Five of the seven models I committed to are now posted under the Tips+ section. You’ll have KenPom, Final Four/Champ, Outcome Matching, Efficiency/Coaching and the Baseline Strategy.
I’ll have at least three more…and I think I’m going to bow to pressure and do my Keeper bracket too. Special thanks to Ryan Tressler, who spent a lot of time doing the traditional Seed Match-up model, which is wasn’t going to provide–and now will.
That’s it for tonight. Gotta make some CBS deadlines. Thanks everyone for hanging in there.
Peter Tiernan has been using stats to analyze March Madness for 22 years. He writes for 
Thanks a lot Pete, I can’t wait to check them out!!!
Also, for all the readers:
all the information on the seed by seed model is contained in the comments section for the previous post by Pete (*version 1 of excel sheet is posted in tips+ section*), I run down things for each region
Could we make the region by region forum to see what everyone is thinking?!?!?
Jared – I’ll get to it tomorrow. Totally swamped at the moment.
just sent the email to you, let me know what you think
You might think that I am crazy… but I am incredibly impatient with this stuff… so could you give us a time when the region by region stuff will come out so I can have something to look forward to during this dreadfully boring Monday please
Patience, Jared. Patience. I probably won’t open these up until later this afternoon or tonight. Still crunching through models (just did Pulse Check) and need to do the Seed Quality Curve. Then I go on TV for three hours. Busy day. My busiest day.
Pete, I have paid for various sites in the past which had a “scientific reasoning” behind their method. This is by far the best site that I have ventured across the last few years. Simply amazing.
Thanks Beau. Nice to hear…particularly after the night I’ve had!
Does anybody know how Pete determines how tossups work in his “Outcome Matching” model? He says, “And for toss-up games, I followed historical win averages of the higher and lower seed to make the results look like a real bracket should,” but I don’t know which article or tool to use to figure “historical win averages”. When I look at toss-up rules, it seems that Illinois should be a victim, but that’s not what Pete’s showing… so I clearly don’t know where to look. Anybody know?
Great site Pete!
Scott – upsets determined by nearest efficiency differential in 4/13, 5/12 and 6/11 games. Efficiency numbers sort out most of the other matchups. This is a relatively new model, so I haven’t written an extensive article on it. If you feel like Illinois must go down, then don’t be constrained. I don’t qualify a 7v10 is an upset matchup.
Thanks Pete. I have another question about your Final 4/ Champ rule model… you say: “Eliminate key underachieving outliers before their seed-‐expected win totals (if both teams are eliminated in a match-up, use upset/toss-‐up rules)”. Does this mean that I look at each matchup (5/12 e.g.) and look to see which one of them has one of the outlier characteristics, and advance the other? (and if tied, go to upset rules?)
Or am I supposed to look at higher seeds only and just not advance them as far as their seed would indicate?
Thanks again-
Scott
Pete,
Excellent analysis, thanks for the site.
Question about the models: should there be some accounting for the location of the games? I believe I’ve seen studies that show home court advantage is worth between 2.5 and 5.5 points. Granted the tournament games are technically neutral site contests, however won’t Kansas State have an advantage in KC, Ohio State in Dayton, etc.? I imagine this would change the outcome matching bracket in particular.
Thx,
MattyS
I have done analysis of proximity in the past. Surprisingly, it’s not a big factor…though it does yield slight PASE overachievement when a team is within 250 miles of campus. That said, didn’t Ohio State tank in Dayton a few years back?