It occurs to me that the idea of a intra-California results can be generalized to “virtual conference”. Selecting a pool of teams from which to draw intramural results out of the larger results pool might be a more useful general problem. I swear the general idea only presented itself after working on the (absurd) specific idea, which is some kind of software dev lesson right there.
So I will work on that and polish the API a little. The obvious idea is to make the engine agnostic about what pool of teams it gets, and I think that’s totally feasible now.
The only thing about the pool of teams is the FBS/FCS split, or more generally, the fact there are larger abstract tiers of team quality that may intersect with your chosen pool. In our case, there are FCS California schools (well, Davis at least) that play FBS schools. Over the years the NCAA has has had the D-1A / D-1AA split and before that the “University Division” / “College Division” split.
When I first conceived of this I thought the FCS data would be harmless, based on my recency bias of the last 10 years where there’s only the odd game with FCS opponents. Fortunately the real data pool is much more interesting. The first run of results shows varied data with much more diversity of winners than I anticipated, clear “eras” and team dynasties and all kinds of avenues to explore.
But there are eras when many more FCS teams were playing the FBS teams, and more intriguing from an algorithm standpoint, the same team might be FCS one year and FBS the next. San Diego State in the 60s promoted at some point. UCSB was up and down before disappearing entirely.
It appears College Football Data has an API to query FBS status on a year-by-year basis, which is great. (I’m using “FBS” as a catch-all term which is what they use, I’m assuming that for the 60s where the NCAA actually used the term “University Division” that’s what the API tests for.)
The most conservative approach is simply to disregard FCS results. The MCC can only be won by a team playing at the highest level and the results counted are those against opponents from that level.
Next step is to generalize the API with a flag for “inspect FBS” and then compare the results when running it both ways. I don’t expect to change much… the Cal State system in the 80s was still largely playing at the FBS level… Cal State Fullerton and Cal State Long Beach were fielding D-1A teams in the 1980s, which is a mildly shocking thing to hear in the context of the last decade.