uOEC 2018 Programming Challenge

Overview

Todays companies are sitting on mountains of data, and are increasingly looking for cleaver engineers to structure this data and produce useful insights.

Since the rise of bitcoin people are more comfortable with gambling their money away online. This years challenge is to channel you inner addictive gambler and marry that with your dank coding skills to design an algorithm to make you rich (or entirely broke)

Problem Statement

You are developing an algorithm to predict the outcomes of tonights NBA games based on any pre-grame data. Your application must be able to read a csv file with the following schema

Month,Day,Year,Start (ET),Visitor,Home

And output computed predictions in the csv format

Month,Day,Year,Start (ET),Visitor,Home,Outcome,Confidence

Outcome	a value of either 'home' or 'away' representing the winning team
Confidence	an associated confidence score (0 to 1) of the prediction (bonus points)

At the end of the competition you will be required to predict tonights games only using any data sources you wish. You may also use any programming language of choice, but a jupyter notebook enviorment is recommended.

Provided Data

The following two datasets are provided along with the scrapper that generated them which can be found in data/scrape_standings.py. You are not limited to these datasets and are extremely encouraged to construct your own based on the provided scrapper for things such as injuries, local vs visitor weather, player stats etc,.

data/2017schedule.csv

This file contains a list of all games in the 2017 regular season along with the points for completed games.

Month,Day,Year,Start (ET),Visitor,Visitor PTS,Home,Home PTS,Score,,Notes

For games which are yet to be completed the points value is empty along with the Score

data/nba.db

This sqlite database contains team standings for various dates throught the 2017 regular season The following values are included:

date
team_name text
games integer
minutes integer
field_goals integer
field_goal_attempts integer
field_goal_percentage decimal
three_points integer
three_point_attempts integer
three_point_percentage decimal
two_points integer
two_point_attempts integer
two_point_percentage decimal
free_throws integer
free_throw_attempts integer
free_throw_percentage integer
offensive_rebounds integer
defensive_rebounds integer
total_rebounds integer
assists integer
steals integer
turn_overs integer
personal_fouls integer
points integer

Evaluation

The following is the grid for evaluation, the team with the most points after evaluation wins

Criteria	Max Points	Description
Code Quality	30	Readability of the code Functionality of the code This score is based entirely on judges
Additional Data Sources	20	The use of additional relevant data sources no points given for using irrelevant data sources regardless of how correlated (ex. babies born before game day) 5 points per data source
Prediction Algorithm	40	The understanding and ability to explain the Prediction Algorithm used The quality of features selected for algorithm Presentation of feature research in the form of graphs and charts to identify key features
Predicting the games for the night	10	`score = sum(confidence * outcome)` where outcome is +1 for correct geuss and -1 for incorrect geuss

All teams will conduct a 10 minute presentation of which 5 minutes are a designated question period.

Submissions

In order to create a submission you must fork this repositiory by clicking the Fork button on the main page As soon as you have forked the repository please submit the link to the competition organizer, your repository must be public to be graded you can then push commits to your forked version.

NikhilPeri/nba-betting