538 Weighs In on the WCBB Tourney

[ocoandasoc]

Nate Silver’s fivethirtyeight.com site is usually amazingly accurate with their predictions on political races across the country. This year he is taking on the NCAA Women’s Tourney and giving probabilities of each team advancing to each level. The web site is

http://fivethirtyeight.com/interactives/march-madness-predictions-2015/#womens

But here are some highlights:

Chances of winning it all: (Teams with at least a .5% chance or better)

UConn 74%

SC 10%

ND 9%

Md 2%

Baylor 2%

Tenn 2%


Chances of making the Final Four: (Teams with at least a .5% chance or better)

Spokane:

Md 37%

Tenn 33%

Ore St 17%

Duke 9%

Prncton 2%


Albany

UConn 96%

Ktky 1%

Lville 1%


OK City

ND 58%

Baylor 30%

Stnfrd 5%

Iowa 4%

DePaul 1%


Greensboro

SC 67%

FlSt 17%

UNC 7%

ArizSt 4%

TxA&M 1%


Best chances for first round upset:

DePaul 77%

Nebraska 48%

Gonzaga 48%

Arkansas 43%

Miami 42%

OKSt 41%

Pittsburgh 41%

JMU 40%

 

[Wbbfan1]

If Ohio State is able to beat North Carolina, which they should, they have a great shot at knocking off South Carolina. Ohio State has the ability to outscore South Carolina. IMHO its a repeat of last years North Carolina vs South Carolina game where a dynamic young player leads her team to victory over the #1 seed. Last year it was Diamond DeShield will it be Kelsey Mitchell this year?

 

[Huskies Rule]

Thank you very much, oco- I have been following 538 for years- in politics, not sports. They are historically excellent 'predictors.' Perhaps more accurately they should be called polling aggregators- which hopefully extends to WCBB. 538 factors in Sagarin, Sokol, Moore and Massey data into its numbers. The story behind Silver's bracket percentages (found on your link):

http://fivethirtyeight.com/features/how-we-made-our-forecasts-for-the-womens-ncaa-tournament/

 

[meyers7]

What? Only 74%? Tony's not gonna be happy about this.

 

[Zorro]

Tony ain't gonna believe it. (Neither do I.) Still, a fascinating exercise in prediction.

 

[Orangutan]

The article on how they made the predictions is fascinating. Two things that caught my eye:

Last year's UConn team was the top team of the last 10 years in terms of "standard deviations from average talent." Last year's Notre Dame is 5th on that metric, which shows just how untouchable last year's Huskies were.

I hadn't seen the Moore or LRMC ratings before. DePaul is 11 in one of them and 12 in the other. And ND has to potentially play them in round two. Yikes.

Choice quote to make Huskies fans smile "it’s almost as though the top women’s teams (and particularly UConn) are playing in a whole different league." Tell me something I don't know, Nate...

 

[JoePgh]

The background article is indeed fascinating, and it introduces a whole new subject of discussion that is likely to be of interest to Boneyarders. It attempts to rate and rank teams from the 2005-2015 era based on "talent" (unfortunately it does not say how that term is defined and measured), and the results are:

1. UConn '14
2. Baylor '12
3. UConn '10
4. UConn '09
5. ND '14
6. ND '12
7. UConn '15
8. UConn '13
9. Stanford '12
10. ND '13
11. Stanford '10
12. UConn '11
13. Tennessee '08
14. Maryland '06
15. Tennessee '07
16. UConn '12
17. North Carolina '06
18. Baylor '13
19. Texas A&M '11
20. Duke '06

So this year's team is the 7th most talented of the last eleven years, and the only 2015 team in the Top 20. Surprisingly, last year's UConn team is at the top of the heap, ahead of the Tina/Maya teams of 2009 and 2010 (one of which also included Renee Montgomery).

They also give numeric ratings, which I didn't reproduce above, which indicate UConn '14 was at 4.5, whereas this years team is 3.5 -- quite a dropoff because of the departure of Stef and Bria, it seems.

UConn has had 5 of the 10 best teams in this era, and 7 of the best 20. Notre Dame had 3 of the top 10 and 4 of the top 20. Tennessee had only the two Candace Parker teams, and they were both in the second set of 10.

I would be very interested to know the details of the metric used for this comparison.

 

[DobbsRover2]

Really great and deeply analyzed stuff. I take it though that all the rankings shown above have to be the final year power rating rankings for the teams and that it is only UConn 2015 that is pre-Tourney, or the UConn 2013 would not be rated that far ahead of Baylor (and of course ND). So the UConn 2015 final score and ranking could enjoy a big jump in the next three weeks.

 

[JoePgh]

Oops, I see that 538 did explain what their numbers mean:

Since 2005, there have been 20 women’s teams with a Massey rating at least three standard deviations above average, when there only “should” have been five such teams based on a normal distribution.

So that means that UConn 2014 was 4.5 standard deviations above the mean in its Massey rating, which basically should never happen if the "talent" on the teams were normally distributed. (Probabilistically, it would be like a DNA match between two unrelated individuals.) By contrast, UConn 2015 is "only" 3.5 standard deviations above the Massey mean -- also highly improbable.

Describing this metric as a measure of "talent" is somewhat misleading, since the Massey rating reflects results achieved (yes, including margin of victory) as well as talent (aka potential). That is, unless you count coaching as part of the "talent".

 

[RegisteredUconn]

ND with 3 of the top ten teams and none of the 11 titles. That's got to be a little disheartening. Still looking for JoePgh's 4th ND team.

 

[RockyMTblue2]

Do we still have to play the games?

 

[JoePgh]

ND with 3 of the top ten teams and none of the 11 titles. That's got to be a little disheartening. Still looking for JoePgh's 4th ND team.

Note that in each of the years in which ND had a top-10 team, there was another team in the same year that is ranked higher. That may happen again in upcoming years -- ND will have very good teams, but probably not the best team in any year.

Sorry about that phantom 4th entry for the Irish.

 

[MilfordHusky]

The Kentucky men are at 41%. We're almost double that. Yikes!!

 

[Tidepool]

South Carolina (67%) has a better chance of reaching the Final Four than Notre Dame (58%)?

 

[RegisteredUconn]

Sorry about that phantom 4th entry for the Irish.

But less sorry than the Irish I bet.

 

[DobbsRover2]

South Carolina (67%) has a better chance of reaching the Final Four than Notre Dame (58%)?

Most of the difference seems to be with the expectation that USCar would play FSU, while ND would play Baylor. Baylor is considered a stronger team with a 74% chance of getting to the Elite 8 to FSU's 61% chance, so ND's chances in that game rate less than USCar's. The power ratings do generally support that evaluation if it does happen, but a bunch of games need to be played before the E8 matches roll around.