kyleskom / NBA-Machine-Learning-Sports-Betting

NBA sports betting using machine learning
1.16k stars 431 forks source link

Different result without -odds #253

Closed AllInOneYT closed 9 months ago

AllInOneYT commented 1 year ago

I don't understand why am I getting different result with the exact same data?

1 python3 main.py -xgb -odds=fanduel -kc

------------------fanduel odds data------------------
Denver Nuggets (-136) @ Miami Heat (116)
---------------XGBoost Model Predictions---------------
Miami Heat vs Denver Nuggets (55.0%): OVER 215 (70.6%)
------------Expected Value & Kelly Criterion-----------
Miami Heat EV: -2.71 Fraction of Bankroll: 0%
Denver Nuggets EV: -4.63 Fraction of Bankroll: 0%
-------------------------------------------------------

2 python3 main.py -xgb -kc

Denver Nuggets vs Miami Heat: 215
Denver Nuggets odds: -136
Miami Heat odds: 116
---------------XGBoost Model Predictions---------------
Denver Nuggets (69.8%) vs Miami Heat: UNDER 215 (56.1%)
------------Expected Value & Kelly Criterion-----------
Denver Nuggets EV: 21.09 Fraction of Bankroll: 28.94%
Miami Heat EV: -34.72 Fraction of Bankroll: 0%
-------------------------------------------------------
jlucas03 commented 10 months ago

Correct me if wrong, I havent looked at this code in a long time but I used it a bit last season. But -odds=parameter is what tells the code which odds you would like to see.

So for example in the first one you are seeing the odds that fanduels has because of -odds=fanduels

but the 2nd one doesnt have one set so it uses the default which could be like -odds=draftkings or another betting platform.

hope this helps

cpjolicoeur commented 10 months ago

@jlucas03 yeah that isnt fully correct. When you dont provide the -odds prompt, the app script prompts you to manually enter the odds data (it doesnt pull the odds from a "default" sportsbook).

The OP is asking why the model gives different results if he enters the same manual odds values that the -odds=fanduel is fetching.

cpjolicoeur commented 10 months ago

@AllInOneYT I'm not personally seeing that same effect you are. I get the same output when running the same model against manually inputted odds or using the -odds parameter.

Are you still seeing this issue when running the latest code and models?