Penn State basketball loses to Rider

And I think that Beilein has chosen the certainty of tomato can wins vs the small possibility of a loss that, while not affecting RPI much (compared to a win vs a terrible team), would have an outsized negative impact on a tournament resume. My point was that bad losses, or lack thereof, have a huge impact, and I think Northwestern's case last year is a great example of that. So Beilein has tried to set up schedules that eliminate the possibility of a bad loss, and he has largely been rewarded compared to where RPI suggests Michigan should be seeded. This is why he has not been scheduling a lot of RPI 100-200 teams (like Oakland, or much of the MAC), and I think he's right not to risk losses in games that his team might be prone to take less seriously. I also think the B1G's unusual level of suck this year is magnifying the impact on RPI as well. I do agree that there is a significant RPI difference between wins over RPI 200ish teams vs RPI 350ish teams without a corresponding difference in win expectancy, but it is also tough to predict the quality of teams year to year, and it can also be difficult (or expensive) to schedule a lot of those games without giving up home dates.

College hockey years ago tweaked the RPI formula to drop from consideration games that teams win that would hurt the team's RPI anyway. While RPI is convenient in that it is somewhat easy to calculate, RPI would become a better tool if that change were made. It would still not be the be-all end-all tool that it is largely used as in hockey, but it would remove the issue that Michigan is facing with these three games against bad competition. Hockey has also added bonuses for road results and deductions for home results, which might be nice to have quantified in basketball rather than leaving it to a nebulous committee of ADs with varied agendas.