/sci/ - Science & Math » Thread #12302250

64KiB, 1000x727, 1000+monte+carlo+simulations+of+sports+betting+history.jpg

View Same Google iqdb SauceNAO

Anonymous Tue 03 Nov 15:16:20 2020 No.12302250 View Reply Original Report

Quoted By: >>12302379 >>12302676 >>12303703

what would be the most useful/accurate statistical method of predicting outcomes for sports betting?

Anonymous

Anonymous Tue 03 Nov 2020 16:09:18 No.12302379 Report

Quoted By: >>12302436

>>12302250
Nothing. It's literal pseudostatistics used sonthat vrtting houses can have a justification for ripping you off.

Anonymous

Anonymous Tue 03 Nov 2020 16:30:49 No.12302436 Report

Quoted By: >>12302511 >>12303582

>>12302379
not really, there's some degree of corellation between certain types of data and the outcome of a game. i've had decent success modeling football games after a poisson distribution so it "works", i'm just asking if there are more advanced statistics that i could use.

by your logic nothing can be modeled and predicted so we might as well not try to

Anonymous

Anonymous Tue 03 Nov 2020 16:57:06 No.12302511 Report

Quoted By: >>12302528

>>12302436
Correlation is often pure coincidence.
Just google "weirdest correlations".

Anonymous

Anonymous Tue 03 Nov 2020 17:01:52 No.12302528 Report

Quoted By:

>>12302511
i agree, which is why i think statistical analysis most of the time is a futile endeavour. such as mine. however if by "pure coincidence" i'm able to get a few percentages in my favour in terms of chance than why should not at least entertain the idea a little

Anonymous

Anonymous Tue 03 Nov 2020 17:59:11 No.12302676 Report

Quoted By: >>12303591

>>12302250
you can't, if you don't track every player and have access to time series with their position (which you obviously don't)

Anonymous

Anonymous Tue 03 Nov 2020 22:19:25 No.12303582 Report

Quoted By:

>>12302436
>modeling football games after a poisson distribution
what the fuck lol

Anonymous

Anonymous Tue 03 Nov 2020 22:22:24 No.12303591 Report

Quoted By:

>>12302676
It's obviously not completely random though, it's one game with more or less set rules and two carefully cultivated teams consisting of players near the upper bounds of proficiency, so it must trend towards some model.

Anonymous

Anonymous Tue 03 Nov 2020 22:54:10 No.12303703 Report

Quoted By:

>>12302250
A simple classifier I guess.
Like a vanilla neural network with a logistic output layer, or a support Vector Machine.
Well, some ensemble of them would be even better.

Capcode	All Only User Posts Only Moderator Posts Only Admin Posts Only Developer Posts
Show Posts	All Only With Images Only Without Images
Deleted Posts	All Only Deleted Posts Only Non-Deleted Posts
Ghost Posts	All Only Ghost Posts Only Non-Ghost Posts
Post Type	All Only Sticky Threads Only Opening Posts Only Reply Posts
Results	All Grouped By Threads
Order	Latest Posts First Oldest Posts First

Your latest searches