Five Thirty Eight's Prediction Model

[ochoopsfan]

Her Hoops Stats predicts UConn will reach the Final Four.

Nice to get some positive results for my $20 annual subscription.

 

[Golden Husky]

And one: this is getting waaay over my head...not hard to do.

I'm not a betting person but I do read occasionally the betting column in the NY Post.
Their 'expert' has lots of info and strategies.

One is about 'Joes' and 'Sharks'. Joes are regular type fans and tend to bet lower amounts.
'Sharks' are bigtime gamblers who do lots of research and tend to bet much larger amounts.

Example: men's tourney...round one. Vermont [13 seed] plays Arkansas [4 seed] in Buffalo.
Arkansas is favored by a few points...say 5. The Post 'expert' says that most bets are coming in
are on Vermont...but they tend to be small bets....from Joes. He says that while there are fewer bets on Arkansas they are larger ones...coming from Sharks. More money is on Arkansas...not more bets....but more money.

Result: the odds went down some...maybe to 4 points for Arkansas...trying to urge more Vermont bets whether small or not....to even out the amount bet on each team.

Uhhh...glad I do not bet.

BTW: rooting for Vermont! They are from New England....and if UConn wins Rd 1, they would face the winner of Vermont/Arkansas on Saturday...I figure it would be harder to take down the Razorbacks.

A bit to unpack here:

First, unsophisticated bettors are called "squares." Sophisticated bettors are referred to as "sharps." A very sophisticated bettor is called a "triple sharp." (Don't ask me why there's no such thing as a "double sharp"). Of course, the Post journalist can use his own terminology but "squares" and "sharps" are the more widely used words.

Except for the Super Bowl, squares have little impact on the line; they just don't bet enough. Yes, it's the amount of money wagered that influences the line, not the amount of bets, @Blakeon18 .

Making and adjusting a line is a reasonably complicated task, which helps explain why today's bookmaker is far more likely to have a BA in sportsbook management and a working knowledge of mathematics than be some colorful character with a fedora and a diamond pinky ring. Adjusting the line involves concepts such as "the ladder principle," THP (theoretical hold percentage), PHP (practical hold percentage) and "extension" (the amount of money a book is willing to risk losing on a game).

As mentioned earlier in the thread, the line's purpose is to predict the wagering, not the outcome of a game. Oddsmakers work for the house, designing a structure that usually provides a profit for the house, irrespective of which team wins the game.

I hope this helps.

 

[BlueSkyGLW]

In my working life we were contracted by Fortune 500 companies to evaluate their methodologies and models for performance management and capacity planning of their computer and network systems. In almost all cases the models severely reflected the analysts' bias. I suspect 538 is guilty of the same error.

Agree but in this case, the error is not just analysts' bias but the data used....historical for the season. UConn became a whole team after all their OOC games, the greatly improved performances of the team were against lesser teams....no analyst would include projections based on unknowns without publishing the expected error band. So, the prediction cannot reflect the expected performance of this team as it is now constituted against better opponents nor is it able to account for the intangible gains it made in surviving the ordeal.

Having said all that, given the UConn team where it is now, I feel much, much better about their chances than I did of the US Olympic Hockey team against the Soviets in 1960 at Lake Placid. We know what happened to all the predictions for that game.....

 

[GoUconn7961]

Didn't see this thread and posted the link to the article in another. I found the percentages interesting.

 

[LoboDays]

My local purveyor of probabilities who can be found in the back booth at O’Malley’s Bar has an easier way of determining odds and point spreads. He says if I want the favorite, I give 5 points. If I want the underdog, I get 4.

 

[Golden Husky]

My local purveyor of probabilities who can be found in the back booth at O’Malley’s Bar has an easier way of determining odds and point spreads. He says if I want the favorite, I give 5 points. If I want the underdog, I get 4.

With that much vig, the rounds should be on him.

 

[LoboDays]

With that much vig, the rounds should be on him.

They have to be on him. He’s the only one with money.

 

[TeamFirst!]

I read that Uconn has a much better chances of winning against NC State if they play each other because of the Bridgeport home court advantage for Uconn. Apparently they must think uconn has a high probability of not making it to the NC state level game if NC State is almost 3 times more Likely to win the NC than Uconn. Also, as much as we all respect Nan's analysis, I find it hard to believe that it's all based on just which team people are betting on. With the kind of money involved I would think there are many other factors including real sports analytical expertise.

 

[fomalhaut]

Any model based on the entire season, with no adjustments for recent play, is likely to be wildly inaccurate.

 

[LoboDays]

If NC State and UCONN meet each other, it means they are both playing well. Can’t see either team being a large point spread fav. Maybe a couple of points for the “home” team.

 

[Redux]

If NC State and UCONN meet each other, it means they are both playing well. Can’t see either team being a large point spread fav.

Correct. The probabilities listed are not just winning the championship game but include winning all the intervening games. As two teams win, their probabilities will converge. SC, for example, is regarded as a much bigger favorite to win its early games and get there than is UConn; that's why they're rated at 46% vs 6%, in NO WAY implying that SC is anywhere near 8X more likely to beat UConn, as that differential will start to fall away with each intervening win.

So in your example of NC St and UConn, if they actually meet in regional, they will likely be rated very close. And if Paige turns out to be Paige and/OR the UConn players currently playing above form continue to do so, UConn is likely to be favored.

 

[jchusky]

concern for Uconn to me has always been the level of competition to finish the season. This isn't 2016 anymore, teams have much more depth and transfer portal has leveled the playing field. I look at Nc st and they have 3 5+ year srs, that experience just can't be duplicated. The notion is that Uconn has looked good the past month, but the best win is Nova a team that barely got in the tournament. Meanwhile Sc has played 5 or 6 tournament teams in teh span of 4 weeks. The big 10 and Sec have had numerous teams in the tournament. The 538 to me not my favorite, but BPI saying 7% is definitely interesting because BPI is maybe the most accurate thing for predictions. Just to use an example last season in the Men's tourny, the 3 favorites to win the ntl title were : Baylor, Gonzaga, and Houston. All 3 made the final 4.

 

[ballerfan]

With Paige Bueckers Back, How Far Can UConn Go?

After the Connecticut women’s basketball team lost Paige Bueckers in December, and the Huskies returned to the court with a lackluster 44-point effort in a loss…

fivethirtyeight.com

New 538 article about UConn.

 

[LETTERL]

Nice to get some positive results for my $20 annual subscription.

I hope, for your sake, Her Hoops Stats doesn't turn out to be another Literary Digest.

 

[darkstar1]

Las Vegas isn’t predicting the winner of any sports competition - it’s predicting bettor behavior. The odds just mean that Las Vegas thinks a lot of people will bet on UConn. That’s why odds change before any competition rather than remain fixed, they’re adjusting to ensure 50% of bettors are betting on each team.

Yes, and that is why they are much truer odds then what any expert or computer model will give you. Would love to see the experts put their $ on their picks. Something tells me their picks might have a different order. The favorite is determined by the odds.

 

[darkstar1]

A bit to unpack here:

First, unsophisticated bettors are called "squares." Sophisticated bettors are referred to as "sharps." A very sophisticated bettor is called a "triple sharp." (Don't ask me why there's no such thing as a "double sharp"). Of course, the Post journalist can use his own terminology but "squares" and "sharps" are the more widely used words.

Except for the Super Bowl, squares have little impact on the line; they just don't bet enough. Yes, it's the amount of money wagered that influences the line, not the amount of bets, @Blakeon18 .

Making and adjusting a line is a reasonably complicated task, which helps explain why today's bookmaker is far more likely to have a BA in sportsbook management and a working knowledge of mathematics than be some colorful character with a fedora and a diamond pinky ring. Adjusting the line involves concepts such as "the ladder principle," THP (theoretical hold percentage), PHP (practical hold percentage) and "extension" (the amount of money a book is willing to risk losing on a game).

As mentioned earlier in the thread, the line's purpose is to predict the wagering, not the outcome of a game. Oddsmakers work for the house, designing a structure that usually provides a profit for the house, irrespective of which team wins the game.

I hope this helps.

I have never seen ESPN or any other sports network list a favorite/team favored to win as any team other than the one with the best odds. Watch where the “smart” money goes.

 

[Skeets]

Yes, and that is why they are much truer odds then what any expert or computer model will give you. Would love to see the experts put their $ on their picks. Something tells me their picks might have a different order. The favorite is determined by the odds.

I think you need to define "favorite". Certainly you do not mean the better team. The whole intent of the "odds" is to balance the book. Balancing the book is the surest and safest way for the bookie to turn a profit. Bookies never, ever gamble (something to think about before you lay your next $50 down). Bookies run a business. Now, your neighborhood bookie doesn't control the odds. If his book gets out of balance he'll lay off to a higher, regional bookie. The regional guy would probably lay off to Vegas if necessary. Not sure how this works in the era of legalized gambling. Who does the giant commercial sportsbook houses lay off to? Or are they so big that they can withstand their books being slightly out of balance?

 

[Conndog]

Has anyone seen NCSt play? I haven’t, so I’m curious what are their strengths. I know they have 6’5 Cunane at center, what is her game like, is she a power player like Boston, or is she more finesse like Liv? Who else is dangerous, who will be a matchup problem for UConn? Do they rely on outside shooters and have lots of them? Do they play a primary group of 6-7 players or do they use more like UConn can now? Is their defense their strong suit? I know very little about them and am curious who is likely to guard who.

 

[Waquoit]

Yes, and that is why they are much truer odds then what any expert or computer model will give you. Would love to see the experts put their $ on their picks. Something tells me their picks might have a different order. The favorite is determined by the odds.

Line makers don't just try to "even out the money", these days they try to entice the public to bet on the loser. If all bets are evened out, they just make 10%. They want to do better.

 

[eebmg]

Nice to get some positive results for my $20 annual subscription.

For $50, they can make a custom prediction where UConn wins the NC.

 

[LoboDays]

I think you need to define "favorite". Certainly you do not mean the better team. The whole intent of the "odds" is to balance the book. Balancing the book is the surest and safest way for the bookie to turn a profit. Bookies never, ever gamble (something to think about before you lay your next $50 down). Bookies run a business. Now, your neighborhood bookie doesn't control the odds. If his book gets out of balance he'll lay off to a higher, regional bookie. The regional guy would probably lay off to Vegas if necessary. Not sure how this works in the era of legalized gambling. Who does the giant commercial sportsbook houses lay off to? Or are they so big that they can withstand their books being slightly out of balance?

The online sports books lay off just like the Vegas books do. They get burned (or have a nice payday) when the majority of late money is lopsided on one team throughout and there is no one to lay off to.

 

[LETTERL]

Has anyone seen NCSt play? I haven’t, so I’m curious what are their strengths. I know they have 6’5 Cunane at center, what is her game like, is she a power player like Boston, or is she more finesse like Liv? Who else is dangerous, who will be a matchup problem for UConn? Do they rely on outside shooters and have lots of them? Do they play a primary group of 6-7 players or do they use more like UConn can now? Is their defense their strong suit? I know very little about them and am curious who is likely to guard who.

NC State runs mainly a "1-in...4-out" system. If Cunane is defended closely, she'll pass the ball out to the perimeter, where any of the 4 outside shooters can sink the THREE. We play mainly 8 players. The first two off the bench are speedster Diamond Johnson and then Jada Boyd, who brings an extra level of physicality to the lineup when she replaces Kayla Jones. Kai Crutchfield is our best defender. Junior Camille Hobby spells Cunane, either for breaks or in the rare instances where Cunane gets into foul trouble.

If we have to go deeper, freshmen Azaiah James and Jessica Timmons and sophomore Genesis Bryant are all capable shooters; James and Timmons are both solid perimeter shooters.

When both are in the game, Cunane and Boyd will often switch off, so that it is Boyd playing down low and Cunane haunting the perimeter, where she is more than willing and able to shoot the THREE when she's open.

If there is a down side to Cunane's game, it is that she can be dominated by more physical players....and often plays rushed, sometimes having games where she misses a lot of "bunnies" and with low shooting percentages. However, down the stretch of the regular season and in the ACCT, Cunane was outstanding, seeming to raise her game to another level.

I am, at the same time, anxious and excited for a potential showdown with Connecticut, if the seeds hold and both teams advance with no injuries or illness. A lot would be riding on that outcome...in what I would expect would be an intense, close affair. For that to happen, though...NC State will have to get over the hurdle that seems to be the Sweet 16 round; the Wolfpack Women have made 15 trips to the Sweet 16 but have advanced to the Elite Eight just once.

 

[VA Huskie]

NC State runs mainly a "1-in...4-out" system. If Cunane is defended closely, she'll pass the ball out to the perimeter, where any of the 4 outside shooters can sink the THREE. We play mainly 8 players. The first two off the bench are speedster Diamond Johnson and then Jada Boyd, who brings an extra level of physicality to the lineup when she replaces Kayla Jones. Kai Crutchfield is our best defender. Junior Camille Hobby spells Cunane, either for breaks or in the rare instances where Cunane gets into foul trouble.

If we have to go deeper, freshmen Azaiah James and Jessica Timmons and sophomore Genesis Bryant are all capable shooters; James and Timmons are both solid perimeter shooters.

When both are in the game, Cunane and Boyd will often switch off, so that it is Boyd playing down low and Cunane haunting the perimeter, where she is more than willing and able to shoot the THREE when she's open.

If there is a down side to Cunane's game, it is that she can be dominated by more physical players....and often plays rushed, sometimes having games where she misses a lot of "bunnies" and with low shooting percentages. However, down the stretch of the regular season and in the ACCT, Cunane was outstanding, seeming to raise her game to another level.

I am, at the same time, anxious and excited for a potential showdown with Connecticut, if the seeds hold and both teams advance with no injuries or illness. A lot would be riding on that outcome...in what I would expect would be an intense, close affair. For that to happen, though...NC State will have to get over the hurdle that seems to be the Sweet 16 round; the Wolfpack Women have made 15 trips to the Sweet 16 but have advanced to the Elite Eight just once.

Thanks for the great summary.

 

[darkstar1]

I think you need to define "favorite". Certainly you do not mean the better team. The whole intent of the "odds" is to balance the book. Balancing the book is the surest and safest way for the bookie to turn a profit. Bookies never, ever gamble (something to think about before you lay your next $50 down). Bookies run a business. Now, your neighborhood bookie doesn't control the odds. If his book gets out of balance he'll lay off to a higher, regional bookie. The regional guy would probably lay off to Vegas if necessary. Not sure how this works in the era of legalized gambling. Who does the giant commercial sportsbook houses lay off to? Or are they so big that they can withstand their books being slightly out of balance?

We don’t disagree. We just look at this issue a little differently. I understand how it works, but the favorite is simply the team/horse, etc. predicted to win as established by the odds set. The favorite IS usually the perceived “better”team, player or horse, etc at least by the oddsmaker when the line is established. Yes, the line may move, depending on how $ is invested by the betting public. Think of a horse race. Is the favorite the better horse? Who knows. Upsets happen all the time and the “better” horse doesn’t always win. Look what happened to Kentucky. St Peters won, but few would say they are the better team. I look at odds as if the public is contributing their version of rankings, just like any expert. So to me the futures line making SC 1 and UConn/Stanford 2 before a pretty big drop off is akin to a ranking of favorites. Anyway, we’ve this horse to death and none of it matters to UConn’s succes, whether they are rated as 1, 2 or 20. Just win baby.