Week four of Elo predictions are now ready. As a recap of last week, the model went 1-1 with predictions. It correctly predicted the close Dartmouth victory over Yale, but got wrong the amazing Cornell upset against Harvard. A couple positive things about the model include that the Dartmouth-Yale game was the closest one of the year so far and was also our closest model prediction up to that point (64% to 36%). More importantly, though, was the Cornell upset (Go Big Red!). Even though our model was technically incorrect at picking the winner I think that an upset like this is actually a win for Elo. The projections gave Cornell a 26% chance of winning, which is a lot higher than I think most people would have expected.
The Big Red had not beaten the Crimson in 11 straight match-ups, but were able to pull the large upset last weekend. Cornell was in desperate need of a win after going winless to this point in the season. The victory last Saturday will hopefully help the team get on the right track. In terms of Elo scores, Cornell had a great week, improving 23 points to sit at 1383, sadly still last in the league. Depending on the outcome of this week’s games, Cornell could soon be out of its last place spot that has been haunting the team since last season.
Week 4 predictions are below.
Princeton at Brown, Friday, October 14 at 12:30pm
This weekend starts off with our closest match-up so far. Elo is projecting this one to be close to a toss-up. Both teams have only played one Ivy League game so far and both lost. Overall, Princeton sits at 3-1 and Brown is 2-2. Princeton seems to have a stronger team overall, but has lost to Columbia at home this season. We will have to wait and see how Brown plays at home in this match up.
University of Pennsylvania at Columbia, Saturday, October 14 at 1:30pm
The other Ivy game this week should also be a close one. An undefeated Columbia (4-0) is taking on a 2-2 UPenn team. Even though Columbia’s Elo rating is not stellar, they have not lost yet this year. If Elo factored in non-conference games I would expect Columbia’s rating to be much higher. I do not want to count the Quakers out here, but based on some of the model’s limitations I would expect this one to be very close to a toss-up.
The Elo model sits 3-3 in game projections so far. Check back in again next week for more Elo predictions.
Comments