Frankly, though anything from 3-7 looks pretty murky and really a reasonable miss from this top-level model could have a team like Brown ending up 3rd best.
The good news for Yale is that a reasonable comp (Cornell 2010) lost about the same number of minutes/key players from the same starting point and stayed in the Top 200. And that's a team that was better offensively than defensively. Offense is more dependent on continuity than defense, so you'd expect a more defensively-oriented squad like Yale to fall a bit less with the same amount of turnover. In other words, I think Yale will now fall to the pack, but will probably have the highest floor of any of the teams down there.
I'm struck by the calculation that 6 of the Ivy's 8 teams project at 170 to 243. Looks like there will be a very competitive 6 team race. If Harvard doesn't gel in time to avoid falling behind Princeton, then the race for the two remaining playoff spots will be the focus.
From an efficiency margin perspective (since that's how KP judges things now), here's the expected efficiency margin from above and then the actual:
Team Expected / Actual
Brown (7.55) / (9.86)
Columbia (4.03) / (6.27)
Cornell (4.83) / (8.72)
Dartmouth (4.91) / (12.96)
Harvard 8.03 / 4.75
Penn (4.96) / (0.21)
Princeton 13.01 / 12.59
Yale (1.22) / 1.00
All in all, it's not a bad first blush estimator. The biggest misses were Dartmouth (overpredicted) and Penn (underpredicted) but every other team finished within 4 pts/100 poss of the expected Efficiency Margin.
In the future, I'd probably add a frosh mins expectation based on how strong frosh contributions can overcome the need for more continuity (I plugged Harvard's continuity to account for the frosh - and slightly over projected - because without it, Harvard would have been near 200, which didn't seem right).