Peach Bowl Thoughts

While we will have more to discuss as the Peach Bowl draws closer, our initial take is rather consensus.  Against inferior opponents, Ohio State tends to “out-athlete” the competition and accumulate impressive statistics, but when faced with ranked opponents, its overall production falls significantly.  The severity of the splits is what is most revealing.  Even if one assumes away the Michigan game, Ohio State is just a +1 NYPP team against ranked opponents, which is good but not great.  By contrast, UGA is 2.43 NYPP against ranked teams. 

This is most visible in CJ Stroud’s passing statistics verses ranked opponents in 2022.  8.1 ypa and a near 150 passer rating is good, but it is hardly great.  Further, it is really a combination of two average games and one exception game against PSU.  While PSU boast a UGA-quality CB in Joey Porter Jr, they also have a defensive coordinator know for aggressive blitzing and allowing big plays. UGA’s defense is significantly more athletic and complex than any of the defenses Stroud has faced.  If he doesn’t produce a game with a 9+ YPA, it will put extraordinary pressure on Ohio State’s defense.

Obviously, if the QB is less efficient against ranked teams, it stands to reason that his primary WRs would be less productive as well.  Of Ohio State’s WR’s only Harrison was equally effective against ranked teams as non-ranked teams, but Egbuka and Fleming were not.

Championship Week Update

There is a reason that analytics and football don’t always go together.  Sample sizes of total number of games are relatively small, such that statistical outliers from peaks and valleys often impact that season long averages.  (More on this later).  Further, some of those peaks and valleys can occur within a single game and can create anomalies.  As we all know, “garbage” time results are usually not indicative, as teams simply on both offense and defense with a priority on running out the clock.  Thus, some of the advanced statistical systems screen out garbage time statistics to eliminate the effect of these anomalies. 

The SEC CG contained one of those anomalies, where garbage time significantly altered the statistical picture of the game.  In building a 35-7 lead, UGA gained 290 yards at a 7.8 ypp while allowing LSU 140 yards at a 5.0 ypp for a NYPP of 2.8.  From that point forward, LSU gained 409 yards at 9.3 YPP thereafter whereas UGA gained just 239 yards at a 6.3 ypp.  The contrast in passing couldn’t be more stark, as LSU gained 345 yards on just 26 pass attempts, while UGA attempted just five passes over that same time period.  The impact of that “garbage” time was significant because it reduced UGA’s NYPP% from 38% to 32% FOR ALL ITS P5 GAMES.  In short it was a tremendously outlier as indicated below. 

While we realize football games consist of two halves of 30 minutes, it would be foolish to infer much from UGA’s second half.  To be clear, there were coverage breakdowns mistakes that must be addressed, but the notion that UGA’s performance against LSU upon building a 35-7 lead is predictive of what we will see on 12.31 isn’t compelling to me.

Net YPP and Relative Net YPP

As we concluded the regular season and championship week, UGA has a sizable NYPP gap over its playoff competition as well as the rest of CFB, and again, omitting the second half of the LSU game would take UGA’s relative NYPP to 2.24, which is a sizable gap to Ohio State and Michigan and more than double TCU.

Strength of Schedule Movements and Dissection

As we updated our numbers in recent weeks, Michigan’s and Ohio State’s opponents NYPP obviously benefitted from playing each other, which made intuitive sense.  That said, given Michigan and OSU faced two and three ranked teams, respectively.   Further, their bottom three/four opponents that rank at the bottom of P5 (Rutgers, Indiana, Michigan St., and Northwestern).  Thus, it seemed odd to us that UGA and TCU would have an opponent NYPP below that of Michigan and OSU, particularly given that both played 10 games against P5 teams.

So, we looked at the underlying data.  We found that facing Vanderbilt and Colorado had an outsided impact on UGA’s and TCU’s averages.  While Colorado was clearly the worst P5 team, Vanderbilt had wins over Kentucky and Florida, which belied its -2.8 YPP.  For comparison, Rutgers and Indiana each finished the CFB season at a NYPP that was 1.2-1.5 ypp BETTER than Vanderbilt.  Of course, both finished the season by losing 8 of their last 9 games.  So UGA’s average was being penalized by a team that was probably better

Additionally, Penn State ended the year as the 7th highest team in P5 on a NYPP basis at positive 1.3 NYP, which is a statistically impressive season and included dominating Auburn, a blowout loss to Michigan, and tight game with to OSU.  That said, Penn State didn’t face any other ranked teams and and is the only team among the top 15 to not beat a single ranked team.  That suggests to us, that PSU’s underlying numbers are boosted by their schedule and aren’t representative of a team ranked in the top 15.  Nonetheless, Penn State is strong statistical boost to Michigan’s and Ohio State’s opponent NYPP.  Compare that to the third best resume in the SEC – Alabama.  Alabama has wins over Texas and Miss St and narrow losses to LSU and Tennessee. 

The Big Ten in context.  The entire conference has one win over a ranked OOC foe.  In contrast, excluding Ole Miss’ win over Troy, the SEC has wins over Oregon, Utah, Texas, and Clemson with losses to ranked teams FSU, Kansas St, and Penn St.  Not only does the B10 not have many ranked OOC wins, they don’t even have many games.  The entire Big 10 has played a total of two ranked OOC opponents (Notre Dame and Washington).  In aggregate, the B10 has played 8 P5 (including ND and BYU) OOC conference games and won five.  The SEC has played 16 (including BYU) winning 11 along with another game and win over future P5 Cincinnati.

As one might expect as the sample size grows, the weekly movement in Opponents Net YPP were not particularly significant.  As much as we love data, we wrestle with how informative the strength of schedule data based upon opponents NYPP because of the insular nature of conferences.  To be clear, we don’t want to come across as thesis-biased.  On one hand, we can’t just dismiss data, but ultimately, the challenge in comparing teams from different conferences is that metrics that may look similar really aren’t. 

Combined

The same tiers remain. Both UGA and OSU declined a bit, but the gap between the two and next tier significant.  Upon taking a 35-7 lead on LSU, UGA was at a 38% Relative YPP% for the season.

How do they do it?

Georgia remains the most balanced team, with an equally good offense and defense.  Michigan is similarly balanced, and OSU and TCU are more heavily reliant on their offenses.