/sci/ - Science & Math » Thread #13723157

310KiB, 1242x807, E-o7GG5WUAUzI8G.jpg

View Same Google iqdb SauceNAO

Anonymous Wed 06 Oct 22:00:42 2021 No.13723157 View Reply Original Report

Quoted By: >>13723222 >>13723227 >>13723229 >>13723368 >>13723373 >>13723380 >>13723393 >>13723738 >>13724006 >>13724016

Will statistical machine learning eventually be able to predict the outcome of any sports game?

Anonymous

Anonymous Wed 06 Oct 2021 22:18:50 No.13723195 Report

Quoted By:

Yes

Anonymous

Anonymous Wed 06 Oct 2021 22:23:17 No.13723211 Report

Quoted By: >>13723218

no
you cant account for luck in a game
at the highest level luck is often the deciding factor

Anonymous

Anonymous Wed 06 Oct 2021 22:26:48 No.13723218 Report

Quoted By: >>13723730

>>13723211
This is how people who don't understand high level competition always interpret things, luck isn't the big issue, it's more just that some times people have good and bad days, even in poker a game that should be mostly luck, high level players have skill gaps

Anonymous

Anonymous Wed 06 Oct 2021 22:27:07 No.13723222 Report

Quoted By:

>>13723157
It will be able to tell the equity and ROI of any sports bet by calculating the
probabilities of the possible outcomes. This has been done numerous times already.
Look up the documentaries on the horse race betting incident in Hong Kong
https://www.youtube.com/watch?v=efBG1YfuNhg

Anonymous

View Same Google iqdb SauceNAO 1629537985606.png, 790KiB, 720x757

Anonymous Wed 06 Oct 2021 22:30:06 No.13723227 Report

Quoted By: >>13723248 >>13723307

>>13723157
it doesn't even need to be right 100% of the time, If a computer can predict 51% of outcomes it is pretty much game over for sports betting.

Anonymous

Anonymous Wed 06 Oct 2021 22:30:34 No.13723229 Report

Quoted By:

>>13723157
back to /b/ schlomo
https://www.youtube.com/watch?v=jrWLEjPLqVg

Anonymous

Anonymous Wed 06 Oct 2021 22:39:49 No.13723248 Report

Quoted By: >>13723263

>>13723227
No. 85% would be better than most people.
51% is trash.

Anonymous

Anonymous Wed 06 Oct 2021 22:46:04 No.13723263 Report

Quoted By: >>13723267 >>13723286

>>13723248
Even the best gamblers struggle to win 50% of the time, if you look at the worlds best poker players they only win like 50-51% of hands but over a long enough period of time they come out as famous millionaires.

Anonymous

Anonymous Wed 06 Oct 2021 22:47:39 No.13723266 Report

Quoted By: >>13723280

It seems like a tall order to predict when someone is going to throw a game because they were offered a lot of money or an unfortunate accident.

Anonymous

Anonymous Wed 06 Oct 2021 22:47:52 No.13723267 Report

Quoted By:

>>13723263
If you win 50% of the time, and lose 50% of the time in sports betting, you will lose a ton of money.

Anonymous

Anonymous Wed 06 Oct 2021 22:52:27 No.13723280 Report

Quoted By:

>>13723266
Italian soccer teams are famous for this. An AI can still learn from the patterns enough
to make predictions that beat the bookie odds. Similarly for when referees in German soccer leagues are paid to favor specific teams.
The AI can figure out this kind of stuff from the raw data.

Anonymous

Anonymous Wed 06 Oct 2021 22:54:45 No.13723286 Report

Quoted By: >>13723315

>>13723263
Football team wins 8 times, and loses 8 times
Game 1, Bet 100, Win 110
Game 2, Bet 100, Win 110
Game 3, Bet 100, Win 110
Game 4, Bet 100, Win 110
Game 5, Bet 100, Win 110
Game 6, Bet 100, Win 110
Game 7, Bet 100, Win 110
Game 8, Bet 100, Win 110
Game 9, Bet 100, Lose 0
Game 10, Bet 100, Lose 0
Game 11, Bet 100, Lose 0
Game 12, Bet 100, Lose 0
Game 13, Bet 100, Lose 0
Game 14, Bet 100, Lose 0
Game 15, Bet 100, Lose 0
Game 16, Bet 100, Lose 0
Net bets: $1600
Net outcome : $880
Net profit: $-720

Oof

Anonymous

Anonymous Wed 06 Oct 2021 23:08:38 No.13723307 Report

Quoted By: >>13723315 >>13723323

>>13723227
Wow, I didn't even realize how stupid your logic is until I actually crunched the numbers.

If I bet on every home team to win in the NFL last year, I would have won 134 of 256 times, which is 52.34375%
LMAO

Anonymous

Anonymous Wed 06 Oct 2021 23:11:34 No.13723315 Report

Quoted By: >>13723323

>>13723286
The likelihood of a $1.10 team losing is close to zero though, 8 times in a row would never happen.

>>13723307
nice, why didn't you?

Anonymous

Anonymous Wed 06 Oct 2021 23:14:04 No.13723323 Report

Quoted By: >>13723347

>>13723307
2019 was 136 home wins
2018 was 157 home wins

>>13723315
Wow, you are actually retarded

Anonymous

Anonymous Wed 06 Oct 2021 23:22:03 No.13723347 Report

Quoted By: >>13723354 >>13723363 >>13723366 >>13723947

>>13723323
forget single wins, but say an AI can predict a multi bet with 5+ legs with over 51% accuracy that is revolutionary.

Anonymous

Anonymous Wed 06 Oct 2021 23:23:43 No.13723354 Report

Quoted By:

>>13723347
That would be pretty good.

Anonymous

Anonymous Wed 06 Oct 2021 23:30:02 No.13723363 Report

Quoted By:

>>13723347
>forget single wins, but say an AI can predict a multi bet with 5+ legs with over 51% accuracy that is revolutionary.
and then another group predicts what your AI will predict, thus betting on the same as you

Anonymous

Anonymous Wed 06 Oct 2021 23:31:22 No.13723366 Report

Quoted By:

>>13723347
51% total accuracy would have to be accurate about 87.4% for each of the 5 leg parlay,
89.4% for 6
90.8% for 7
91.9% for 8
92.8% for 9
93.5% for 10
94.1% for 11
94.5% for 12
95.0% for 13
95.3% for 14

Anonymous

Anonymous Wed 06 Oct 2021 23:31:47 No.13723368 Report

Quoted By: >>13723389

>>13723157
Lol even the white girl in the top left is looking with intense interest towards the stud brothas. White girls really love chocolate if you know what I mean. That's why you have to make sure to feed your girlfriend/wife chocolate because happy wife, happy life.

Anonymous

Anonymous Wed 06 Oct 2021 23:33:04 No.13723373 Report

Quoted By:

>>13723157
No. A system can only accurate simulate a less complex system than itself. A game is more complex than any non speculative simulated. Also any sufficiently complex simulate will in all probability be constrained by size requirement and subsequent light speed delay of signal exchange.

Anonymous

Anonymous Wed 06 Oct 2021 23:37:21 No.13723380 Report

Quoted By: >>13723386

>>13723157
Statistical prediction. Game will always result in an outcome as defined by a cessation of gameplay in any real meaning. There. Easy. We can take it one step further and predict that any game will originate from a defined pre-state of existence.

Anonymous

Anonymous Wed 06 Oct 2021 23:39:22 No.13723386 Report

Quoted By: >>13723394

>>13723380
What if the computer can change the pre state of existence by putting shock collars on players of the team it predicts will lose.

Anonymous

Anonymous Wed 06 Oct 2021 23:41:06 No.13723389 Report

Quoted By:

>>13723368
That is an illusion caused by the instantaneousness of the photo.
In reality she just turned her head to take in the enticing view of the handsome trouper
standing next to her.

Anonymous

Anonymous Wed 06 Oct 2021 23:42:02 No.13723393 Report

Quoted By:

>>13723157
i don't know but my neural network predicts that you are a faggot

Anonymous

Anonymous Wed 06 Oct 2021 23:43:08 No.13723394 Report

Quoted By:

>>13723386
Then the game improved in entertainment value. Why is this not a real thing yet? Make it like that camgirl thing where donations activate a vibrator. I'd pay-per-view any game that lets me shock a player that burns a penalty kick. Imagine the hivemind shocking some unpopular player for the entire 90 minute duration of the game. That would be hillarious.

Anonymous

Anonymous Thu 07 Oct 2021 02:32:17 No.13723730 Report

Quoted By:

>>13723218
Ok but from the perspective of a bettor who has no insight into the lives of the athletes, this is in fact a random variable and is therefore "luck." You do understand that, right; you don't think people are saying luck because they think the universe bends in some quantum way around a game, right? Like what do you possibly think luck is if not a random variable in a betting scenario?

Anonymous

Anonymous Thu 07 Oct 2021 02:37:08 No.13723738 Report

Quoted By:

>>13723157
No, but it could make really good guesses after a lot of training.

Anonymous

Anonymous Thu 07 Oct 2021 04:44:55 No.13723947 Report

Quoted By:

>>13723347
What do you think the average accuracy is for the current best model?

Anonymous

Anonymous Thu 07 Oct 2021 05:23:11 No.13724006 Report

Quoted By:

>>13723157
Take your bait to /pol/

Anonymous

View Same Google iqdb SauceNAO back-to-the-future.jpg, 96KiB, 1200x675

Anonymous Thu 07 Oct 2021 05:25:55 No.13724016 Report

Quoted By:

>>13723157
of course not. that's like asking "can i take a sports almanac and predict the future unambiguously?"

5040vision

5040vision Thu 07 Oct 2021 05:37:34 No.13724038 Report

Quoted By:

There is an interesting tendency in today's society for people to vastly overestimate a complicated model's ability to be precise and vastly underestimate how well a well thought out yet simpler model can predict something.

I have looked at how some of the system for determining "expected goals" in association football and they are outright laughable, yet they get treated in media as some sort of "precise science". As a teacher of Mathematical Statistics I can't avoid doing repeated facepalms when watching how really bad implementations from people with lacking abilities in statistics and mathematics somehow think they can counteract for it by being self-confident, use quasi-bullshit terms for their methodologies and the do massive PR campaigns for their nonsense.

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