Summer League Stats You Can Trust

Which ones are sticky and which ones are icky

Welcome back to another edition of The F5. This week were talking summer league and how to separate the signal from the noise.

As a monthly reminder, this newsletter exists because hundreds of readers like you have deemed it worthy of supporting with their own money. If you find the contents of this email interesting or worthwhile, then I encourage you to consider upgrading to a paid subscription for $5/month. Paid subscribers to The F5 receive weekly tutorials in R (and now also Python) that detail how I create the charts and analysis found throughout this newsletter. If you’re a student or can’t justify the cost of a subscription at this time, get in touch and we’ll work something out.

Now let us turn our attention to summer league. God help you if you’re reading this from Vegas.

I hardly watch any college or international basketball. So summer league is my first chance to form on-court impressions for most of the rookie class. If you’re in the same boat as me, then summer league can be a useful viewing experience. When you know next to nothing about a player there’s a lot you can learn, even in a series of meaningless exhibition games. Things as obvious as “That guy looks bigger/faster/stronger than everyone else” will jump off the screen immediately. But there are deeper and more interesting insights to be found if you know where to look.

Back in 2021 I published an analysis of which stats from summer league can tell us something about a player’s corresponding rookie season. It’s been a few years since then and the data coming out of summer league has gotten better (read: my ability to access and analyze said data has gotten better). So I thought I would use this as an opportunity to refresh which summer league stats are “stickiest” — or rather, which ones are most predictive of how a player will perform during their rookie season.

To put it another way, what can we reasonably expect from a player’s rookie season if all we have to go on is their summer league performance? Which box score stats should we pay attention to and which ones should we disregard? Whose statistical indicators coming out of this year’s summer league are promising and whose raise concerns?

To find out, I tracked 25 different statistics across summer league and the NBA for 485 different rookies since 2008. Since summer league games are only 40 minutes long, I used per 36 versions of familiar box score stats (i.e, Points per 36 Minutes instead of Points per Game) and other rate-based stats that are less sensitive to differences in raw playing time.

The multi-faceted chart below shows the relationship between a player’s summer league and rookie season across the various indicators I tracked.

Each dot on the chart represents a player’s value in a given statistical category at summer league (x-axis) and their value in that same category in their rookie season (y-axis). For this analysis, I’m only looking at rookies since 2008 that logged at least 50 minutes¹ in their first summer league and 250 minutes in their corresponding rookie season².

(Note: when I say “Summer League” I’m referring to not just Vegas Summer League, but also the California Classic and Salt Lake City Summer Leagues).

The stats in the chart are ordered from most to least predictive, as measured by R-Squared. As a refresher, R-Squared values range from zero to one and in this case a value of one would mean a player’s summer league performance in a specific statistical category perfectly explains their performance in that same statistical category in their rookie season. Meanwhile, a value of zero would mean there’s no discernible relationship between a player’s summer league performance and rookie season performance.

While none of the stats have a perfect one-to-one relationship, I think there’s a clear hierarchy of stats we should pay attention to at summer league. To emphasize which stats are most predictive more clearly, here’s a bar chart of the different stats ordered by their R-Squared value. I’ve arbitrarily grouped the stats into three buckets: Sticky Stats, Icky Stats, and Other Stats.

Sticky Stats (stats we can trust)

Three-Point Attempt Rate, which is the percentage of a player’s shot attempts that come from three, has the single highest R-Squared value. Not far behind is Threes Attempted per 36 Minutes. You can interpret the placement of these stats on the chart to mean a player’s shot selection in summer league is fairly representative of what their shot selection will look like in the NBA. So if a player is bombing threes in summer league then chances are they’ll do the same at the NBA level.

Nothing about this should come as a surprise. Guards and Wings that run and gun are going to do the same regardless of the environment. Meanwhile Bigs that strictly rim run are going to continue to play their game even if the misses don’t count.

So let’s take a look at which rookies profile as three-point gunners and which ones are allergic to the three-point line based on their summer league stats. The interactive table below shows the 2025 draft picks by their three-point attempt rate across the summer league games they’ve played as of this writing³.

Egor Dëmin’s name stands out from the rest. Dëmin, who Brooklyn drafted with the 8th overall pick, took a whopping 85 percent of his shots from beyond the arc at summer league on 11 attempts per 36 minutes⁴. Those two numbers give us more insight into what his game might look like at the NBA level than anything else he showed at summer league, including his outstanding three-point percentage (more on that later).

Only three players last season in the NBA took more than 85 percent of their shots from three: AJ Green, Sam Merrill, and Nicolas Batum. Brooklyn should have no trouble finding minutes for Dëmin alongside their other players that need the ball as long as he’s willing to take threes at close to the rate he did in summer league. And based on historical precedent, there’s good reason to believe Dëmin will be letting it fly inside the Barclays Center next year.

It’s worth emphasizing that the inverse is also true. That is, if a player isn’t letting it fly in Vegas then it’s unreasonable to expect them to develop the confidence or willingness to start chucking them once October rolls around. Which is why I’m even less bullish on the fit of Derik Queen in New Orleans after seeing him in summer league.

Queen, who New Orleans selected with the 14th overall pick, attempted just two threes in summer league. I was already having a hard time imagining a lineup featuring both Queen and Zion Williamson (another offense-first, non-shooting Big) functioning on defense. But now I have concerns about the duo’s viability on the other end of the court as well. In fairness, Williamson has made chicken salad out of chicken shit while playing next to non-shooting Bigs like Steven Adams and Jonas Valančiūnas throughout his career. So if you’re a Pelicans optimist then maybe you see Queen’s irregular fit as a feature rather than a bug.

More Sticky Stats

A few other summer league stats that stand out for their stickiness include measures of how often a player blocks shots (Blocks Per 36), creates second chance opportunities (Offensive Rebounds Per 36), and creates scoring chances for others (Assists Per 36). What all of these stats have in common is that they are useful for describing how a player plays rather than how well they play. Meaning, if a player shows a willingness to pass the ball at summer league then they’ll likely carry over that same willingness into their rookie season. Same goes for players that swat shots and crash the glass.

The interactive table below shows this year’s crop of rookies and their stats in these stickier categories.

There are two names that stand out that I want to talk about for different reasons.

Cooper Flagg did not have a summer league to remember. In addition to poor shooting (46 percent True Shooting), Flagg had only five assists, one block, and no offensive rebounds across two games. Tom Thibodeau likes to say you don’t need to shoot well to play well and Flagg’s all-around game should have lent itself well to filling up the stat sheet. But that wasn’t the case. I’m not sounding any alarm bells, but I may (slightly) re-calibrate my expectations for him next season.

Meanwhile, Collin Murray-Boyles was the number nine overall pick by the Raptors and has looked as advertised in summer league. He’s a do-everything small ball Big that’s drawn comparisons to Draymond Green. He can’t shoot for shit, but he can pass, defend, and he leads all rookies in offensive rebounds at summer league. When you’re not a threat from beyond the arc you have to be close to perfect in every other facet of your game. So far, the early returns on Murray-Boyles’s game outside of his three-point shot look pretty good.

Icky Stats (stats we can’t trust)

Then there are the stats on the other end of the spectrum. These are the ones that are least useful for projecting what will happen in a player’s rookie season. Things like Plus Minus and Three-Point Percentage have almost no connection between the exhibition games in Vegas and the real games in the regular season.

That’s not really surprising. Plus-Minus, which tells us how well a team performs when a specific player is on the court, is more of a team stat than an individual one. And since teams at summer league bear little resemblance to their regular season counterpart (excluding Washington and Utah who seemingly sent half their roster to summer league) it’s no wonder the R-Squared is so low.

Meanwhile, Three-Point Percentage is a notoriously noisy stat that takes many more tries than one could reasonably attempt at summer league before it stabilizes. So it follows then that how well a player shoots on a limited number of attempts in Vegas hardly predicts how well they’ll shoot over a much larger sample during the regular season.

That might come as a relief for a player like VJ Edgecombe, the Sixers 3rd overall pick, who shot 15 percent on threes at summer league. With real NBA spacing and a NBA caliber point guard feeding him easier looks, Edgecombe’s accuracy from beyond the three-point line has essentially nowhere to go but up.

The last thing I want to mention is that all-in-one metrics like Game Score and Daily RAPM Estimate (DRE) have at best a weak relationship between summer league and the regular season. These metrics attempt to summarize a player’s performance by assigning weights to different box score stats and summing them up into a single number. Both metrics reward scoring efficiency and penalize turnovers and fouls, which tend to be more common at summer league. Ultimately, I wouldn’t recommend putting too much stock in them unless of course they paint a favorable picture of your favorite player.

Closing Thoughts

Remain calm if a rookie doesn’t perform well at summer league in the conventional sense. Shooting efficiency (2P%, 3P%, eFG%, and TS%) and production (Game Score and DRE) in summer league are not usually representative of a player’s shooting efficiency and production during the regular season.

Instead, focus on the stats that describe what kind of player they might be. You don’t need more than a few games to get a sense for who might be typecast as a floor-spacer, a shot blocker, or a threat on the offensive boards. Whether they’ll be any good in those roles is a different story that is yet to be written.

1. I used 50 minutes as a cutoff because that’s typically about the amount of minutes the top prospects tend to play at summer league. This year, Cooper Flagg, the number one overall pick, played just 63 minutes across two games in Vegas before getting shut down for the remainder of summer league.

2. This does not include players like Chet Holmgren that played in summer league, but missed their entire rookie season. Nor does it include players that appeared in an NBA game before their first summer league game. The sample here is limited to players that went straight from summer league to their rookie season, which means no data for the 2020-21 season since there was no summer league that year due to Covid.

3. There’s a still a few days left of summer league, but many of the most intriguing rookies have already been shut down. So I think there’s no harm in taking a sneak peak at these stats now.

4. As a comparison, only 50 percent of Dëmin’s shots attempts came from three during college where he launched about six threes per 36 minutes.