72% chance we lose 1st rd

[Ricardo del Rio]

This chart is not as sophisticated as TMH's chart, but I found the results as interesting. Based upon this raw data, I struggle to understand the rationale that there is a 72% chance of a WSU upset. Particularly notice the shooting percentages.

[1972Shocker]

Ok, there is a 72% chance we lose. But what about the other 80% (using the Yogi Berra mathematical model of predicitivity). Yogi should know how many World Championships did he win?

[ShockerPrez]

They would probably beat you up if you tried. And that's just the female graduate students. :p :D

No way would I take on the female students!!

The males probably have set up an impenetrable perimeter with lasers beams and ray shields. I'd have to go Flash Gordon on them.

[SpanglerFan316]

No way would I take on the female students!!

The males probably have set up an impenetrable perimeter with lasers beams and ray shields. I'd have to go Flash Gordon on them.

What would (Greg) House do? Or Sheldon Cooper, B.S., M.S., M.A., Ph.D., Sc.D.?

[Anthroshock]

The physics majors at UNLV were all juiceheads. Ever seen a raging physicist?

[SHOXAAC]

There is a 100% chance I think this system is flawed beyond recognition.

[thegalen]

Actually reading from the modeler blog he has warned that vcu result us a big outlier and may not do as well as predicted - nice waffling. Basically he saying "even though I predict them to win they might very well lose". :)

That's with respect to its value as an upset pick. That is, coming off a final four VCU will be a trendy upset pick.

Keep in mind that's what this guys whole model is about: identifying UPSETS. Not in predicting win/loss in general. The model may or may not suck, but until its results for picking UPSETS are compared side-by-side to the other models results for picking UPSETS, posting general win/loss prediction success rates is comparing apples to oranges. As for the models success rate, there's a difference between false positives and false negatives. The latter is much higher and the former (relatively) low/lower. That is, of the teams identified within this model's framework, those teams are pretty likely to pull an upset with varying distributional chances of doing so. The model doesn't though do as well with identifying the broader pool of teams likely to off the upset and therefore failed to identify several teams that did so leading to the higher aggregate error rate. Clearly then there's a covariate he hasn't accounted for. Maybe its lady luck, maybe its offensive efficiency including pacing. Who knows. Anyway I hope this helps. I'm not saying VCU is actually a favorite to win. Far from it. You guys are, all around, the far better team (our coach has voted for you guys in the regular season poll several times). Just trying to point out what this model does and does not attempt to do.

[jocoshock]

And Nate Silver says we have a 74% chance of WINNING the first game: http://www.nytimes.com/interactive/2...h-madness.html

Their is nobody better than Nate Silver...

[thegalen]

He uses a composite of Sagarin, Pomeroy et al's computer ratings combined with pre-season AP and USA Today votes. I'm not knocking Nates stat skills and this seems like a good model that will generally outperform Kenpom etc because it accounts for other factors such as location and injuries/other roster problems(see Cuse for instance). Again though this is a general/win loss probability forecasting model and not an attempt to isolate upsets specifically. You might find this, which Nate linked to in that piece, to be interesting as well: http://fivethirtyeight.blogs.nytimes.com/2011/03/11/in-n-c-a-a-tournament-overachievers-often-disappoint/

[60Shock]

And if anyone of these HaaaVarrrrrd yokels truly believe their statistics, they should put their money where their mouth is.

[SB Shock]

That's with respect to its value as an upset pick. That is, coming off a final four VCU will be a trendy upset pick.

Actually it is not. If you look at his model Turnover Margin is heavily weighted. He called VCU an "outlier" because VCU leads the NCAA in turnover margin. I suspect VCU was ran against the whole field, they would have a good chance from this model to upset most teams.

Keep in mind that's what this guys whole model is about: identifying UPSETS. Not in predicting win/loss in general.

46% is less than flipping coin, so there is not much value of his model unless you don't have a coin.

[thegalen]

Actually it is not. If you look at his model Turnover Margin is heavily weighted. He called VCU an "outlier" because VCU leads the NCAA in turnover margin. I suspect VCU was ran against the whole field, they would have a good chance from this model to upset most teams.


46% is less than flipping coin, so there is not much value of his model unless you don't have a coin.

" A note of caution, however: VCU is a big outlier. Since this model is looking for average effects, it may not do as well out of sample predicting teams far away from the average, like VCU. The Rams fit the profile of a tournament David, but between their popularity and their extreme profile, understand that the expected value of picking them might be lower than this model implies."

That's pretty black and white. As for the 46% part, again, that's aggregate error rate including false negatives. This models worth, if there is any, is not in running simulations of random team x against random team y for standard win-loss forecasting. For what it's worth, the performance of identified teams or those picked to upset was 15/21, or 71% which mirrors almost exactly kenpom et als predictive success for general win-loss forecasting based on composite performance rankings. The R2 value isn't through the roof though so that's something to consider. Anyway, not trying to flame. The suspense is just killing me and I have nothing better to do than obsess over numbers until Thursday! :)

[forevershockerfan]

There is a 100% chance the world won't come to an end on December 21st, 2012. However, the History Channel did have an intruiging show on that topic, speaking of prognosticaters through the Centuries that thought so on this day that draweth nigh. I thought Harvard was going to beat Georgetown during the 'Ewing Era' many moons ago, but they faced very unsupportive whistles until the cows came home and they finally lost.

[ABC]

That was Princeton.

This is the first time since 1946 that Harvard has been in the tourney.

[thegalen]

That was Princeton.

This is the first time since 1946 that Harvard has been in the tourney.

:) the point about whistles is well taken though. Our win margin is far lower than yours and that separates good teams from great teams. Were far more susceptible to collapsing in the face of adverse calls. It directly disrupts our defensive flow which is of course what our team is built around. Speaking of which, was trying to rewatch the UNLV game but no dice on espn3. Am watching the Ill State game instead. Some brutal fouls!