this series we provide a framework for the CFP selection committee to generate a Projected Net Yard Per Play output that will allow them to compare different teams from different conferences. Our hope is that it catalyzes change.