Knicks fans, are you tired of the boring, vanilla assist-to-turnover ratio? Does it bother you that it equates the cost of turnovers to the benefits of assists? Have you ever thought to yourself, “I wish there was a statistic that properly captured the rewards and penalties of assists and turnovers have on the outcome of a game? If your answer is “what weirdo actually thinks about these questions?” I in no way can blame you. With that said, this weirdo did wonder these questions and I did, in fact, play a part in creating such a statistic.
Reintroducing hPPR, a statistic that is a mathematically more grounded improvement upon John Hollinger’s PPR. I co-created the statistic back in late November when I contributed at Holyfield. You can find the entire methodology here; however, I am going to provide everyone a quick-and-dirty summary of the process as I did with my zone true shooting metric. First, let’s look at Hollinger’s PPR formula:
“Where does the two-thirds (0.667) come from?” That’s an excellent question. The answer: for all intents and purposes, it’s arbitrary. In order to determine the actual coefficient/weight for the assist and turnover variables in play, a robust OLS regression (like zTS%) was performed. The regression was modeled after Kevin Ferrigan’s DRE metric, but with three distinct alterations:
- We used single-season RAPM over 14 years instead of 14-year RAPM. The reason was to account year-over-year fluctuations in assist and turnover totals.
- We broke out the field goal variables in Ferrigan’s model into two-point and three-point variables to distinctly account for the differing impacts two-point and three-point shooting respectively have on winning.
- We broke out the total rebounds variable in Ferrigan’s model into offensive and defensive rebounds for the same reason as the previous point.
The coefficients for the assists and turnovers variables are 0.15984 and the -0.26318, respectively. These results were then normalized — dividing both coefficients by the original assists coefficient from the regression results (0.15984 ÷ 0.15984 and -0.26318 ÷ 0.15984) — which gives a turnover coefficient value of 1.65 (took the absolute value due to Hollinger’s formula already subtracting turnovers). Below is the hPPR formula:
Though the change may seem subtle in that Hollinger’s two-thirds figure (0.667) isn’t that far off from our normalized assist figure (0.60734), the difference is significant enough to affect the results. Furthermore, striving for accuracy and testing the results of a study is, you know, a foundational element of scientific principles. But let me stop myself from hopping on that soapbox and give you guys what you really want: the results. Below are the hPPR figures for the New York Knicks in 2017–18:
If you visit the Tableau Dashboard, you can view historical hPPR figures for this decade, which includes some elite hPPR figures from Knicks legend Priggy Smalls.
This may come as a shock to some, but New York had two point guards in the 95th percentile for hPPR. Trey Burke’s 12.55 figure and Jarrett Jack’s 10.77 figure ranked fourth and ninth in the entire NBA, respectively. It may also come as a surprise that Emmanuel Mudiay during his time in New York had a better hPPR figure than Frank Ntilikina. What shouldn’t be a surprise is how bad Kristap Porzingis’ figure is given his inability to pass out of double teams while posting up.
But what do these figures mean? What does it mean when Burke and Jack are in the 95th percentile in hPPR? How do we properly contextualize a player’s hPPR figure? This is how my colleagues and I wrote in the original article:
More than anything else, the best way to use hPPR is to determine what players run their offense at minimal cost to offensive possessions; it does not determine the skill level of a passer as much as it indicates their ability to make a smart pass.
Looking at the Knicks’ hPPR figures, that conclusion makes sense, right? I’d argue that Ntilikina is the Knicks “best, most skilled” passer, but he did not run the offense as efficiently or efficiently as either Trey Burke or Jarrett Jack in terms of cutting down on turnovers. There were many times where French Sinatra would telegraph passes out the pick-and-roll or have passing lanes jumped because the defender knew that he wouldn’t shoot. Burke and Jack, on the contrary, not only knew when to pass but also probed the defense with veteran-like savviness in order to get a better look for a teammate. I believe the French Prince will improve his hPPR as his career progresses and he learns the nuances of NBA point-guard play; this is in no way a slight, but rather one assessment of his rookie season.
Ron Baker being third on the team isn’t surprising given how he rarely turns the ball over and generally makes the smart pass. What is a bit surprising, though, is Emmanuel Mudiay being fourth on the Knicks during his stint and ahead of Ntilikina. Mudiay improved his hPPR during his time in Denver from 0.29 (60th percentile) up to 3.5 (80th percentile) to close out the season. If there is one
and only positive to Mudiay’s season, it would be his passing/cutting down on turnovers.
As alluded to earlier, Porzingis must improve as a passer, and hPPR really brings to light this glaring weakness in his game. Porzingis’ appalling -6.18, 5th percentile figure is alarming, especially when you look at the raw assist and turnover figures. The lack of passing vision is something Ben Falk detailed in his January piece on The Unicorn back in December:
Here’s an even crazier stat. We can look for games that seem to be on the extreme of shooting without passing: games with 25 or more shot attempts and 1 or 0 assists. Let’s call this a “Carmelo”, for reasons that will become clear in a minute. Porzingis leads the league this year in Carmelos, with 7 in just 31 games played. The league leader last year was Anthony Davis, who had 11 Carmelos in 75 games played.
Porzingis is never going to be a playmaker like an Al Horford or a Nikola Jokic — first and second in hPPR for centers, respectively, as well as in the 85th percentile — but if he could at least get to a point where he has more assists to turnovers, he should be fine. Hell, both Anthony Davis and Carmelo Anthony had more assists than turnovers in 2017–18, and their respective 20th and 25th percentile hPPR figures would be a considerable improvement.
To wrap up, like any statistic, hPPR should in no way be used as a “tell-all” passing statistic that measures “best playmakers” or “best passers” (though players like Chris Paul, Rajon Rondo, and Rick Rubio do happen to make many appearances at the top of the historical list). This is a passing statistic to help measure “who is running the offense well/efficiently by not committing many turnovers.” Trey Burke and Jarrett Jack did just that for New York this past season. With Jack probably not coming back, let’s see if Burke continues efficiently running the offense next season as well as see if Ntilikina can cut down on his turnovers and if Mudiay’s improvement is real.