Updated Big 12 Projections

A lot can happen and quite frankly a lot needs to happen. KU needs to step up the play as we get into conference. We have in the past and I believe we will again. But right now the DPPI says it is a long uphill battle. One tweak I am contemplating for the DPPI is to perhaps diminish even more blowouts against poor teams. West Virginia has feasted on a very poor schedule thus far. So I am not sure how real their No. 1 DPPI ranking is. But Sagarin has them No. 1 as well. The one thing for sure is KU needs to step up the pace.

Whether we win or not I just do not see OU and UT coming in tied for 9th.

Wow! What an interesting look from top to bottom. I will be very surprised if things are not shuffled around from this and would need to see the teams get a few conference games under their belts before committing to that order.

I agree with Illhawk regarding UT and OU. I also have a tough time seeing WV winning by such a margin. In addition, we have all seen how the early non-con play has seemingly over-inflated the overall view of the Big-12 in the past with our teams laying eggs in the NCAA tourney.

It will be a dog fight, to be sure. If this team does not want to be "the one" who breaks the streak, they will need to be ready to step up each and every conference game.

As a "stat head," I truly appreciate and enjoy your posts here. But, with all due respect, I find these projections and probabilities difficult to have much confidence in. I just cannot accept that the probability of KU winning the conference title is less than 2% -- 47 to 1 odds? Really? If I was a betting man, I'd take all that action I could get! Finally, the boys have won 11 in a row, It's difficult to see how they can "step up the pace." Perhaps "keep up the pace" would be more accurate?

Merry Christmas and Happy New Year, Corpus!

I was wondering the same thing . . . . .


RC

Don, I feel your pain. I have been cussing and discussing my algorithms with my son all weekend. I have decided to tweak my algorithm a bit. As a result the probabilities have changed a wee bit. The issue at hand is that West Virginia is just blowing everyone away this year and their rating is obscenely high as a result. Without giving exceedingly too much detail here is a recap of my algorithm.

1. Sum op the scoring margin o a teams opponents.
2. Add to that the sum of the scoring margin for the opponents opponents.
3. Add to that the sum of the scoring margin for the opponents opponents opponents.
4. Add to that the sum of the scoring margin for the opponents opponents opponents opponents.
5. Add this to 71 (the average score per game for all games)
6. Adjust all the games above for a home/away/neutral adjustment
7. Adjust for trending (a team improving or getting worse) (This is usually a small adjustment of less than 5%)
8. Adjust weighting to increase weight of results against tougher teams (mental toughness)
9. Adjust for statistical outliers and consistency (i.e adjust for very large margins of victory or defeat assuming they are not necessarily indicative of future predictability)
10. Finally adjust for Actual performance margins compared to predicted margins.

There are a few other tweaks and twitters but this is the basic algorithm. So the fact that West Virginia is rated so high is adjusted for blowouts and trends and weak schedule. And when the dust settles, West Virginia is simply playing very well thus far.

Here is an updated Big 12 projection based on my updated algorithm. KU currently is up to about a 14% probability of winning. Do I believe that? Sort of. Last year KU got as low as a 15% before coming back to win. The graph below shows the win probability after every conference game.




As for the probability of winning equation, the graph below shows the source. I plotted the probability of winning all 5000+ games last season against projected margin. It is a normal distribution. I simply fit the curve and use that normal distribution equation to calculate the probability of winning a game based on the projected margin. So for instance, KU is a 5.6 point favorite against TCU which equates to a 70.6% probability of winning.

WVU SOS 168 www.cbssports.com/collegebasketball/bracketology/sos

KU 24th

RC

Thanks for the info, Corpus, and for the insights into your methodology. I want to ask you a couple of questions. First, why add 71, a constant, to the index? Since it is a constant, I don't see the necessity or value in using it. Second, I'm having difficulty seeing the correspondence between the table and the graphs in your post. I'm assuming that the x-axis is conference games played. We are now at game 0, right? If so, the graph shows KU with about a 25% chance of winning outright and WVA about a 38% chance. Shouldn't those values be the same as the ones in the table (6% and 84%, respectively)?

Sorry to be a bother and thanks again for your analysis.

The graph was from last season. The 71 is to make the rating relative to an actual score. All of the calculations result in a score "relative" to the mean or average team. So KU right now would have a rating of 22 or so and UC-Santa Barbara has a rating of -13 or so. I could keep the ratings like that but it seems more sensible to add in the mean score from all teams and rate around the mean score rather than around 0.

Got it. Thanks again, Corpus! Happy New Year!

Interesting stuff, but this is one of those instances where the data doesn't seem to pass the "smell test". By "smell test" I'm referring to gambling.

I don't know what the vegas odds are of KU winning a 13th consecutive crown, but I'm guessing that odds aren't reflective of WVU's cakewalk that your data suggests. In fact, I'd assume that if you were to place a bet on it, the odds on KU would make for the worst payout.

In fact, it'll be interesting to see them play at Stillwater on Friday. If they blow out OSU in their own gym, that would be significant. Sagarin has them winning by around 6, but also has them with a SOS of 338 (out of 351). So it's not unreasonable to expect WVU to lose on Friday.

Anyway, lots of games to play. Nothing is guaranteed.

WVU?

Baylor? Yuck. I don't like crybaby drew at ALL!!!!!!

1/24, 2/1, 2/13, and 2/18. That when we play @WVU, vs. BU, vs. WVU, and @BU. While there will still be 4 games left after that group, those 4 games will be the critical stretch for us this year based on these projections.

Of those, the bookends, starting with a road game in Morgantown and ending with the game at Baylor are the real biggies. Assuming that we hold home court, those road games are huge.

As we all know, winning your home games and roughly half of your road games is usually enough to get the job done.

Finally, look at the full 8-game stretch:
1/24 @WVU
1/28 @Kentucky
2/1 vs. Baylor
2/4 vs. ISU
2/6 @KSU
2/11 @Tech
2/13 vs. WVU
2/18 @BU

The ISU, @KSU, @Tech "reprieve" in our schedule is far from an easy cruise, with two road games and knowing that the KSU game is always tougher in person than it should be on paper.