538.com's Sweet Sixteen predictions/projections

[Knicks411]

UConn women are 76%, why shouldn't they be?

When you factor in Brazil hadn't lost a competitive match in Brazil since 1972 before the 2014 World Cup yes I could understand that.
The betting lines had Brazil as a heavy overprotect before that tournament.

These are the same odds the USA basketball had at the 2012 Olympics. They were an 85% favorite to win Gold.

Brazil soccer was a heavy favorite at home in 2014 when no European team previously had won in the North/South America.

The mens olympic basketball team is a lot further ahead of the rest of the world than Brazil in soccer. (So yes, I probably did overstep my original statement, I should have clarified further). Same point for the Uconn women.

The last world cup in either North or South America prior to 2014 was in 94, not sure what happened 20 years ago is that relevant. Brazil beat Italy in penalty kicks. In 86 Argentina beat Germany in Mexico 3-2 in the final. In 1978 Argentina beat the Netherlands in the final in extra time. So sure, a European team didn't win those, but it's not like we're talking huge margins here. Plus, we're talking 3 tournaments over the course of 36 years.

I have no problem with Brazil being the favorite. Bu tI think there's no way they should have had (essentially) the same odds as the US men's olympic basketball team or the UConn women.

 

[mscanlo]

You realize analytics don't decide games they are just a tool. I trust 538 to give an idea of the teams numbers. Upsets always happen.

Sports are decided on the field even if the numbers say differently. Nobody would have predicted Middle Tennessee beating Michigan State. All computers had Michigan State a 96% favorite but MTSU won that day.

Silver does a good job giving info. Nothing more just info.

thus my last sentence, "sport is very difficult to predict"

i imagine nate silver has cool and interesting models, i'm just saying, when he and his team apply it to sports, it's not very great. then again, probably not too many models are because it is sports and anything can happen.

what i mean to say, his models essentially just pick the favorite to win. it's not that challenging/interesting if you think about it.

the likelihood of winning is the interesting stat, but there is nothing to compare if it is correct. they only play the game once, not one-million times like 538 probably models out (i.e. mich st was 96% favorite, maybe they were really only 80% favorite in actuality, or 99% favorite, nonetheless, the amount they weren't favored ended up being correct.


http://fivethirtyeight.com/interactives/world-cup/

 

[OttoinGrotto]

Thank you, sir. This will appear in a slide deck for an upcoming meeting.

 

[HoustonCuse]

UConn women are 76%, why shouldn't they be?

But only about 4 schools even try in women's basketball. It just isn't a competitive enterprise at this point. Maybe in 15-20 years, but not now.

 

[Knicks411]

thus my last sentence, "sport is very difficult to predict"

i imagine nate silver has cool and interesting models, i'm just saying, when he and his team apply it to sports, it's not very great. then again, probably not too many models are because it is sports and anything can happen.

what i mean to say, his models essentially just pick the favorite to win. it's not that challenging/interesting if you think about it.

the likelihood of winning is the interesting stat, but there is nothing to compare if it is correct. they only play the game once, not one-million times like 538 probably models out (i.e. mich st was 96% favorite, maybe they were really only 80% favorite in actuality, or 99% favorite, nonetheless, the amount they weren't favored ended up being correct.


http://fivethirtyeight.com/interactives/world-cup/

Ideally what you would want to do is take a large sample, look at all the games where they gave say, a 75-80% chance to win, and see how often that team won.

As for them just picking the favorite; I sometimes think the favorite isn't as obvious as you think though. (Thinking to the ND/Stephen Austin game, for instance, or Wichita State and Gonzaga in round 1)

 

[pulaski74]

I don't know if you follow soccer or not. However, homefield advantage matters more in soccer than sport I have ever followed.

The home team in soccer has a huge advantage. Brazil is the most successful soccer nation of alltime and they were hosting the World Cup. Ask any of our soccer people Forza Azzurri IthacaMatt before the 2014 World Cup the heavy heavy favorite was Brazil.

Argentina, Spain, Germany were the other top teams but everyone thought Brazil would win at home. Once the first games were played though it was obvious Brazil was too depend on Neymar and he got hurt in QFs. Going into the Semis I doubt Brazil was an 80% favorite.

Live soccer is awesome, tv not so much. Soccer needs to be seen live, no way tv can convey what is happening. Sorry for going ot

 

[pulaski74]

Ideally what you would want to do is take a large sample, look at all the games where they gave say, a 75-80% chance to win, and see how often that team won.

As for them just picking the favorite; I sometimes think the favorite isn't as obvious as you think though. (Thinking to the ND/Stephen Austin game, for instance, or Wichita State and Gonzaga in round 1)

Funny thing, if SFA's best player would have stayed in control down the stretch they would have won that game.