Hey y'all! I hadn't been planning on making another FanPost so soon, and was all set to go into hibernation mode until college football starts up again (THANK GOD that Breaking Bad returns, or I don't know what I would do with myself).
However, I read the Football Outsiders Almanac that just came out (which you should as well, it's awesome!), and it got me thinking about my win/loss projections I made for each Big Ten team. I projected a score and went from there, but that's not really the best way to do this. If a team is favored by 1 pt in every game, you would "expect" them to win every game, but their record at the end of the year would probably be more like 6-6, reflecting the uncertainty from game to game.
I kind of referenced this when making "Best Case/Worst Case" guesses, and one can sort through the Projected Score Differentials and say "yeah, that game's probably a toss-up, that game's probably an easy win," but that wasn't good enough for me. I think it's better to list each game as a "probability of victory," which gives leeway to the actual results come fall and gives a better tell of what record we can expect from teams at the end of the year. Like I mentioned in my Wisconsin preview, I don't expect the Badgers to go 12-0, and going by win probability gives a better projection to the naked eye.
NOTE: Nebraska, y'all still get pessimistic projections. The schedule is BRUTAL.
Now when converting "Projected Score Differentials" to "Probability of Victory," I didn't overthink myself. I just did a simple linear regression, with 50% at 0 pt "Projected Score Differential," and 99% at 30 pt "Projected Score Differential."
I used 30 pts as a cap, just because that's about when the probabilities went to 100%. 99% cap, just because I think that there should always be some possibility of upset.
I used 2011 data to formulate the... formula, so if last year was really weird, apologies. (NOTE: for whatever reason, 4-7 pts "Expected Score Differential" was consistently less certain than 0-3 pts "Expected Score Differential". Perplexing, but whatevs).
Also, Best Case/Worst Case was made using a 20% threshold. I just kind of grabbed that out of the air, but it seemed okay. If I'm completely off-base with that, let me know,.
Now, I'm not going to go through and post each Big Ten Teams schedule, because that's a lot of work, and I already wrote those other previews. I can however, give an updated Total Win Projection, complete with the MATLAB plots I just made to give an idea of the distribution. They are as follows:
BIG TEN EAST
BIG TEN WEST