Can a rookie survive a tank?

Did the Process leave any survivors?

Photo credit: Keith Allison from Owings Mills, USA [CC BY-SA 2.0 (]

Tune in to a Suns/Knicks game in 2019 and you’re excused if you don’t recognize the game in front of you. In the modern NBA, tanking, or losing “intentionally”, can become a viable franchise plan for future building. Popularized by former 76ers GM Sam Hinkie, tanking has become pop philosophy itself; a legion of rabid Philly fans point to Ben Simmons and Joel Embiid as proof that God is real and that, while maybe not water into wine, turning Michael Carter-Williams into anything useful is just miraculous. Losing on purpose allows a team to maximize their odds at a top pick. These GMs can excuse putrid on-court performance for the payout later on, arguing that sometimes the only way to the top is digging all the way through rock bottom.

Imagine any other industry adopting this approach. A restaurant could only get the top culinary talent out of college by flunking a health inspection and giving every patron food poisoning. A law firm has to lose 85% of their cases to finally snag that Harvard grad with a ruthless pedigree and a vast capacity for mind-numbing case work. For the NBA, the teams need to sell some hope in the form of young player development: sure, we might lose 70 times, but in those teaching moments, our 19 year old rookie could practice being a player without any stakes attached.

Yet, in these lost seasons, can a rookie actually flourish? If a tree falls in a forest, and the only veteran presence around to hear it is Chris Andersen, can that young phenom still find success? Let’s investigate.


We’re looking at our dataset of every rookie in the NBA since 2000, but for our analysis, we’re going to fine-tune the pool a bit. Only the rookies on the league’s worst team will be eligible for analysis. In isolating these players, we can try and measure the tank’s impact on their development. A reminder on scoring: I’ve pulled several impactful stats and scored each player by their z-score, or count of standard deviations from the mean for that stat. For example, Chris Paul has a z-score of 4 for his assist averages in his rookie season, meaning he’s four full standard deviations above the rest of the rookies. All of these z-scores summed together generates one composite score that helps rank our rookies across positions. Data is thanks to the kind folks at Basketball Reference.

Overall Picture

Things aren’t too rosy for our collection of rookies on the league’s worst teams. We’re working with forty-two eligible rookies since 2000, and, given our scoring system, a zero would be the exact league average for a rookie–for context, Larry Nance Jr. and T.J. McConnell are the two rookies nearest to a perfectly average zero. Let’s first show the distribution of values, where the largest bar will indicate where most rookies end up, and compare that to the histogram of just our tank commanders.

The majority of rookies are average to below average, with their negative scores weighed down by the superlative seasons of the rookies to the right of our graph. In fact, the average score in our dataset sits ever so slightly below zero. You’ll see in their distribution of our tanking rookies that they’re skewed a bit more to the left, with their average score at --2.4. They’re heavily congregated toward the negative end of the spectrum.

Is this stat sig?

However, before claiming omnipotence, let’s try and run some statistical analysis on my theory through hypothesis testing. Everyone tosses around the phrase statistically significant these days, but, if you’re like me, that often just means you double-checked your averages instead of the actual mathematics required to prove your theories. Resources like this great article on Minitab and this informative walkthrough on Analytics Vidhya explain the value and the methodology for proper hypothesis testing, but the brief summation is that this technique always you to vet what level confidence you should have in your conclusion.

For my theory that rookies on tanking teams struggle to do well, my hypothesis is that the worst team will produce poor rookies; the null hypothesis is that losing has no effect on their performance. I am using a target p-value, essentially the significance level, of 0.05, which will give us a threshold to say that 95% of our the time, our results are based on my theory (and not randomly occurring).

To calculate the p-value of my hypothesis, I’ll need to define these variables. The score we’re measuring is -2.4, as we want to see if the average tanking rookie is worse off than the rest of their counterparts. In the overall dataset, the mean is right around zero, with a standard deviation of 7.3. You’ll see the formula broken out here, as we snag the z-score for this value and produce -0.328: our tanking rookies barely differ from the typical ones, under half a standard deviation worse. Applying our p-value calculation, helpfully built in a calculator from a Georgetown professor here, we get an abysmal p-value of 0.25, far above our p-value cutoff and an indicator that our theory does not hold up statistically.

Being a rookie on the worst team in the league does not necessarily mean you’re doomed for failure, but it certainly doesn’t help. I’ve found some fun examples to showcase the variety of potential outcomes. Within this group, we can identify some fun outliers–there are some roses that grow from the fractured concrete–and doomed casualties of the losing debacle.

Unaffected by the chaos

For these three rookies, the turmoil of a losing season did little to stop their growth and success. Our top three scores are Nene Hilario (a 17-65 Nuggets team in 2003), Carlos Boozer (a 17-65 Cavs team also in 2003), and Gilbert Arenas (a 21-61 Warriors team in 2002). Boozer and Nene ended the season sixth and seventh in rookie of the year voting, per Basketball Reference, and were named to the first and second All-Rookie team for that season. The Arenas scoring is a bit of an exciting surprise, however. Arenas, a second round pick and unheralded rookie, did not place in the rookie of the year rankings or in the all-rookie team.

Yet, his total statistical profile tells the story of an efficient scorer contributing meaningfully to his team’s bottom line. His leap to stardom would happen in year two, jumping up to over 18 points per game and winning the Most Improved Player in his second season. Our rookie modellng shows his potential, and I’m excited to see some of the predictive potential that we’ll cover in a future post.

Run over by the tank

For every Nene, there’s another Denver rookie failing to find their footing in the league. Our bottom three guys are Junior Harrington, (Nene’s teammate in 2003), Jamal Crawford (a 15-67 Bulls dumpster fire), and Nikoloz Tskitishvili, also part of the abysmal Denver team in 2003. Tskitishvili was a top five pick, an infamous bust, and unfortunately one of the worst rookies in my entire dataset. His graph points almost entirely to the left.

I don’t want to be rude but the man had a 37.4% true shooting rate in his rookie year, 3 full standard deviations below his colleagues.

Poor Crawford. He struggled here but still managed to carve out a career as a hyper-enteraining, score-first combo guard with unbelievable handles. Show me a more devastating fake than the move he pulled on Kirk Hinrich.

Special call-outs

After gathering all of these players, I did find three guys that I wanted to call out specifically. Jahlil Okafor ended his rookie season with a score of 4.7, incorrectly indicating that he was set for a contributing role on a good team once the Process finished processing. Instead, his empty stats hid defensive liabilities that nearly ended his career.

If we’re stretching the tank imagery to the absolute maximum, I’d posit that Kemba Walker took down an aircraft carrier worth of weaponry. His 7-59 Bobcats team could not be more pitiful, and yet poor Kemba cobbled together a 2.5 season score. As for his teammate in despair, Bismack Biyombo ended at nearly -6 on the year.

My all-time favorite find? Look no further than the 2013-14 Milwaukee Bucks, owners of a 15-67 record. Their first round pick Giannis Antetokounmpo scored -2.33 in our rookie rating system. I clearly couldn’t factor Greak Freakiness into my methodology, but it’s pretty fascinating to see the early stages of his game flash in his rookie season. Giannis was perhaps too deferential on offense, with a ridiculously low usage rate, but his ability to block, steal, and rebound while drawing fouls consistently hint at the gamebreaking insanity that he’s turned into today.


For fellow fans of tank commanders like the Knicks, hope isn’t all lost. Some rookies in bad situations thrived, some never found their footing, and others followed a bit more circuitous route to competency. In future posts, I’ll look at how the 2018-19 class is faring so far, with a special shout out to my boys Kevin Knox, Mitchell Robinson, and Allonzo Trier.

Check out the list of all eligible rookies here.