What Actually Predicts Winning in the NBA
Final Project
Who am I and what is this?
Hello! My name is Tiernan Shibles. I’m from Providence, Rhode Island, and I am a junior at Xavier University studying Business Analytics and Information Systems. I chose to analyze NBA team-level game data because I have always been a huge basketball fan and wanted to use data to answer a question I find myself debating all the time: What actually separates winning teams from losing ones? I hope you enjoy!
Introduction
Over the last decade, the NBA has undergone one of the most dramatic tactical shifts in professional sports history. The mid-range jumper has given way to a three-point explosion, pace has accelerated, and analytics has quietly reshaped how teams are built. But beneath all of that evolution, one question stays the same: What actually separates winning teams from losing ones?
Is it shooting efficiency? Rebounding? Taking care of the ball? In this analysis, I examine over a decade of NBA team-level game log data, spanning the 2012-13 through 2024-25 seasons, to identify which in-game statistics are most strongly associated with winning. Every row in the data-set represents one team’s box score in one game.
My two main research questions are:
- Which in-game statistics are most strongly associated with winning?
- Has what it takes to win in the NBA changed over the last decade?
The dataset was sourced from Kaggle and can be accessed here.
Data Dictionary
The dataset contains one row per team per game. Below is a breakdown of each variable:
| Variable | Description |
|---|---|
GAME_ID |
Unique game identifier |
TEAM_NAME / TEAM_ABBREVIATION |
Team name and abbreviation |
HOME_TEAM |
The home team for the game |
FGM / FGA / FG_PCT |
Field goals made, attempted, and percentage |
FG3M / FG3A / FG3_PCT |
Three-pointers made, attempted, and percentage |
FTM / FTA / FT_PCT |
Free throws made, attempted, and percentage |
OREB / DREB / REB |
Offensive, defensive, and total rebounds |
AST / STL / BLK / TO / PF |
Assists, steals, blocks, turnovers, fouls |
PTS |
Points scored |
PLUS_MINUS |
Point differential |
EFG_PCT |
Effective field goal % — adjusts for the added value of three-pointers |
PIE |
Player Impact Estimate — a composite efficiency metric |
COVID_FLAG |
1 if the game was played during the COVID bubble season |
RESULT |
Game outcome: 1 = Win, 0 = Loss |
SEASON |
Season identifier (e.g., 24 = 2024-25) |
WIN_PCT |
Team’s win percentage at the time of the game |
Part 2: Descriptive Analysis
Now that we have a solid understanding of the dataset, let’s start exploring the data to answer our research questions. We’ll begin by looking at the most fundamental question in basketball: how does scoring relate to winning?
No surprise here, winning teams score more points. But what’s interesting is how tight the distributions are. The median winning score sits around 110 points while losing teams hover around 100, meaning the difference between winning and losing on any given night is often just a handful of buckets. This raises the question: if the margin is that thin, which specific aspects of the game push a team over the line?
Effective field goal percentage adjusts for the fact that three-pointers are worth more than two-pointers, making it a more accurate measure of shooting efficiency than standard field goal percentage. The separation here is striking, winning teams consistently shoot at a higher EFG% than losing teams, with very little overlap in the middle of the distributions. In the modern NBA, getting good shots and making them is key.
Turnovers are often called self-inflicted wounds. While the difference here is less dramatic than shooting efficiency, the trend is clear: losing teams give the ball away more. Every turnover is a missed scoring opportunity on one end and a potential fast break points opportunity for the opponent on the other. Taking care of the ball is a quiet but consistent ingredient in winning basketball.
The data tells a story that I as a basketball fan, have seen unfold in front of my eyes. Since the 2012-13 season, the average number of three-point attempts per game has nearly doubled. What was once considered a low-percentage gamble is now the cornerstone of modern offensive strategy. This shift raises an important question, has what it takes to win changed along with it?
Looking across multiple statistics at once paints a fuller picture of what winning basketball looks like. Winning teams lead in assists, rebounds, steals, and blocks, while committing fewer turnovers. Assists stand out in particular, suggesting that ball movement and team cohesion are hallmarks of winning teams. Defense, often undervalued in the highlights era, shows up clearly here through steals and blocks.
Home court advantage is one of the most debated concepts in sports, and the data confirms it is very real in the NBA. Home teams win at a noticeably higher rate than road teams, a pattern that has held consistently across the entire data set. Crowd energy, travel fatigue, and familiarity with the arena all likely play a role. It is worth keeping this in mind when evaluating team performance, where a game is played matters just as much as how a team plays.
Part 3: Secondary Data Source
To supplement the primary data-set, I scraped league-wide per game averages from Basketball Reference, one of the most trusted sources of NBA statistical data. While our primary data-set contains individual team-game observations, the Basketball Reference data provides season-level league averages, giving us a broader lens to evaluate how the NBA as a whole has changed over time.
Each row in this data-set represents one NBA season and includes league-wide averages for key statistics like three-point attempts, assists, turnovers, pace, and shooting efficiency. This makes it a perfect complement to our primary data, rather than just asking what winning teams do differently on a given night, we can now ask whether winning teams are simply ahead of league trends or whether they are doing something fundamentally different from the rest of the league.
By combining both sources, we can make comparisons like:
- As the three-point revolution took hold, at what point did shooting more threes actually start translating to more wins?
- As the league has gotten faster, have winning teams been able to score more points along with it?
These comparisons give us a different angle and help us understand whether the traits of winning teams are truly distinctive or simply a reflection of broader league trends.
The story this chart tells is more nuanced than expected. Back in 2013, teams that shot a lot of threes were actually slightly less likely to win, the strategy simply wasn’t proven yet and most rosters weren’t built for it. By 2016, everything changed. Three-point volume became the single strongest predictor of winning in our entire data-set, teams that bombed away from deep won at a dramatically higher rate than those that didn’t. This was the golden window for early adopters.
But by 2019 and 2022, the advantage had shrunk considerably. As every team in the league started launching threes at a record pace, simply attempting more of them stopped being enough to stand out. The three-point shot went from a secret weapon to a basic requirement, and the teams still winning are the ones doing everything else right too.
As league pace has climbed steadily since 2012, winning teams have scored more points right along with it. More possessions mean more scoring opportunities, and winning teams have taken full advantage. What’s striking is how closely the two lines track each other, every uptick in pace corresponds with an uptick in winning team scoring. In the modern NBA, speed and points go hand in hand, and the teams best equipped to play at a high pace are the ones putting up the biggest numbers on the winning side of the box score.
Conclusion
So, what actually predicts winning in the NBA? After examining over a decade of team-level game data and cross-referencing it with league-wide trends from Basketball Reference, a few clear themes emerge.
Shooting efficiency is the single biggest separator. Winning teams consistently shoot a higher effective field goal percentage than losing teams, and that gap has held steady even as the entire league has gotten more efficient. Getting good shots and making them is the most reliable path to winning on any given night.
Ball movement and ball security go hand in hand. Winning teams record more assists and commit fewer turnovers than their opponents, not occasionally, but consistently across every season in the data-set. These aren’t flashy stats, but they show up in the win column night after night.
The three-point revolution has changed what winning looks like, but not what it requires. Teams that embraced the three-point shot early reaped massive rewards, but as the whole league caught up, simply shooting more threes stopped being enough. Today, three-point volume is a baseline expectation, the teams winning are the ones combining high volume with high efficiency.
The bar to win has never been higher. As pace has increased and scoring has climbed, winning teams have had to keep up every step of the way. The modern NBA rewards teams that can play fast, shoot well, move the ball, and protect it, all at the same time.
For all NBA fans the takeaway is simple: winning in the NBA has always been about efficiency, discipline, and execution. The style of play has changed dramatically since 2012, but the formula for winning really hasn’t.