The Fastest Path to the Top 50: Fantasy Basketball Statistical Analysis

The NBA season is not even halfway over, but the lull in the schedule created by the NBA Cup gave me time to start thinking about what types of data I could research in this coming offseason to become a more informed analyst and a better overall fantasy basketball manager.

And then I started thinking, "Why do I need to wait until the offseason? What if I forget about this and get caught up in other projects?" When it comes to covering fantasy sports, I research and write about baseball, football, and basketball - though I have always maintained that basketball is my favorite of the three and the one that I would like to spend more time digging deeper into.

One thing that I love about fantasy baseball is the number of different statistics that we have to research, graph, plot, and dissect. Sabermetrics and advanced statistics have created endless opportunities to examine the performance of baseball players to try to predict the likelihood of their improvement, decline, or continued success. So, I set out on a little project to try to do something similar with some advanced NBA statistics, and in this piece, I'll share what I was able to find.

Can Advanced Stats Help Predict Top-50 Performance?

After having some conversations with other fantasy basketball analysts, including my good friend Dan Besbris, I decided to go on the hunt for some statistics that usually aren't considered when we are building our rankings or making our selections in drafts.

And I chose to use the top 50 as the group of players to analyze. Why? Because those are the difference-makers, the league-winners, the players who can propel us to fantasy basketball championships.

In a 10 or 12-team league, you are only going to get the opportunity to draft four or five of these players from the previous season. One path to a winning season is finding other players who are being drafted outside the top 50 who have the potential to finish there and, of course, to avoid any landmines (aka players who could drop in the rankings the following season). It's about finding value - plain and simple.

Injuries happen, and I'm not interested in trying to predict them. Yes, we can avoid players who we may perceive to be injury-prone, but what I'm attempting to do in this experiment is locate the types of players, based on their statistical profiles, who are going to hit more often than not when they are healthy.

So, which stats did I focus on? I'll get to that in a minute, but the stats that I wanted to avoid were the main stats that we often look at when evaluating players for fantasy purposes. In 9-category leagues, we hone in on counting stats (points, rebounds, assists), defensive stats (blocks and steals), shooting percentages (both field goal percentage and free throw percentage), three-pointers made, and turnovers.

One advanced statistic that is now commonly cited (especially for my DFS peeps) is usage rate (how often a team's possession is used by a player). We love high usage rates because you can't score points or hand out assists without having the ball in your hands during a possession. Low-usage players can still be very productive but likely have to do so through efficiency and/or defense.

Stats and Methodology

So besides usage, what other advanced statistics are readily available for us to analyze? The NBA tracks a lot of stuff, and if you've never explored their stats page, you should!

I ended up with four advanced statistics that I thought COULD be important and that I wanted to investigate further.

Assist to Turnover Ratio (AST/TO)

It's a simple statistic and not a complicated formula to calculate - ha! However, with one stat, we can measure how much a player contributes in two categories. It's an efficiency metric. High-assist players are also usually high-turnover players, but players who can deliver value in assists without racking up turnovers are surely better, right? And assists are the rarest of the counting stats.

Effective Field Goal Percentage (EFG%)

In the modern NBA, not all shots are created equal and we continue to see the number of three-point shots taken each year go up. So comparing a player who shoots 45% from the field with the majority of those shots being taken behind the arc to a player who shoots 55% from the field on mainly two-point shots might not be all that useful.

EFG% adjusts for three-point shooting in an attempt to capture the added value of making three-pointers. An efficient three-point assassin like Garrison Mathews has the same EFG% (64%) as a big man like Goga Bitadze, who makes his living on shots around the rim.

True Shooting Percentage (TS%)

Another shooting efficiency metric is true shooting percentage factors in free throws, whereas EFG% does not. In essence, how efficiently does a player score points every time the ball leaves their hand heading towards the basket? Since we value field goal percentage and free throw percentage in 9-cat, I figured it was worth investigating further.

Player Impact Estimate (PIE)

This one is the most complicated and nuanced as PIE attempts to measure a player's overall positive contribution to his team during the game. The NBA created PIE, which also incorporates defensive statistics, unlike John Hollinger's PER (player efficiency rating). Any metric that includes some defense interests me, as blocks and steals are much more scarce statistics than points, rebounds, and threes.

This data was tabulated nearly a week ago, so most of the players listed here have played a game or two since then, but their rankings and metrics are not likely to have changed much since then.

To capture the top 50, I used a composite ranking made up of each player's Yahoo! ranking and Basketball Monster 9-cat ranking. To account for the players not matching up perfectly, I extended our sample group to the top 55.

Advanced Stats: The Top-50

(click to enlarge)

I added some color coding to the chart to help the best performers in each category stand out.

The only regular NBA starter with a better AST/TO than Chris Paul and Tyrese Haliburton is Tyus Jones (4.9), who is 75th on Yahoo. Jimmy Butler sports an impressive 3.7 mark while not being a point guard.

A couple of Knicks (Jalen Brunson and Josh Hart) and Derrick White are the only others in this group over 3.0. Other notable NBA regulars over 3.0 are Fred VanVleet (3.9), Mike Conley (3.6), and Payton Pritchard (3.1), and the league average is around 1.7.

Those who are strong in EFG% are mainly the same guys who stand out in TS%, too. The big men dominate here, with Jarrett Allen and Walker Kessler shining, followed by Nikola Vucevic and Domantas Sabonis. Josh Hart and Pritchard have impressive marks for non-bigs.

Other rotational players not listed here who are excelling this season would be Daniel Gafford (71.6, rank of 116 on Yahoo) and Jalen Duren (70.2, ranked 162).

We have some inefficient shooters in this group who are below the league average in EFG% (51.8) - LaMelo Ball, James Harden, Trae Young, Dyson Daniels, Tyrese Maxey, Alperen Sengun, Brandon Miller, DeMar DeRozan, and Jamal Murray.

The average TS% is 54.7, so the excellent free throw shooting on significant volume from Harden, Ball, and DeRozan move them ahead of the mean, while the others are still below.

And finally, we have some very interesting PIE results. Of all four of the statistics, each player's PIE rating was the most highly correlated with their composite ranking (-.57 correlational coefficient). The other three statistics showed little to no correlation within the sample group.

In layman's terms, as we go up the chart, we are more likely to get higher PIE ratings near the top, and the color gradient helps illustrate that here, as well as the darkest green is concentrated in the top 15 or so of the rankings. Sengun and Giannis are the only two players above 17 who are outside the top 10. We may have found something here.

Results

What I did first was compare the averages from our sample group (top 50) with the rest of the league to see how much better our top group was than the rest of the league in each statistical category.

The least impressive statistic for the top 50 group was EFG%, as the average was only 9% higher than the league average. We already discussed the fact that nine guys in the top 50 group were below league average in that metric, so that tracks.

TS% was a bit better at 10.6%, but there's still not enough distance between the top 50 and the rest of the pack, and we still have six players under the league average.

The top 50 group was nearly 20% better in AST/TO than the league average. However, the number of players who are below the mean jumps to 17! This is problematic and likely due to the nature of how concentrated assists are among a relatively small group of players.

A lot of our most efficient passers are in the top group, but not all of them, and we have Wemby all the way up at No. 2 overall, averaging as many turnovers as assists.

But look at PIE! Our average PIE rating for the top 50 was 55% higher than the league average - that's substantially more concentrated among our top 50 players!

The NBA average is around nine, so we have only two players in our top 50 group (OG Anunoby and Daniels) who fall below the league average. Those two are good examples of inefficient offense players who derive a lot of their value from defensive stats. Heck, if you cut Daniels' league-leading 3.1 steals per game in half, he'd drop out of the top 50 altogether.

Let's look into PIE a little more. Here's another way to display the leaders (PIE of 14 or higher) - with every player who is averaging at least 15 minutes per game displayed and their current Yahoo rank.

(click to enlarge)

I went ahead and highlighted those players from outside our top 50 group with two different colors. The yellow group consists of players who we probably wouldn't expect to make this list based on their role or reputation.

We have all big men here, with Mark Williams leading the way. Moritz Wagner is done for the year and is easily the most one-dimensional player in the group, but he certainly had a big role on an injury-ravaged Magic team for the first few months of the season.

Robert Williams III (aka the Time Lord) simply hasn't played enough in Portland to be worthy of rostering, but this metric certainly underscores how impactful we know he can be as a scorer, passer, and defender. Jonas Valanciunas is stuck in a timeshare and coming off the bench, but he is still close to cracking the top 100 in just 20 minutes a night.

Goga has pretty much maxed out by getting to the top 60 and has benefitted from all the injuries in Orlando. Santi Aldama is having a fine year but is now hurt and is going to constantly be battling for minutes in Memphis when all the other big men on that squad are healthy.

The red group consists of players who we expect to rank higher based on their traditional stats. Ja Morant averages 21 points and nearly eight assists but is only playing 27 minutes a game and continues to deal with inefficient shooting and high turnovers.

Paolo Banchero was having a massive breakout before getting hurt, but is being held back by 64% shooting at the free throw line on nearly 12 attempts - that's almost Giannis-level bad - and a lack of defensive stats.

Ivica Zubac is having a career year, but his elite offensive numbers, FG%, and rebounding can only offset so much of the negative (no threes, 51% from the line, nearly two turnovers). He is impactful on the court, but if he can't crack the top 75 in his best season as a pro, then he's not likely to ever be a top fantasy player.

What I like about this list of top PIE performers is that it's made up of a lot of different types of players. It's not exclusively scorers or high-usage players. It's not dominated by just guards or big men, either.

As someone with a huge DFS background, I think this list makes perfect sense because these are the top fantasy points per minute leaders that we look for in our DFS contests. These days, I value efficiency more than ever, and PIE represents a different type of efficiency - the ability to accumulate stats on a per-minute basis.

If we can locate those high PIE players who simply need more minutes (the Time Lords and Jo-Vals), then perhaps that's one way to find value both within the season and from season to season.

Conclusion

The most fun part of doing research like this is if you're willing to keep an open mind, you're almost always going to learn something - even if it's not what you expected.

I am not sure that any of these four stats are predictive of top-50 performance, but a larger study that examined more players across more seasons (instead of just one partial season) would likely lead to the best results. But here are some takeaways for me anyway.

• Of the four stats analyzed, PIE was by far the most overrepresented in the top 50 group.
• AST/TO is a great stat for efficiency, but only a handful of players are well above league average as elite production is consolidated among those top-tier pass-first players.
• TS% is probably more valuable than EFG%, considering that it is a factor in free throw shooting, but plenty of volume scorers can still crack the top 50 without being efficient.
• Usage rate is still a great indicator of potential top 50 finishers (85% higher in the top 50 than the league average); however, PIE can help identify elite fantasy production among low-usage players. Examples: Josh Hart (14% usage, 12.4 PIE), Walker Kessler (12.6% usage, 12.6 PIE), and Jarrett Allen (14.7 usage, 14.2 PIE).

I hope you learned something here and perhaps begin to investigate some advanced NBA statistics on your own. I think I am probably just scratching the surface and this whole project makes me want to keep going with some different angles, testing out some other hypotheses.

NBA DFS News and Injury Alerts