Elo 2017: From past to present, predicting Ivy League football

Updated: September 22, 2017

Ivy League football Elo ratings are back for the 2017 season! As a quick refresher, Elo is a rating system that scores teams with numbers to try and calculate how strong different teams are. The higher the score the better the team and a theoretically “average” team will have a score of 1500. For more information and for those who are not familiar with the Elo system take a look at our first Elo article. The formulas used this year are the same as last year.

The graph below includes ratings from 2009 to 2016. There is a gap in our Elo ratings from 1982 to 2009 due to missing game results. To calculate Elo ratings, all that is needed are the teams playing, the location of the game and the final score. However, the games also have to be calculated in chronological order and the lack of a composite schedule for Ivy League football from 1982 to 2009 made that quite difficult. We decided to begin calculating ratings again in 2009 in the interest of bringing projections up to date. The Elo ratings will be accurate for the upcoming season because they have had 7 years to develop from their starting points.

The only minor change that was made is that instead of starting all teams at 1500, I used the previous season’s standings to rank the team’s ratings. For example, in the 2009 season, Harvard began at 1525 and Brown started at 1520 while Dartmouth began at 1475. These small corrections were made to partially adjust for the lack of historical ratings previous to 2009.



The team rankings and historical Elo ratings are interesting, but the true power of Elo rests in its ability to predict games. This season we now have access to current Elo scores and will be able to give win probabilities for all of the Ivy League football matches. The formula for win probabilities is also explained back in Elo’s first article, but it’s important to remember that the only things that matter are the difference between the two teams’ ratings and that home teams get a 65-point boost to their rating.


Cornell vs. Yale, Saturday, September 23 at 1:00pm


Let’s break right into the projections and look at Cornell’s first Ivy League game. The first conference game looks like it will be a tough one for the Big Red. Cornell, with a last place rating of 1377, will be on the road against Yale, who holds a rating of 1480. Our model only gives the team a 28% chance of victory. The model also projects Yale to win by about seven points. It is important to note, however, that Cornell was able to pull out a victory last year when both teams carried similar ratings. This is one situation where I hope the model is wrong, but it looks like it will be a tough game down in Connecticut for our Big Red football team.


Brown vs. Harvard, Saturday, September 23 at 12:00pm


The Brown-Harvard game is the only other Ivy football game this weekend and is the more lopsided of the two. Brown sits at a 22% chance of victory and is expected to lose by just under 9 points. Harvard is the better team and leads Brown in Elo rating by over 150 points (1595 to 1443) without even taking into account that Harvard will be playing at home in Cambridge. In the teams’ last meeting, which took place in Providence, Harvard won by 10 points, 32-22.



Elo ratings will be back on a weekly basis with updated ratings and game projections for the week. We will also be keeping track of Elo’s accuracy to see how well our model fares when put to the test.

Leave a Reply

Your email address will not be published. Required fields are marked *