2023-2024 MBB computer projections thread

[NorthCyd]

You just multiply the %s together. Like if you flip a coin four times. The odds of each flip being heads are 50% but the odds of every flip being heads are 0.5*0.5*0.5*0.5 = 6.25%

Right. Some look at odds on our schedule and think we are favored in every game so winning every game should be the most likely outcome but it's going to be much more difficult than that.

 

[clone4life82]

You just multiply the %s together. Like if you flip a coin four times. The odds of each flip being heads are 50% but the odds of every flip being heads are 0.5*0.5*0.5*0.5 = 6.25%

Again I’m not a stats nerd (I like stats but only took a couple stat classes in college ) but is it really just as simple as multiplying out the odds like that flipping a coin like that? If you told me Kansas has 60-70% odds like that for their last four games every season for the last 20 years, how many of those seasons did they win out vs not? I’d bet it’s more than 23% don’t you think?

 

[NorthCyd]

Again I’m not a stats nerd (I like stats but only took a couple stat classes in college ) but is it really just as simple as multiplying out the odds like that flipping a coin like that? If you told me Kansas has 60-70% odds like that for their last four games every season for the last 20 years, how many of those seasons did they win out vs not? I’d bet it’s more than 23% don’t you think?

That's more a question of how accurately whatever model you are using predicted the odds of outcomes. It's never going to be easy to determine if a team simply outperformed their normal ability over a stretch of games or if the model is not accurately predicting outcomes. It's why folks like kenpom are constantly tweaking their models.

 

[rosshm16]

Again I’m not a stats nerd (I like stats but only took a couple stat classes in college ) but is it really just as simple as multiplying out the odds like that flipping a coin like that? If you told me Kansas has 60-70% odds like that for their last four games every season for the last 20 years, how many of those seasons did they win out vs not? I’d bet it’s more than 23% don’t you think?

Basically you are asking (for example), do we really have a 66.3% change of winning @ KSU? No one can answer that. If we accept those percentages as correct, then it is correct that mathematically our odds of winning all four games are 23%, by the statistical definition of the term "odds".

 

[clone4life82]

That's more a question of how accurately whatever model you are using predicted the odds of outcomes. It's never going to be easy to determine if a team simply outperformed their normal ability over a stretch of games or if the model is not accurately predicting outcomes. It's why folks like kenpom are constantly tweaking their models.

That’s why it feels to me (again just a feel) that just multiplying each games’ predicted outcomes isn’t that accurate and that I would think it’d be more likely that (than the percentage when multiplying them out-what was it like 23%?) that ISU wins out.

 

[MeanDean]

Again I’m not a stats nerd (I like stats but only took a couple stat classes in college ) but is it really just as simple as multiplying out the odds like that flipping a coin like that? If you told me Kansas has 60-70% odds like that for their last four games every season for the last 20 years, how many of those seasons did they win out vs not? I’d bet it’s more than 23% don’t you think?

Just look at the last 3 games. Basically a 1/3 chance to lose each one. So multiplying it out - even though we're favored in each game - there is a 100% STATISTICAL likelihood we lose one of the three. That's at this point in time.

If we win the first one then it's only a 2/3 chance of losing one of the last 2.

Then if we win the next, then there's only a 1/3 chance of losing another game.

As the results actually occur, it resets the totals/percentages.

 

[ebState]

That's more a question of how accurately whatever model you are using predicted the odds of outcomes. It's never going to be easy to determine if a team simply outperformed their normal ability over a stretch of games or if the model is not accurately predicting outcomes. It's why folks like kenpom are constantly tweaking their models.

KP and Torvik are based on Bayesian probability. Basically it has an idea of how good or bad it thinks you are and when you out perform or under perform it adjusts what it thinks of you.

 

[CloniesForLife]

It is weird to me that the same system that projects us as the #8 team in the country also simultaneously projects us as a #3 seed.

Maybe the system is building in the "you are not a national brand so the committee's going to f*** you" factor.

The system that has us as the #8 team is purely analytics based on our efficiency. The other one takes into account our actual resume and who we have beaten

 

[NENick]

Honestly they need to win out.

Need?
I'm pretty confident they can, but losing any of the remaining 4 wouldn't shock me. The road games in particular will be tough.