I take your presentation of your back of the envelope calculations here to be what you see as the most likely scenario. I'm not disregarding the possibility that a full CFB season would be no worse than what you describe; I'm saying what you lay out is much closer to a best case scenario and is minimizing the true risk. If you were to say, "based at what it looks like, we wouldn't expect 100s of players to die of COVID-19," I'd probably agree with the interpretation.
The MLB anecdotes are to illustrate a potential worst case scenario that would be worse than what you described. If there are zero more MLB COVID cases for the rest of the season you arrive at the 3% that you just came up with; obviously, seeing zero cases in the 8 weeks that follow the 27 that occurred in the first two weeks is not a worst case scenario. If you want to engage in a good faith discussion, it is bad practice to obviously distort the argument presented to you; I'll try to be more explicit going forward to minimize that possibility.
What the Cardinals and Marlins example illustrates, is that a team of healthy young men that travel, practice, and play together can see up to 60% of them contract COVID in as little as a week. As the season goes on, maybe we see only another 4 or 5 teams experience similar outbreaks with a few sporadic cases on other teams that don't spread and 10-15% of MLB players get infected this season so we're in the ballpark of your estimate. Maybe we see almost every team catch an infection that blooms into an outbreak and 30-45% of MLB players get infected. My point is not that your estimate is an impossible outcome, it's that there is no reason to be certain that 15% represents a reasonable ceiling for an outcome worth considering.
The biggest question I have is, where does your case fatality rate of 0.0001-0.0002% come from? Here's a reported CFR of 0.5% (would translate to 0.05% based on how you decided to 10x the number of cases for your cumulative incidence estimate) for the 20-29 year old cohort or 0.1% for the 10-19 year old cohort.
SOURCE
View attachment 38653
Even if we take the lowest CFR we can derive from this data and use it in your scenario (10x as many cases), we'd be going with 0.001%, which is 100x higher than what you're using.
I can pull numbers from reasonable sources and make some unqualified assumptions and say, as many as 30% of players could end up infected because college players will be more irresponsible than MLB players, and play a much higher contact sport in larger teams where they will be more cramped on the sidelines and using facilities in the locker room in a higher density; and that we could see an IFR (infected fatality rate) of maybe 0.025% (half of the rate on the table above) which means that 0.0075% of college football players would be expected to die if we played a season, that works out to about 1 player. However, that's just a silly first run through of the numbers that could be way high or way low.
I gather you will most likely argue that my fear mongering estimates arrive at a measly number of deaths. We'd probably both agree that if there was a CFB season we shouldn't absolutely expect dozens of kids to die. I just think a more clear eyed view of available information places the best estimate of deaths much higher than what your calculation did. There's a non-trivial chance that we'd see at least 1 COVID-19 death of a CFB player if the P5 teams played a season; I'd also say there's a non-trivial chance that it could be worse in many ways that we wouldn't see from cobbling together some back of the envelope calculations in a non-systematic approach that doesn't construct a proper statistical model.