The SportsCenter clips vs. the spreadsheet. The eye test vs. the
analytics. "Seeing is Believing" vs. "Numbers Never Lie."
Sports fans have created this "rivalry" between film geeks and
analytics nerds when, in reality, those two libraries of knowledge
complement each other seamlessly. No human can watch every single
moment of every single contest, and likewise, no data point can
capture every literal aspect of a basketball game. The best writers
and experts I know can slide willingly between both schools, using
them to understand the nuances of the sport while covering for
blind spots and bias.
Stats are cool. They're easy to visualize, they make for
convenient comparing, and they can boil a lot of basketball down to
a single number. But there's a reason people don't have computers
running professional franchises. Metrics are often misunderstood,
misused and overemphasized, leading to some crazy conclusions
simply because a random set of numbers points in that direction.
There is no such thing as a bad stat, but it can quickly be
invalidated if the user doesn't understand it.
I'm not a statistician — I actually dropped my Statistics minor
in college because I was struggling to stay on top of a computer
programming class. But I love basketball, I love reading about
basketball, and I love trying to learn how some of the world's
greatest athletes do what they do. So I'm putting together my
philosophies on using stats to better understand the game, and
hopefully it'll at least give you a way to approach debates with
the giant asteroid field of numbers out there.
To me, almost all basketball stats fall under one of three
- Impact Estimators
No one category is better than another, but they each colorize a
player or team's impact in different ways. Let's get into each, and
I'll give you one of my favorite (free!) "advanced" stats from each
Event stats are pretty straight-up: What happened? Most of your
box-score stats are here; points, rebounds, assists, steals, fouls,
etc. all total up what went down in a single game or season.
They're often attached to a scoring play, a missed shot or a
turnover — something you can easily point to and say, "Oh,
something just happened there."
The problem with event stats are that they usually get
attributed to a single player, when in reality, all five players on
the team could have contributed. Did LeBron James receive a bad
pass that he had to hoist up late in the shot clock? Welp, that's a
missed field goal. Did Giannis Antetokounmpo have a wide-open dunk
because Khris Middleton drew two defenders away with a nice cut?
Middleton gets no credit. It's one of the many reasons people watch
film — to see what the stats don't capture.
However, event stats can be useful for collecting major plays,
and many more exist that don't get covered in your typical box
score. I've always been a fan of deflections, which you can find on the NBA's
official stats site. Deflections are, as written in the NBA's
glossary, "The number of times a defensive player or team gets
their hand on the ball on a non-shot attempt."
Most would agree that, even if a defender doesn't generate a
steal, getting an active hand on the ball disrupts the offense and
takes time off the shot clock. Here were the NBA's deflections
per-game leaders from last season:
1. Fred VanVleet, Toronto Raptors — 3.8 per game
2. Robert Covington, Portland Trail Blazers — 3.6 per
3. T.J. McConnell, Indiana Pacers — 3.5 per game
4. Jimmy Butler, Miami Heat — 3.5 per game
5. Ben Simmons, Philadelphia 76ers — 3.5 per game
How did Russell Westbrook average a triple-double? How did
Stephen Curry lead the league in threes made? Explainer stats dive
into how a player or team made the plays that they made and add
some context to the event stats.
These categories can range from the very general, such as field
goal percentage (that stat is a whole different conversation), to
the extremely detailed, i.e. three-point percentage after taking
seven or more dribbles. I love a good explainer because it allows
me to get specific as far as what I like about a player. Instead of
saying, "Nikola Jokic is a bucket because he scores a lot," I can
say, "Nikola Jokic is a bucket because he shoots threes
off-the-catch well, drives well, and is one of the best post-up
players in the league based on these stats." Which sounds more
convincing to you?
What's wrong with explainers? The nitty-gritty ones often rely
on player tracking data, which can range from iffy to downright
wrong at times. If you're tracking shooting defense, for example,
how does the tracking system know what the player's goals are and
who they should match up against? Maybe a play breaks down and a
defender has to sprint out to contest a shooter they weren't even
supposed to be guarding, but they might be labeled as the primary
So while these stats would ideally be extremely helpful, the
data should always be taken with a grain of salt. Watch the film
At-rim finishing is one of my favorite explainer stats. It's
easy to track and NBA teams are always looking for players who can
get easy points near the basket. Here were the leaders in
restricted area field goal percentage last year (per NBA Stats, min. 100 shot
1. Giannis Antetokounmpo, Milwaukee Bucks — 80.7%
2. DeAndre Jordan, Brooklyn Nets (now Los Angeles Lakers) —
3. Robert Williams III, Boston Celtics — 78.0%
4. Michael Porter Jr., Denver Nuggets — 77.5%
5. Daniel Gafford, Chicago Bulls/Washington Wizards —
With any stat, the ultimate question you're trying to answer is:
How good is Player X? But impact estimators are pretty explicitly
trying to assess a player's impact, whether it be in a specific
skill or their overall on-court value. Some of the most basic
estimators out there include offensive and defensive rating, as
well as plus/minus.
But impact estimators can get complicated and daunting.
Companies, front offices and branches hire folks whose sole
responsibility is to come up with a metric that can accurately
protray basketball impact. Formulas are tested and re-tested by
professional analytics minds who work extremely hard to create the
gold standard of a hoops quantifier. It's an admirable, albeit
To effectively wield impact estimators, you have to fully
understand how they are built and what they are trying to say. A
classic misstep is when people use defensive rating to say Player X
is a good defender. Defensive rating is a misleading name; it's a
team-based stat that shows how many points per 100 possessions the
team gave up while Player X was on the court. There's only so much
a player can do for their defensive rating if they're on a terrible
So don't be fooled because someone slapped a cool name on a
stat. Understand what other stats go into the formula and what it's
Here's a popular impact estimator: True Shooting percentage. The
formula is relatively simple and it provides a summary of a
player's efficiency for all shot values (here's a good article about it).
While most efficiency stats skew heavily toward bigs because they
take closer shots, True Shooting percentage has a slightly better
balance while still making sense. Here were the leaders from 2020-21:
1. Ivica Zubac, Los Angeles Clippers — 69.3
2. Rudy Gobert, Utah Jazz — 68.2 TS%
3. Joe Ingles, Utah Jazz — 67.2 TS%
4. Richaun Holmes, Sacramento Kings — 66.9 TS%
5. Mikal Bridges, Phoenix Suns — 66.7 TS%
Is Ivica Zubac the best shooter on the planet? Probably not. But
he was extremely efficient with the shots he got last season.
Stephen Curry is ranked 11th (65.5 TS%), and with him taking so
many threes, that's quite impressive.
So in my overall approach to using NBA numbers, I sort an
interesting stat into its larger category, I remember the pitfalls
of using that type of metric and I understand that a single number
will never provide the full flavor of seeing what the player did on
the court. And like I said, I am no expert; the goal should always
be to keep a growth mindset.
Stats, just like highlights, should never end a basketball
discussion — only make it more insightful.