NFL Betting, are you REALLY in it to win it?

Describe the problem

Sports betting is huge in American culture, and so is the NFL! However, no one likes to lose money, so finding strategies, superstitions, and lucky bets are a few ways that people try to get the best bang for their buck when it comes to betting on NFL games. The NFL offers numerous different features that makes betting on their games unique, including point spread betting, football squares, and weather affecting different stadiums in different ways. Click here if you are interested in learning some more about NFL betting, or here if you are curious as to how many people really bet sports in America. While there may not be one foolproof solution to always make the best bet, my aim with this project is to draw insight on the history of predictions and outcomes of NFL games and offer insight and advice to those looking to make a little money betting on the league where they play, for pay!

The Data

Here is a link to the data set I will be using. It has three different files, one with information on different stadiums, one with information on different teams, and one with information on every NFL game played since 1966 with some relevant info included like whether or not it was a playoff game and what the weather was like. For my project I was going to narrow it down to everything post 2000 or some other more recent metric because I don't think we need to look through data before that.

import pandas as pd

df_stadiums = pd.read_csv('nfl_stadiums_new.csv')

df_stadiums.head()

	stadium_name	stadium_location	stadium_open	stadium_close	stadium_type	stadium_address	stadium_weather_station_code	stadium_weather_type	stadium_capacity	stadium_surface	STATION	NAME	LATITUDE	LONGITUDE	ELEVATION
0	Acrisure Stadium	Pittsburgh, PA	2001.0	NaN	outdoor	100 Art Rooney Ave, Pittsburgh, PA 15212	15212	cold	65,500	Grass	USW00094823	PITTSBURGH ASOS, PA US	40.4846	-80.2144	366.7
1	Alamo Dome	San Antonio, TX	NaN	NaN	indoor	100 Montana St, San Antonio, TX 78203	78203	dome	72000	FieldTurf	NaN	NaN	NaN	NaN	NaN
2	Allegiant Stadium	Paradise, NV	2020.0	NaN	indoor	NaN	NaN	dome	65000	Grass	NaN	NaN	NaN	NaN	NaN
3	Allianz Arena	Munich, Germany	NaN	NaN	outdoor	NaN	NaN	moderate	75,024	Grass	NaN	NaN	NaN	NaN	NaN
4	Alltel Stadium	Jacksonville, FL	NaN	NaN	NaN	NaN	NaN	NaN	NaN	NaN	NaN	NaN	NaN	NaN	NaN

df_teams = pd.read_csv('nfl_teams.csv')

df_teams.head()

	team_name	team_name_short	team_id	team_id_pfr	team_conference	team_division	team_conference_pre2002	team_division_pre2002
0	Arizona Cardinals	Cardinals	ARI	CRD	NFC	NFC West	NFC	NFC West
1	Atlanta Falcons	Falcons	ATL	ATL	NFC	NFC South	NFC	NFC West
2	Baltimore Colts	Colts	IND	CLT	AFC	NaN	AFC	AFC East
3	Baltimore Ravens	Ravens	BAL	RAV	AFC	AFC North	AFC	AFC Central
4	Boston Patriots	Patriots	NE	NWE	AFC	NaN	AFC	NaN

df_games = pd.read_csv('spreadspoke_scores.csv')

df_games[10:]

	schedule_date	schedule_season	schedule_week	schedule_playoff	team_home	score_home	score_away	team_away	team_favorite_id	spread_favorite	over_under_line	stadium	stadium_neutral	weather_temperature	weather_wind_mph	weather_humidity	weather_detail
10	9/11/1966	1966	1	False	Pittsburgh Steelers	34	34	New York Giants	NaN	NaN	NaN	Pitt Stadium	False	64.0	5.0	70.0	NaN
11	9/11/1966	1966	1	False	San Francisco 49ers	20	20	Minnesota Vikings	NaN	NaN	NaN	Kezar Stadium	False	60.0	25.0	75.0	NaN
12	9/11/1966	1966	1	False	St. Louis Cardinals	16	13	Philadelphia Eagles	NaN	NaN	NaN	Busch Memorial Stadium	False	72.0	5.0	70.0	NaN
13	9/11/1966	1966	1	False	Washington Redskins	14	38	Cleveland Browns	NaN	NaN	NaN	RFK Memorial Stadium	False	65.0	8.0	52.0	NaN
14	9/16/1966	1966	2	False	Los Angeles Rams	31	17	Chicago Bears	NaN	NaN	NaN	Los Angeles Memorial Coliseum	False	72.0	10.0	52.0	NaN
...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...
13511	1/22/2023	2022	Division	True	Buffalo Bills	10	27	Cincinnati Bengals	BUF	-6.0	48.5	Highmark Stadium	False	32.0	4.0	100.0	snow
13512	1/22/2023	2022	Division	True	San Francisco 49ers	19	12	Dallas Cowboys	SF	-3.5	46.5	Levi's Stadium	False	55.0	19.0	47.0	NaN
13513	1/29/2023	2022	Conference	True	Kansas City Chiefs	23	20	Cincinnati Bengals	KC	-1.5	48	GEHA Field at Arrowhead Stadium	False	22.0	13.0	55.0	NaN
13514	1/29/2023	2022	Conference	True	Philadelphia Eagles	31	7	San Francisco 49ers	PHI	-2.5	45.5	Lincoln Financial Field	False	52.0	14.0	48.0	rain
13515	2/12/2023	2022	Superbowl	True	Philadelphia Eagles	35	38	Kansas City Chiefs	PHI	-1.0	51	State Farm Stadium	True	76.0	8.0	8.0	retractable (open roof)

13506 rows × 17 columns

With this data set I will clearly be able to draw conclusions based on how accurate the spreads and predictions are and how the game ended up. There is so much data here in regards to what factors may go into a winning team like age of stadium, weather, home turf, what kind of game it is, and how accurate the lines are and whether or not people should make their bets based on that.

Specific Examples to Help Solve the Problem

• We'll cluster the top 5-10 teams from the past 10-20 years and look at how often they were the favorite to win a game, what their spread was, and how much they actually won or lost a game by to test the accuracy of the spread predictions.
• We'll cluster the top 5-10 teams from the past 10-20 years and look at their win percentages at home vs away vs neutral, as well as look at characteristics of the stadiums of these teams such as indoor/outdoor or how new the stadiums are, even the capacity to see how many fans could attend.
• We'll cluster the top 5-10 teams from the past 10-20 years and look at what conferences and divisions they are to see if having competitive rivals in their same division or conference is an indicator of success.

As you can see there are endless possibilities with this data set and many ways to look at the different factors going into what makes a winning NFL team. Based on our conclusions, we will be able to make recommendations to those who bet on NFL games and hopefully win some people some money (or at least stop them from losing as much ;)).