An experiment in ranking

First off, let me start out by saying that I have been reading this blog for a while now and have always enjoyed reading the views that are expressed here so in a way I consider this my first post. I haven't been posting much myself mainly because I haven't found much to put forth in discussions, but I felt that this was something everyone would want to see. A while back  I started an experiment with ranking each team in the NFL, with the idea of taking the schedule of each team and using Excel, creating a rating for each game, and totaling them into a score for each team. My big goal was to create something I felt was fair and would rate each team's performance with as little bias as possible, then I could see how good or bad my boys were in the league.

The system I chose to use to rate each game for each team by using two "scores" for both wins and losses.

First is the opponents wins/losses and second by the strength of victory/loss as compared to the average for that team. So for example, in the Cowboys win over the Raiders in week 12, I would take the wins of the current win/loss for the Raiders 5 and multiply by 15, then to rate how the Cowboys performed compared the the other teams that beat them, I took the average amount that the Raiders loose by which is 19.33, and the amount that the cowboys beat them by which is 17 points, got the percent of 17 into 19.333 which is approximately 87.9% and then added 87.9% of the 5x15 the wins of the Raiders.

Now for losses I would take similar numbers but instead of adding to the total I would take the current losses that team has and put in as a negative and then add in the percent of loss (if the team looses by more that the average loss it would put in a negative number and a positive for less than the average). now to avoid div/0 errors in excel, thank you Colts, I would add 1 in losses( that's why I multiply by 15 to prevent this from effecting the results.

I been running this for a couple of weeks to see how well the results work from week to week. Just a few facts I think everyone would like to know the saints were #2 until the Pats game and the cowboys bounced from the 12th and 8th spot until of course the saints game.

I view these rankings as more of a look at how the teams have performed for this season and not an indicator of how each team is this good or bad.  but to satisfy my curiosity I took a look at how it predicts victories in the next week and since week 10ish(when I started). the rankings has performed at about a 89% accuracy with last week being the worst. but to see how good this works I will post how this is predicting next week. I have a feeling that this will not work well with so many teams vying for playoff spots but lets see anyway.

Hopefully I managed to explain how the rankings work well enough. I do look forward for feedback to see if my line of thinking is valid or not in this setup.