Game by Game Ratings

Hello folks, sorry for the long hiatus, but I'm back and soon (well, after this weekend) I'll be all college basketball all of the time. Anyways, First I'd like to go ahead and endorse, err pick the Patriots in the Super Bowl (sorry about that, too much debate watching). They'll get all 10 of my precious points. Now, onto the real reason I'm posting. I've been toiling away at a game by game measurement of the Tar Heels' offensive and defensive performances. I wanted to find a way to represent how well the offense and defense played in each game to see if the team is in as much trouble as people suspect, and also to see how much of the team's struggles have resulted in playing some stiffer competition. The results? Well, check out the nifty graph:

Now, what does that all mean. Ok, lets start with the easiest to understand first, OE-DE. Basically, game by game, our Offensive Efficiency minus our Defensive Efficiency. This roughly translates to score (as you can see, the lone negative value is the Maryland loss) and is a good barometer of how the game went, from domination (SC St, Kent St, Valpo) to close games (Clemson, GT, Maryland). Now, the slightly more complicated part. The red and green (for stop and go), are what I've deemed ratings. I calculated a rating based on two things: Opponent's Adjusted Efficiency (season average) and Actual Game Efficiency (game results). The rating is the difference between that single game, and the team's average on the season, which is then weighted by a adaptation of z-score. So, since Clemson is averaging 115.3 OE on the season, in the game against UNC they had 99.4, this makes that game's rating (after some z-score adjustment) 14.34. The blue line can be used as a guide to see the results of the game, and the red and green lines the respective strengths of UNC's offense and defense. Obviously UNC has won more games relying on offense than defense. As a side note, UNC's forced TO% and OE are now significantly correlated at a 99% CI. Basically, UNC scores more efficiently when forcing more turnovers's, statistically speaking.