Home-Field Advantage in the NFL (2000-2023)

Course: DTSC-2301 Data Science Foundations (UNC Charlotte)
Tools: Python, pandas, matplotlib, seaborn
Dataset: nflverse game-level data (games.csv)

Why this project matters

People often say NFL teams play better at home. We hear this all the time from fans, TV analysts, and betting discussions. But instead of assuming it is always true, I wanted to check the data.

This project looks at two things:

  1. Does home-field advantage show up in the data?
  2. Has that advantage changed over time?

1) Problem definition

Research question

To what extent does home-field advantage influence NFL game outcomes, and how has that effect changed from 2000-2023?

Project goal

This is an exploratory project. I am trying to understand patterns in past data, not build a prediction model.

Why someone might care

2) Data description and sample design

Data source

What each row means

Each row is one NFL game.

Main columns used

Other indicators that were available but not used in final analysis

The dataset also includes other fields such as overtime (overtime) and additional game context columns. I reviewed these, but I did not include them in the final analysis because they are outside this project’s scope.

This project focuses on one core question: how home win rate changed over time. Adding many extra indicators at once would make the story less clear for this assignment.

Sample size and filtering steps

What this data does not include directly

This dataset does not directly include some things that could matter, like:

So we can describe patterns, but we should be careful about saying exactly why those patterns happen.

3) Cleaning and preparation decisions

Decision A: Use 2000-2023

Why: It gives enough years to study trends, while still focusing on the modern NFL.

Tradeoff: We leave out older seasons and newer seasons after 2023.

Decision B: Focus on regular-season games

Why: Regular-season data is more consistent year to year, which helps trend comparisons.

Important note: playoff rows in this dataset are labeled WC, DIV, CON, and SB (not POST). Because of this, the current postseason export logic produced an empty postseason file. To keep results reliable, I focused final claims on the regular-season sample.

Decision C: Make score columns numeric and check for missing values

Why: This makes sure win/loss calculations are valid.

Result: No regular-season rows were removed, but this check is still important.

Decision D: Handle ties separately

Why: Ties should not be counted as losses.

Decision E: Check for duplicate games

I verified that game_id values are unique in the final sample.

4) Visual analysis

Visual 1: Home win rate by season

Home win rate by season

Why this chart: It shows how home win rate changes year by year.

What it shows:

Visual 2: Home win rate with 5-year rolling average

Home win rate with 5-year rolling average

Why this chart: It makes the long-term trend easier to see.

Why 5-year instead of 2- or 3-year: A 2-year or 3-year average still jumps around a lot and can make normal ups and downs look like major trend changes. A 5-year average smooths short-term noise so the long-term direction is easier to read.

I also did not use a very long window because that can hide important shifts. So 5 years is a good middle option: smooth enough to see the big pattern, but still able to show major changes like 2020.

Visual 3: Home scoring margin distribution

Distribution of home margin

Why this chart: Win rate is only yes/no. Margin adds more detail about game outcomes.

What it shows:

5) Main findings

Finding 1: Home-field advantage is real

Across 6,175 regular-season games, home teams won 56.26% of games.

Finding 2: Home advantage looks weaker than in the early 2000s

Finding 3: 2020 is the lowest season in this sample

Finding 4: There is some rebound after 2020

Home win rate rises after 2020, but still appears lower than many early-2000s years.

Margin context

6) Interpretation

The best conclusion is:

This project describes patterns in the data. It does not prove a single cause.

7) What would be misleading to say

It would be misleading to claim:

The data supports trends, not final cause-and-effect proof.

8) Limitations, assumptions, and reflection

Main limitations

Assumptions

Reflection

A small percentage change can still matter across thousands of games. At the same time, careful analysis means being honest about what the data can and cannot tell us.

9) Next steps (if I continue this project)

10) References and transparency

Data source

Transparency

View full code

GitHub Repository:
DTSC-2301-Project-1