Replies: 26 comments 40 replies
-
|
I don't understand the question |
Beta Was this translation helpful? Give feedback.
-
|
Hey Kyle ,
For the Xg boost model prediction says it predicts at a 67% ml win rate ?
52.5 on under/over ?
…On Thu, Mar 16, 2023 at 7:25 PM Kyle Skompinski ***@***.***> wrote:
I don't understand the question
—
Reply to this email directly, view it on GitHub
<#153 (comment)>,
or unsubscribe
<https://github.com/notifications/unsubscribe-auth/AY7NRWV35ZTQLSV2BG7XTOLW4PDRPANCNFSM6AAAAAAV55YQLE>
.
You are receiving this because you authored the thread.Message ID:
<kyleskom/NBA-Machine-Learning-Sports-Betting/repo-discussions/153/comments/5340737
@github.com>
|
Beta Was this translation helpful? Give feedback.
-
|
We've been training a bunch more models lately and things have changed. I have to go back and revisit how I got my original results. |
Beta Was this translation helpful? Give feedback.
-
|
Do you update the model daily ?
…On Thu, Mar 16, 2023 at 8:32 PM Kyle Skompinski ***@***.***> wrote:
We've been training a bunch more models lately and things have changed. I
have to go back and revisit how I got my original results.
—
Reply to this email directly, view it on GitHub
<#153 (reply in thread)>,
or unsubscribe
<https://github.com/notifications/unsubscribe-auth/AY7NRWTTV4354YVYHVNAGA3W4PLNBANCNFSM6AAAAAAV55YQLE>
.
You are receiving this because you authored the thread.Message ID:
<kyleskom/NBA-Machine-Learning-Sports-Betting/repo-discussions/153/comments/5341127
@github.com>
|
Beta Was this translation helpful? Give feedback.
-
|
Yes you should. It pulls data and if you don't update daily you wont have yesterday's stats. |
Beta Was this translation helpful? Give feedback.
-
|
Kyle can you tell me what prediction is more accurate ? how can the delta
between the two be so despair ?
…On Thu, Mar 16, 2023 at 8:33 PM Info ***@***.***> wrote:
Do you update the model daily ?
On Thu, Mar 16, 2023 at 8:32 PM Kyle Skompinski ***@***.***>
wrote:
> We've been training a bunch more models lately and things have changed. I
> have to go back and revisit how I got my original results.
>
> —
> Reply to this email directly, view it on GitHub
> <#153 (reply in thread)>,
> or unsubscribe
> <https://github.com/notifications/unsubscribe-auth/AY7NRWTTV4354YVYHVNAGA3W4PLNBANCNFSM6AAAAAAV55YQLE>
> .
> You are receiving this because you authored the thread.Message ID:
> <kyleskom/NBA-Machine-Learning-Sports-Betting/repo-discussions/153/comments/5341127
> @github.com>
>
|
Beta Was this translation helpful? Give feedback.
-
|
Hey aaclones, The difference between Money Line and OU results? Is that what you are trying task? |
Beta Was this translation helpful? Give feedback.
-
|
If so... It has to do with the implied probability of events happening. Vegas tends to get ML picks correct in the NBA about 67% of the time, with odds ranging from -105 to -1400 for probability. It is easier to get a higher accuracy when the assumed winner is more obvious. The OU comes with a -110 odds rate because it is the mean of what they modeled. The hit rate for this is 50/50, with a 52.5% rate needed to cover Vegas. Anything above that is killer. My OU model is 61% over the last 5 days. Just blowing away Vegas. |
Beta Was this translation helpful? Give feedback.
-
|
Absolutely love to hear it! |
Beta Was this translation helpful? Give feedback.
-
|
How are you applying the EV number that the model spits out to the OU bet? I get for ML you just bet on the team with positive EV, but wasn't sure how that translates to a point spread. |
Beta Was this translation helpful? Give feedback.
-
|
So since 3/15/2023 I am 66.66% on ML and 61.9% on OU. |
Beta Was this translation helpful? Give feedback.
-
|
The updated daily model has been lights out!!!! |
Beta Was this translation helpful? Give feedback.
-
|
i want to take a moment to thank you for you work ? i want you to be
compensated how do i go about doing so ?
…On Tue, Mar 21, 2023 at 11:50 AM JP23000 ***@***.***> wrote:
The updated daily model has been lights out!!!!
—
Reply to this email directly, view it on GitHub
<#153 (reply in thread)>,
or unsubscribe
<https://github.com/notifications/unsubscribe-auth/AY7NRWX2DCPKW3PJLH2GK53W5H2APANCNFSM6AAAAAAV55YQLE>
.
You are receiving this because you authored the thread.Message ID:
<kyleskom/NBA-Machine-Learning-Sports-Betting/repo-discussions/153/comments/5384977
@github.com>
|
Beta Was this translation helpful? Give feedback.
-
|
The totals more so ... for example sometimes the Xg boost model has a
absurd delta from the neural network , the neural network seems to be a lot
more on the conservative side . I am trying to figure out what seems to be
more accurate as i been monitoring it the last few days
…On Tue, Mar 21, 2023 at 11:51 AM JP23000 ***@***.***> wrote:
Hey aaclones,
The difference between Money Line and OU results? Is that what you are
trying task?
—
Reply to this email directly, view it on GitHub
<#153 (reply in thread)>,
or unsubscribe
<https://github.com/notifications/unsubscribe-auth/AY7NRWV6WPP6LL75GLAPDGTW5H2EVANCNFSM6AAAAAAV55YQLE>
.
You are receiving this because you authored the thread.Message ID:
<kyleskom/NBA-Machine-Learning-Sports-Betting/repo-discussions/153/comments/5384985
@github.com>
|
Beta Was this translation helpful? Give feedback.
-
|
Gotcha, yeah the XG Boost is the more accurate one to go with. |
Beta Was this translation helpful? Give feedback.
-
|
private message me , if possible or send me crypto wallet or anyway I
compensate you for this amazing project
…On Tue, Mar 21, 2023 at 11:54 AM JP23000 ***@***.***> wrote:
If so... It has to do with the implied probability of events happening.
Vegas tends to get ML picks correct in the NBA about 67% of the time, with
odds ranging from -105 to -1400 for probability. It is easier to get a
higher accuracy when the assumed winner is more obvious.
The OU comes with a -110 odds rate because it is the mean of what they
modeled. The hit rate for this is 50/50, with a 52.5% rate needed to cover
Vegas. Anything above that is killer. My OU model is 61% over the last 5
days. Just blowing away Vegas.
—
Reply to this email directly, view it on GitHub
<#153 (reply in thread)>,
or unsubscribe
<https://github.com/notifications/unsubscribe-auth/AY7NRWWLYVPJ7UETAA7LUBTW5H2ONANCNFSM6AAAAAAV55YQLE>
.
You are receiving this because you authored the thread.Message ID:
<kyleskom/NBA-Machine-Learning-Sports-Betting/repo-discussions/153/comments/5385012
@github.com>
|
Beta Was this translation helpful? Give feedback.
-
|
You really don't have to. |
Beta Was this translation helpful? Give feedback.
-
|
no it states on the xg boost model ratings that it's 66% and 52.5 on the
o/u , but Kyle already cleared it up
…On Tue, Mar 21, 2023 at 11:51 AM JP23000 ***@***.***> wrote:
Hey aaclones,
The difference between Money Line and OU results? Is that what you are
trying task?
—
Reply to this email directly, view it on GitHub
<#153 (reply in thread)>,
or unsubscribe
<https://github.com/notifications/unsubscribe-auth/AY7NRWV6WPP6LL75GLAPDGTW5H2EVANCNFSM6AAAAAAV55YQLE>
.
You are receiving this because you authored the thread.Message ID:
<kyleskom/NBA-Machine-Learning-Sports-Betting/repo-discussions/153/comments/5384985
@github.com>
|
Beta Was this translation helpful? Give feedback.
-
|
something seems off with the model the last few days the
statistical significance is very alarming , the totals are just very off
...
…On Tue, Mar 21, 2023 at 12:13 PM Info ***@***.***> wrote:
no it states on the xg boost model ratings that it's 66% and 52.5 on the
o/u , but Kyle already cleared it up
On Tue, Mar 21, 2023 at 11:51 AM JP23000 ***@***.***> wrote:
> Hey aaclones,
>
> The difference between Money Line and OU results? Is that what you are
> trying task?
>
> —
> Reply to this email directly, view it on GitHub
> <#153 (reply in thread)>,
> or unsubscribe
> <https://github.com/notifications/unsubscribe-auth/AY7NRWV6WPP6LL75GLAPDGTW5H2EVANCNFSM6AAAAAAV55YQLE>
> .
> You are receiving this because you authored the thread.Message ID:
> <kyleskom/NBA-Machine-Learning-Sports-Betting/repo-discussions/153/comments/5384985
> @github.com>
>
|
Beta Was this translation helpful? Give feedback.
-
|
for example it gave there are 3 games in the 60 percentile it should win
3-1 , thats not the case and the ev is not even close
…On Sat, Mar 25, 2023 at 2:35 PM Info ***@***.***> wrote:
something seems off with the model the last few days the
statistical significance is very alarming , the totals are just very off
...
On Tue, Mar 21, 2023 at 12:13 PM Info ***@***.***> wrote:
> no it states on the xg boost model ratings that it's 66% and 52.5 on the
> o/u , but Kyle already cleared it up
>
> On Tue, Mar 21, 2023 at 11:51 AM JP23000 ***@***.***>
> wrote:
>
>> Hey aaclones,
>>
>> The difference between Money Line and OU results? Is that what you are
>> trying task?
>>
>> —
>> Reply to this email directly, view it on GitHub
>> <#153 (reply in thread)>,
>> or unsubscribe
>> <https://github.com/notifications/unsubscribe-auth/AY7NRWV6WPP6LL75GLAPDGTW5H2EVANCNFSM6AAAAAAV55YQLE>
>> .
>> You are receiving this because you authored the thread.Message ID:
>> <kyleskom/NBA-Machine-Learning-Sports-Betting/repo-discussions/153/comments/5384985
>> @github.com>
>>
>
|
Beta Was this translation helpful? Give feedback.
-
|
This model isn't giving you Ev for o/u it just gives you a % on what it
thinks it going to happen in totals but it hasn't been as accurate as it
once was , and to further your question this has nothing to do with spreads
this purely gives you Ev Values if you were to make this bet over a course
of time you will be positive E/v but the model needs some tweaking since
the last few days as you can see the model has been getting blown out the
waters
…On Tue, Mar 28, 2023 at 8:32 AM skuan ***@***.***> wrote:
How are you applying the EV number that the model spits out to the OU bet?
I get for ML you just bet on the team with positive EV, but wasn't sure how
that translates to a point spread.
—
Reply to this email directly, view it on GitHub
<#153 (reply in thread)>,
or unsubscribe
<https://github.com/notifications/unsubscribe-auth/AY7NRWW7NWOZE6MYYHMPAVTW6MABVANCNFSM6AAAAAAV55YQLE>
.
You are receiving this because you authored the thread.Message ID:
<kyleskom/NBA-Machine-Learning-Sports-Betting/repo-discussions/153/comments/5455735
@github.com>
|
Beta Was this translation helpful? Give feedback.
-
|
I have read online bookmakers like Draftkings and Fanduel will limit or ban if you profit in the long run. I think we look like normal bettors, that just win more often than not, it would take a significant amount won before ban from what ive read... the unlimited money printer thing is not true even with a profitable model. Its similar to card counting, not illegal, but once they detect you winning too much they will limit you or ask you to take your business else where... Its not like you only get one chance tho, there are hundreds of casinos and bookmakers. |
Beta Was this translation helpful? Give feedback.
-
|
There's going to be a lot of variance in shorter periods of time. As for EV yes, players sitting out has a huge impact on this. It's always best to look at lineups for other things and make adjustments on your own. This is just a single tool. |
Beta Was this translation helpful? Give feedback.
-
|
I hit 60% playing every O/U each day from 3/1 on when I discovered this repository. Made 10k betting $100 a bet. Let's get the ability to scrape the player lineup and I can supply the last 7 years of NBA player data that I got from Big Data Ball. |
Beta Was this translation helpful? Give feedback.
-
|
As for bettor bans.... I will figure out where the article is, but they are not really concerned with people who use systems to win, at least to date. This may change with the proliferation of AI models, but I doubt it. IN some instances yes they ban you or restrict your amount to bet... but this is generally people who are winning really long shot odds bets that shouldn't have hit... and was suppose to be the money maker for the casino. And those they find that are gaming the system by doing something illegal like using other people to place bets for them (Like this dude who had a 17 leg parlay about to hit in the MLB playoffs, but he lived in a state that it was illegal to place a bet and there was no record of him being in the state where the bet was placed.) Or people finding arbitrage situations. Generally speaking with O/U the sportsbooks main goal is to maintain a balanced book. They set the line and then adjust odds and move the line according to where the bets are coming in at so that they can offset them. That's why the odds are -110 instead of -100 or +100. When vegas maintains a balanced book they make 5% no matter the outcome. Really this is those on the right side of the bet vs those on the wrong side of the bet. At least this is what I learned when I interviewed for their US Director for Improvements that they had open a while back. |
Beta Was this translation helpful? Give feedback.
-
|
I disagree, No one that remains profitable for a long enough period is aloud to play. The people that win big parleys will never get limited, why would you limit someone consistently giving money on losing bets?? they get lucky once like the lottery, and then lose it right back. The people that beat closing lines consistently get limited, "sharps" if you wana call it that, and a 10 leg parley isn't very sharp. #161 The only odds maker that doesn't really limit and is widely trusted is pinnacle sports. |
Beta Was this translation helpful? Give feedback.
-
|
You don’t have any scraping of updating lineups since players in/out skews
the data tremendously, and the whole key of proper data is to some extent
minimize variance as much as possible , players in or out actually
increases variance
…On Tue, Mar 28, 2023 at 12:39 PM Kyle Skompinski ***@***.***> wrote:
There's going to be a lot of variance in shorter periods of time. As for
EV yes, players sitting out has a huge impact on this. It's always best to
look at lineups for other things and make adjustments on your own. This is
just a single tool.
—
Reply to this email directly, view it on GitHub
<#153 (reply in thread)>,
or unsubscribe
<https://github.com/notifications/unsubscribe-auth/AY7NRWSCHDT4HAJ56DB6TPTW6M45JANCNFSM6AAAAAAV55YQLE>
.
You are receiving this because you authored the thread.Message ID:
<kyleskom/NBA-Machine-Learning-Sports-Betting/repo-discussions/153/comments/5458090
@github.com>
|
Beta Was this translation helpful? Give feedback.
-
|
Sure does. Adding player data is something im actively thinking about and wanting to work on. Just need the time. |
Beta Was this translation helpful? Give feedback.
-
|
Are you looking for help in adding player data? I'm new in this arena but this project is exactly what I was thinking about creating (with player data). i have a CS background but am green in the ML arena. |
Beta Was this translation helpful? Give feedback.
-
|
@jbolter Yes I am looking for help in incorporating player data. |
Beta Was this translation helpful? Give feedback.
-
|
That would be amazing, i have something started here https://github.com/nealmick/Merge-Data |
Beta Was this translation helpful? Give feedback.
-
|
is there a statistical significance in the case the model starts to be
really off and correction needs to be made ? Is there a threshold number ?
Youve done amazing work I am just wondering if and when it gets majorly
stressed test will the corrections be made ??
…On Tue, Mar 28, 2023 at 1:50 PM Kyle Skompinski ***@***.***> wrote:
Sure does. Adding player data is something im actively thinking about and
wanting to work on. Just need the time.
—
Reply to this email directly, view it on GitHub
<#153 (reply in thread)>,
or unsubscribe
<https://github.com/notifications/unsubscribe-auth/AY7NRWXSO5KXY7GYQIRLSYLW6NFKPANCNFSM6AAAAAAV55YQLE>
.
You are receiving this because you authored the thread.Message ID:
<kyleskom/NBA-Machine-Learning-Sports-Betting/repo-discussions/153/comments/5458684
@github.com>
|
Beta Was this translation helpful? Give feedback.
-
|
Working on some small things now dance cleaning things up. Will hopefully get it it soon. |
Beta Was this translation helpful? Give feedback.
-
|
This project on his own has Made me to take the decission to start venturing into AI. Thanks man ! I have Made fake bets / combi’s with Some nba knowledge and this tool and have for this matchday a completion rate of 80!!!% even a 45odd combi was correct. And this is without the new model. This is a gem 👌 |
Beta Was this translation helpful? Give feedback.
-
|
How will you guys update daily for this ? Are there any reliable sources
for api keys ?
…On Wed, Apr 19, 2023 at 7:30 AM Neal ***@***.***> wrote:
That would be amazing, i have something started here
https://github.com/nealmick/Merge-Data
Basically just merges data from my project and writes top 7 players by
play time to team data csv. ill link my project below aswell.
https://github.com/nealmick/bb
—
Reply to this email directly, view it on GitHub
<#153 (reply in thread)>,
or unsubscribe
<https://github.com/notifications/unsubscribe-auth/AY7NRWULVB2GA3RUQPG577LXB4PZLANCNFSM6AAAAAAV55YQLE>
.
You are receiving this because you authored the thread.Message ID:
<kyleskom/NBA-Machine-Learning-Sports-Betting/repo-discussions/153/comments/5654555
@github.com>
|
Beta Was this translation helpful? Give feedback.
-
|
balldontlie.io |
Beta Was this translation helpful? Give feedback.
-
|
Last time I checked balldontlie didn’t
…On Wed, Apr 19, 2023 at 7:48 AM Neal ***@***.***> wrote:
balldontlie.io
—
Reply to this email directly, view it on GitHub
<#153 (reply in thread)>,
or unsubscribe
<https://github.com/notifications/unsubscribe-auth/AY7NRWUSMO5VCSF4OGQHK7LXB4R3NANCNFSM6AAAAAAV55YQLE>
.
You are receiving this because you authored the thread.Message ID:
<kyleskom/NBA-Machine-Learning-Sports-Betting/repo-discussions/153/comments/5654630
@github.com>
|
Beta Was this translation helpful? Give feedback.
-
|
Pulling data from nba directly is fairly easy |
Beta Was this translation helpful? Give feedback.
-
|
this is true! |
Beta Was this translation helpful? Give feedback.
-
|
And how do you intend to fetch the data for each player daily ??
…On Sat, Apr 22, 2023 at 2:49 AM JP23000 ***@***.***> wrote:
I hit 60% playing every O/U each day from 3/1 on when I discovered this
repository. Made 10k betting $100 a bet. Let's get the ability to scrape
the player lineup and I can supply the last 7 years of NBA player data that
I got from Big Data Ball.
—
Reply to this email directly, view it on GitHub
<#153 (reply in thread)>,
or unsubscribe
<https://github.com/notifications/unsubscribe-auth/AY7NRWVT6SO6JQOQPNLXOMLXCLJC5ANCNFSM6AAAAAAV55YQLE>
.
You are receiving this because you authored the thread.Message ID:
<kyleskom/NBA-Machine-Learning-Sports-Betting/repo-discussions/153/comments/5689304
@github.com>
|
Beta Was this translation helpful? Give feedback.
-
Beta Was this translation helpful? Give feedback.
-
|
It might be easier to input the last game for each team. Would probably offer some performance. Just basically forming last game of each team same as usual would 3x the input data features, and offer alot of momentum short-term type data, kinda like streaks. Could make it trailing 2 or 3 games as-well. Along the same lines, the last game against the opponent team could be added for input as-well. These changes would make the first game of the season break, unless you track last game over multiple seasons, kinda difficult, i just removed first game's for my project. |
Beta Was this translation helpful? Give feedback.
-
|
Also another easy change would be to add categorical inputs for each team id. This would allow the model to tell one team from another. |
Beta Was this translation helpful? Give feedback.
-
|
@kyleskom what do you think of these potential changes ^^ |
Beta Was this translation helpful? Give feedback.
-
|
I think that would be beneficial for the top part. I don't like to know the team and we just predict based on teams stats. I think that could skew the results. Ill think on that as well. |
Beta Was this translation helpful? Give feedback.
-
|
I agree, probably would help account for injuries too if last game performance was bad. The main negative thing is it drives the API request up quickly. |
Beta Was this translation helpful? Give feedback.
-
|
i've tried a bunch of inputs it's so far off ... which categorical inputs ?
and how do you update each player individually on a regular basis ? that
player might not even be playing that day .. i cant find one api key that
allows me to automate
…On Sat, Apr 22, 2023 at 1:34 PM Neal ***@***.***> wrote:
Also another easy change would be to add categorical inputs for each team
id. This would allow the model to tell one team from another.
—
Reply to this email directly, view it on GitHub
<#153 (reply in thread)>,
or unsubscribe
<https://github.com/notifications/unsubscribe-auth/AY7NRWQ6ASSVX74KDRKXMW3XCQ6FFANCNFSM6AAAAAAV55YQLE>
.
You are receiving this because you authored the thread.Message ID:
<kyleskom/NBA-Machine-Learning-Sports-Betting/repo-discussions/153/comments/5696947
@github.com>
|
Beta Was this translation helpful? Give feedback.
-
|
team id categorical input. Please take some time to review API endpoints. |
Beta Was this translation helpful? Give feedback.
-
|
Neal , do you plan on doing it the exact same way but use player stats ?
Ive tried 4 libraries they are underperforming , i am trying to customize
some advance stats currently but again the problem is the api keys , the
rapid api end point doesn't provide the advanced stats for all the players
, nbastats does but can't automate with them
…On Sat, Apr 22, 2023 at 1:38 PM Info ***@***.***> wrote:
i've tried a bunch of inputs it's so far off ... which categorical inputs
? and how do you update each player individually on a regular basis ? that
player might not even be playing that day .. i cant find one api key that
allows me to automate
On Sat, Apr 22, 2023 at 1:34 PM Neal ***@***.***> wrote:
> Also another easy change would be to add categorical inputs for each team
> id. This would allow the model to tell one team from another.
>
> —
> Reply to this email directly, view it on GitHub
> <#153 (reply in thread)>,
> or unsubscribe
> <https://github.com/notifications/unsubscribe-auth/AY7NRWQ6ASSVX74KDRKXMW3XCQ6FFANCNFSM6AAAAAAV55YQLE>
> .
> You are receiving this because you authored the thread.Message ID:
> <kyleskom/NBA-Machine-Learning-Sports-Betting/repo-discussions/153/comments/5696947
> @github.com>
>
|
Beta Was this translation helpful? Give feedback.
-
|
I aggregate and correlate data across multiple api's, both historical and live. |
Beta Was this translation helpful? Give feedback.
-
What about this: https://github.com/swar/nba_api |
Beta Was this translation helpful? Give feedback.
-
|
From my experience the stats.nba.com endpoint ip bans if you make too many requests. I have also found request time to very too much, sometimes taking over 30 seconds to reply making page loads of new games too long. |
Beta Was this translation helpful? Give feedback.
-
|
this doesn't include advanced stats and i believe it's not uptodate
…On Sat, Apr 22, 2023 at 2:30 PM deperto ***@***.***> wrote:
Neal , do you plan on doing it the exact same way but use player stats ?
Ive tried 4 libraries they are underperforming , i am trying to customize
some advance stats currently but again the problem is the api keys , the
rapid api end point doesn't provide the advanced stats for all the players
, nbastats does but can't automate with them
… <#m_-5373284506311994157_>
What about this: https://github.com/swar/nba_api
—
Reply to this email directly, view it on GitHub
<#153 (reply in thread)>,
or unsubscribe
<https://github.com/notifications/unsubscribe-auth/AY7NRWSWCMWREKCHB7FQB3TXCREV3ANCNFSM6AAAAAAV55YQLE>
.
You are receiving this because you authored the thread.Message ID:
<kyleskom/NBA-Machine-Learning-Sports-Betting/repo-discussions/153/comments/5697109
@github.com>
|
Beta Was this translation helpful? Give feedback.
-
|
Includes many advanced stats, pulled directly from nba.com |
Beta Was this translation helpful? Give feedback.
-
|
What needs to be updated exactly? The predictions are used with the most updated team stats every single time you run. |
Beta Was this translation helpful? Give feedback.
-
|
How about for player stats ? In the event they aren’t in the lineup it
skews the whole entire data batch from the team we want to minimize
variance as much as possible , so I feel it’s important to do daily player
stats to match up with the team stats , I’ve been trying but fetching the
player stats is hard to automate
…On Sun, Apr 23, 2023 at 6:34 AM Kyle Skompinski ***@***.***> wrote:
What needs to be updated exactly? The predictions are used with the most
updated team stats every single time you run.
—
Reply to this email directly, view it on GitHub
<#153 (reply in thread)>,
or unsubscribe
<https://github.com/notifications/unsubscribe-auth/AY7NRWXIUKGVW4OFMXT4DXLXCRMITANCNFSM6AAAAAAV55YQLE>
.
You are receiving this because you authored the thread.Message ID:
<kyleskom/NBA-Machine-Learning-Sports-Betting/repo-discussions/153/comments/5697279
@github.com>
|
Beta Was this translation helpful? Give feedback.
-
|
This doesn't include player data and it skewed by injuries yes. I plan on working on that for next season. |
Beta Was this translation helpful? Give feedback.
-
|
I can pitch in I can show you what I’ve done so far the only problem I get
is automating for now I manually punch in the player data and it’s so
tedious , the last 3 days using what you have and including what I did I
successfully predicting 4 games the exact score but I am not sure if it
pure luck or not since my player predictions weren’t accurate
…On Sun, Apr 23, 2023 at 6:40 AM Kyle Skompinski ***@***.***> wrote:
This doesn't include player data and it skewed by injuries yes. I plan on
working on that for next season.
—
Reply to this email directly, view it on GitHub
<#153 (reply in thread)>,
or unsubscribe
<https://github.com/notifications/unsubscribe-auth/AY7NRWSUQDNC3CDIUCN7Q53XCRM7HANCNFSM6AAAAAAV55YQLE>
.
You are receiving this because you authored the thread.Message ID:
<kyleskom/NBA-Machine-Learning-Sports-Betting/repo-discussions/153/comments/5697292
@github.com>
|
Beta Was this translation helpful? Give feedback.
-
|
Team stats will only remain true if the standard deviation is constant ,
but as you see in nba its not and when a key player is out the effect it
has is linear , there needs to be a solution of some sort for these
scenarios.. I’ve explored a lot ways to try and solve this but the hard
part is automated the day to day data to ensure you know whose in or whose
out and updating the data since all these api don’t work even the expensive
ones , unless you manually just punch it in it’s been a tall order finding
a way
…On Sun, Apr 23, 2023 at 7:01 AM Neal ***@***.***> wrote:
I agree, probably would help account for injuries too if last game
performance was bad. The main negative thing is it drives the API request
up quickly.
—
Reply to this email directly, view it on GitHub
<#153 (reply in thread)>,
or unsubscribe
<https://github.com/notifications/unsubscribe-auth/AY7NRWU2XE7YN3TP5SZBM3TXCRPKZANCNFSM6AAAAAAV55YQLE>
.
You are receiving this because you authored the thread.Message ID:
<kyleskom/NBA-Machine-Learning-Sports-Betting/repo-discussions/153/comments/5697347
@github.com>
|
Beta Was this translation helpful? Give feedback.
-
|
Most likely multifactorial, and only linear on a hyperplane described by the coefficients slope and intercept. |
Beta Was this translation helpful? Give feedback.
-
|
I'll work on this and have it fully automated for next season. I have the idea of how to do this. |
Beta Was this translation helpful? Give feedback.
-
|
The top 3 players on the team will cause this
…On Sat, Apr 22, 2023 at 6:22 PM Neal ***@***.***> wrote:
Most likely multifactorial, and only linear on a hyperplane described by
the coefficients slope and intercept.
—
Reply to this email directly, view it on GitHub
<#153 (reply in thread)>,
or unsubscribe
<https://github.com/notifications/unsubscribe-auth/AY7NRWQK2MHWD3HKARUXB7TXCR755ANCNFSM6AAAAAAV55YQLE>
.
You are receiving this because you authored the thread.Message ID:
<kyleskom/NBA-Machine-Learning-Sports-Betting/repo-discussions/153/comments/5697629
@github.com>
|
Beta Was this translation helpful? Give feedback.
-
|
Incredible , cant wait !!!
…On Sat, Apr 22, 2023 at 6:27 PM Kyle Skompinski ***@***.***> wrote:
I'll work on this and have it fully automated for next season. I have the
idea of how to do this.
—
Reply to this email directly, view it on GitHub
<#153 (reply in thread)>,
or unsubscribe
<https://github.com/notifications/unsubscribe-auth/AY7NRWSOBYLFD3B4652W5TTXCSARLANCNFSM6AAAAAAV55YQLE>
.
You are receiving this because you authored the thread.Message ID:
<kyleskom/NBA-Machine-Learning-Sports-Betting/repo-discussions/153/comments/5697636
@github.com>
|
Beta Was this translation helpful? Give feedback.
-
how is this model 66% win rate ?
Beta Was this translation helpful? Give feedback.
All reactions