Notre Dame just beat Duke, 77-73, putting them at 20-2 for the year and a fantastic 8-1 in the ACC. Does that sound like a team that should be ranked #27? According to the RPI, that's where they stand (computed without the rest of today's games).
That's up from #36 before the win—when they were just 19-2 and 7-1 in conference.
It's not like the Irish didn't beat anyone before they topped the Blue Devils. Duke is their 5th top 50 RPI victim, and 2nd top 25 RPI win:
Notre Dame Wins:
42 Michigan State
11 @North Carolina
37 Miami FL
47 @NC State
Notre Dame Losses:
So how come no love from the Ratings Percentage Index? Well, the problems with the RPI have been chronicled for decades, and it boils down to: the RPI doesn't know who you beat and who you lost to.
That's why there's so much attention paid to "Top 50 wins" and "Top 25 wins" etc. The raw RPI doesn't include any of that in the calculation. Sure, you get more points for beating Duke, but most of that (2/3) is just for playing Duke, since 2/3 of the RPI is schedule strength.
Schedule strength is why Notre Dame's RPI is so poor, specifically, their weak non-conference slate. Side note: no one should complain about a team's SOS if their RPI is adequate, since SOS is such a huge part of the RPI. Home/road is factored in, too, but again, not against specific teams, so it's valid to complain that a team has no good road wins, etc. even if their RPI is good.
Anyway, Notre Dame's quality certainly wasn't captured by the RPI in the pre-conference season, and that's the fault of the Irish who scheduled a bunch of cupcakes. But at this point in the year, with the lineup of teams they've defeated, any good ranking system should understand that they're a great team.
Now, it's true that the RPI doesn't use victory margin—by design—so it's a bit unfair to compare the RPI to Ken Pomeroy's ratings, or Jeff Sagarin's Predictor, or our Strength ratings (the Irish were #12 in all three yesterday), but any decent system that doesn't use victory margin should have had the Irish in the top 25. Even before the Duke game our Success ratings put them at #14, and post-Duke they will be around #8.
Since the RPI is so important for seeding in the NCAA tournament, the Bracket Matrix only had the Irish the low 4-seed at 19-2, while our Dance Chance put them at the high 3-seed. Most Bracketologists take into account a team's quality wins since they know the RPI doesn't do it, and therefore the Committee has to do it, too. And the Irish also pass the "eye test" and rank high in MOV power ratings as mentioned above. And their 2 losses? Both are to RPI top 25 teams. Ironically the RPI doesn't see that, either—just that they have two losses. They might as well have been to Fairleigh Dickinson and Binghamton, and not that much would change according to the RPI. In other words, we've all learned to compensate for the RPI's shortcomings, but it still can drag a good team down.
As seen above also, most of Notre Dame's good wins have come at home; they'll undoubtedly take some losses in the 2nd half of the ACC season, and that will push their RPI lower, but at the same time their SOS will get better, too. Maybe they'll end up about where they deserve to be in the end, but the RPI's poor formula has hurt teams before who deserved better—and helped some be seeded much higher than they merited.
When a team schedules cupcakes, garners a dozen easy wins, and later on loses to the tougher teams, then they deserve their resultant low RPI.
But if the same team wins the big games, their easy non-conference schedule is pretty much moot, and in that case the RPI can't see the forest for the trees.