FBS Rivalry Games Decline Parallel to an Increase in Bowls and FCS Opponents

Figure 1. Margin of Victory, 1990-2016

I discussed college football program rivalries in a previous post. Preparing a forthcoming post elucidated a decline in rivalry games occurring parallel to changes in the college football environment. Some operational definitions appear at the close of the post. All data were culled from Sports Reference. 

There was the implementation of the BCS in 1998 to resolve that annual, irrevocable controversy about the national champion. The BCS spawned myriad controversies and intensified that which it was implemented to resolve. There were also concerns that teams were maximizing margins of victory to exploit innerworkings of the BCS algorithm until it was remodeled. However, any broad impact in FBS vs FBS games is not readily deducible from Figure 1. The gradual increase in margin of victory over FCS teams however, is likely attributable to an overall increase of FCS opponents and, thus also, an increase in higher-quality FBS teams playing FCS teams, as seen in Figure 2 (in purple).

We also see in Figure 2 that, although it may be spurious epochal fluctuations, there was a few years from ~2005-10 when the proportion of FBS vs FBS between two top-25 ranked opponents dipped considerably (orange). Indeed, when schedules are set years in advance, come game day, the ranking of either team is unknowable. But, we might also note that teams were playing gradually fewer games versus top-25 ranked opponents (brown). However, there were more FBS teams as time progressed and a lower proportion of those teams could be ranked at a given time as there was still only 25 ranked teams in a given season.

Figure 2. Attributes of FBS Games, 1990-2016

In 2006 and after, FBS programs were permitted to schedule one additional regular season game in any season. Prior to this ruling, programs could schedule one additional game in select seasons. It was around 2006 when the quantity of bowl games began to increase as well. Enter the Playoff in 2014. The concept of a playoff was, itself, accompanied by some controversy. Critics of the CFP cite similar issues as the BCS. Alas, this author is too indolent to cite anything in this paragraph. 

Figure 3 indicates that programs and their teams were pressured to respond to these and other environmental changes. Accompanying the 2006 rule-change permitting one additional game was another statute allowing one win over an FCS team to count toward bowl eligibility in each season. Previously, since at least 1997, one FCS win “every four years” could count toward bowl eligibility.  Thus, programs scheduling an FCS foe essentially guaranteed their teams 12.5% of the six wins necessary to secure an official invitation from one of an ever-increasing quantity of certified licensed bowl games.  

Figure 3. FBS Bowl Team Attributes, 1990-2016

The green line in Fig 3 corresponds to the proportion of bowl teams with 6 or fewer scheduled wins. Annual vicissitudes are evident but only in 2013 or -14 does a trend beginning in about 1998 appear to ossify. The orange and brown lines add some context; until about the inauguration of the BCS the two generally overlap. Although the proportion of bowl teams scheduling multiple FCS games appears relatively constant 1990-2016, we see that even before the 2006 ruling, increasing proportions of bowl teams were distending their win totals with wins over FCS teams, though this may or may not have increased the likelihood they received bowl invites. Lest there is no mention of monetary implications, what was the cost to rivalries?

Summarily, as seen in Figure 4, the programs are scheduling more games versus FCS opponents, but the increase has been gradual, while simultaneously, more teams played in bowls…because there were more bowls. However, we see that, on average, teams are playing fewer games versus rivals, ~15% to ~10% from 1990 to 2016. Realistically, this equates to from ~1.8 to 1.3 rival games per team over that span (out of ~12 to ~13 scheduled games).

Some Operational Definitions 
To examine how the foregoing adaptations are associated with college football rivalries, at the FBS level, I examined the variables described below. The rivalry data are described elsewhere and expounded upon below.  I define scheduled games as games that are not bowl games but include conference championship games as these, when not de facto, are scheduled per se. 

Figure 4.

Minimalists

• Set the minimum wins of team playing in bowls at 6 scheduled games for each season (although that minimum was 5 in several seasons)
• Computed for each season the proportion of teams in bowls with at or less than 6 scheduled game wins
• Subtracted the quantity of wins versus FCS opponents from each team scheduled game win total in each season and computed the proportion of teams playing bowls with < 6 scheduled game wins versus FBS opponents
• Computed the proportion of bowl teams that played ≥1 FCS opponent
• Computed the proportion of bowl teams with ≥2 wins over FCS opponents
• Indeed, this value is meaningless in regards to bowl eligibility. 

Team Games vs Rivals

• Compute proportion of scheduled games versus rival(s) for each FBS team in each season and then compute average for FBS teams for each season.
• I would have included an adjusted version of this value adjusted by subtracting FCS games from the denominator, i.e., rivalry games / (scheduled games – FCS games), but it is essentially overlaps with the main variable
• I included bowl games in these totals because imagine the sensation of a Michigan-Ohio St. playoff game.Non-FBS teams were excluded from the computation of this variable because of the following bullet.
• I used a data set for rivalries generated for a previous posting. In that post, I set the rivalry inclusion criterion at ≥50 rivalry games as major college football programs. This data pertained to games from 1891 to 2015. To apply a similar stringency to the present data, a threshold of rivalry games was set for each season proportional to 50 / (2015-1891), which equals ~.403. For instance, the 1999 rivalry inclusion criterion was set at 43.45 games = (1999 – 1891) × .403 = 108 × .403. So any rivalry with less than 43.45 games by 1999 was excluded in the computation of that season.

FBS Games with Two Ranked Teams

• The quantity of scheduled games pitting two FBS-ranked teams divided by the quantity of scheduled games with at least one FBS-ranked team.
• Indeed, this could have been computed as games with two ranked teams divided by total games. But, there were 107 FBS teams in 1990 and 128 in 2016. However, there were 25 ranked teams in each week of those seasons or 23.3% and 19.5% of total teams, respectively. Likewise, teams played, on average, about 12 games from 1990-2005 but 13.5 games from 2006-2016. Thus, as the eras progressed, there were more games to be scheduled and thus a higher likelihood of facing an unranked team. It seemed to me, most objective to derive the value in this manner as it standardizes across seasons, to some extent.

Games with One Ranked- and One non-FBS Team

• The quantity of scheduled games with one ranked FBS team pitted against a non-FBS team divided by the quantity of scheduled games with at least one FBS-ranked team.
• Indeed, this could have been computed with total games in the denominator but the results are similar.