A quick note about the influence of preseason ratings
12.07.11
There have been a few questions about the influence of the preseason ratings and how this works itself out of the system early in the season. This is handled the same as last season, the details of which are here. You can just subtract one day from the key dates mentioned in that post. One important update is that subscribers can access raw values for both efficiency and tempo. One can go to a team page and click on either of the efficiency values. This will take you to a page with the raw values for every team in Division I.
The raw values are useful from a factual standpoint, but even with the preseason influence, the adjusted values are still your best bet for describing a team’s true ability relative to the rest of the country. Actually, that should read “especially with the preseason influence.” The benefit of allowing the preseason ratings to have some influence at this point in the season is to combat the problems of small sample size.
For instance, right now the Sagarin predictor has Wisconsin at the top and Marquette as the fourth-best team in the nation and I expect my ratings would be the same were it not still including the preseason ratings. Wisconsin and Marquette are very good teams, but I think we can agree they are most likely not the best and fourth-best teams in the country at this point, and the weighting placed on the preseason numbers prevents the ratings here from being quite so bullish on either team.
Power to the people: Harrison Barnes’ minutes tonight
11.30.11
[Update: 144 people responded. The median response was 27 minutes. Thanks to everyone that participated.]
As mentioned in a post a couple of weeks ago, I’d like to do some crowdsourcing from the people that have been generous enough to purchase a subscription to this site. Figuring that those folks are better informed than the average college hoops fan, we should take advantage of this knowledge. This evening I’d like to do an initial test of this concept by having the audience predict the number of minutes Harrison Barnes will play for North Carolina during regulation time in tonight’s game against Wisconsin. You can access the question from the banner on the home page. Obviously, you have to be logged in to participate.
The extremely basic facts pertinent to this question (I want to avoid influencing the jury with my opinions):
- Barnes rolled his ankle against UNLV last Saturday.
- He would return to the game, but leave the arena on crutches afterward.
- He will play tonight.
- Here’s his player card at ESPN for reference on his minutes played the past two seasons.
Feel free to do as much or little additional research as you want. Obviously, Barnes’ minutes will be affected by factors other than his health. His health might even be a non-factor. That is for you to judge. I wouldn’t recommend thinking about this for more than a minute or two, though.
This question will expire at 9 pm eastern and I’ll do my best to reveal the group’s opinion before tip-off. This is really a test for bugs in the code and that logistically this concept can work. I’ll record answers for each user that responds. You can submit as many answers as you want, but only the last response is recorded.
Your guide to the 2011 Big Ten/ACC Challenge
11.28.11
The 13th annual Big Ten/ACC Challenge gets underway this evening. Based on Monday’s ratings, here are the chances of the favored team winning each game…
Tuesday
Michigan at Virginia (62%)
Northwestern at Georgia Tech (52%)
Illinois (55%) at Maryland
Miami at Purdue (79%)
Clemson (55%) at Iowa
Duke at Ohio State (78%)
Wednesday
Penn State (62%) at Boston College
Indiana (53%) at N.C. State
Florida State at Michigan State (68%)
Virginia Tech at Minnesota (62%)
Wake Forest at Nebraska (86%)
Wisconsin (54%) at North Carolina
Obviously there are a few grains of salt to be taken here. Even if Harrison Barnes is less than 100%, UNC should be expected to beat Wisconsin. And Minnesota no longer has the services of Trevor Mbakwe, thus their chances against Virginia Tech may be closer to a toss-up.
Using the game probabilities listed, the following chart describes the likelihood of every possible scenario in the event. Keep in mind, for the first time in the history of the event, there are an even number of games scheduled*. This brings in the possibility of a tie.
This is not only a semi-useful analysis, but an instructive exercise in math! Even with the Big Ten favored in nine games based on the ratings, the most likely outcome is a 7-5 Big Ten victory due to many of the match-ups being close to coin flips.
Add up the numbers, and you get the following chances for each conference winning:
Big Ten: 68.9%
ACC: 13.5%
Tie: 17.5%
Even if we give UNC a 70% chance of winning and Virginia Tech a 50% chance of victory, The Big Ten’s win probability only drops to 61%. For the record, the Big Ten has a 0.18% chance of a skunk, while just two out of one million simulations resulted in a 12-0 ACC win.
The early power ratings have emphatically supported the Big Ten being the strongest conference in the country. The next two days will serve as a great indication whether this is realistic or not.
*This won’t be the first time an even number of games have been played, however. In 2001, an even number of games were played after the Michigan State/Virginia game was cancelled at halftime when condensation from the hockey rink at the Richmond Coliseum was deemed to have made playing conditions unsafe.