The Advanced Stats of Turnovers by Adam Spinella
This article was first featured on Spinella’s Box and 1 Blog.
The little things matter in basketball, and the NBA is no exception to that rule despite common conception. In a league where most teams have talent that can score the ball, the teams that win the most games and have the best offenses maximize their possessions. That can mean a lot of things: getting shots from the right players or places on the floor, winning offensive rebounding battle, having great sets in special situations (end of clock, baseline and sideline out of bounds), or winning the turnover battle.Turnovers are a huge part of the NBA game. Whether based off great pressure from defense or poor execution by the offense, the turnover battle can be very indicative of the result in an NBA game. Additionally, we’ve seen that to maximize possessions and get higher-percentage shots, teams that move the ball well and have higher assist numbers generally have a greater offensive output. We can see within each system what teams do a great job at this: low team turnover rates or assist numbers are indicative of success.
What about on the individual level? Is there a way to quantify a player’s “safeness” with the ball for the role that they serve in? Can we examine it across offensive systems without being dependent on teammates or coaching? Probably not, but knowing what to look for when analyzing a player’s performance is key. I’ll walk you through the process I used to see how strong a player really is with the ball. These numbers are from the 2013-2014 season.
Turnover percentage is a stat that quantifies the turnovers that a player has per 100 possessions. The stat is really only a real indicator for players with a large enough sample size (players on pace for around 1000 minutes played in an entire season). But it goes a long way to showing how responsible they are with the ball and how trusted they can be, particularly in crunch time. Here’s a breakdown of stats from last season with the top five at each position with at least 1000 minutes played.
The table shows that the players that are more catch-and-shoot players at the shooting guard, small forward and power forward positions are not very turnover prone. The players who handle the ball more at those positions, like Boris Diaw, Andre Iguodala and Lance Stephenson, have higher turnover rates – it’s just what you would expect. But there are a couple of aberrations, particularly for players who play such meaningful minutes. To see Dirk and LaMarcus Aldridge in the top five is incredibly encouraging; Tobias Harris and Nick Young both handle the ball quite a bit on the perimeter and don’t have high turnover rates.
The most meaningful data collected here was at the center and point guard positions. At center, we see that Kendrick Perkins turns it over four times more frequently than Al Jefferson despite getting such fewer opportunities to touch the ball. Among point guards, Patty Mills is astoundingly ahead of the pack at 8.2 Point guards have overall higher numbers than all other positions due to the frequency at which they handle the ball. Lillard and Conley, two starters, are on this list despite high usage. Kemba Walker as just outside the top five at 11.6. On the flip side, Ricky Rubio is nearly twice that total with a 21.8 turnover rate. For a player who doesn’t shoot a lot, the turnovers likely are to be higher, but this seems really high for a player that just received a four-year extension.
It’s very telling that the Thunder have three frontcourt players with over a 20% turnover rate: Perkins, Adams and Collison. On the other end of the spectrum is Portland, with three starters in the top five at their position in Lillard, Lopez and Aldridge. It makes up for Nicolas Batum’s fairly high 18.6 turnover rate. The only other team with two starters in the top five at their position was Minnesota with Pekovic and Martin. Turnover statistics, beyond just sheer turnovers and A:TO ratio, are important for frontcourts as well as guards.
But turnovers, by and large, are indicative of guard play. In seeking to balance the TO% numbers with ball-handling metrics, I combined it with ASST% stats, which calculate the percentage of teammates’ field goals that the player assisted on. By dividing the turnover percentage per 100 possessions into the percentage of assists, I got this stat: assist to turnover usage ratio. It focuses mainly on those who handle the ball frequently and have high assist numbers. The question seeking to be answered here: do these players create more buckets for their teammates than they cost them by turning the ball over?
Alright, Chris Paul is absolutely ridiculous. If you had any doubt that he is the best point guard in the NBA, this should wipe that away. He’s a full point and a third ahead of the next-best qualifier, Mike Conley. Only six players are at 2.5 or above, and only Westbrook and Irving had a usage rate above 25% (meaning they took a high volume of their teams shots). John Wall and Stephen Curry, two very different types of players, have a similar impact.
The three players here that caught my eye were LeBron James, Kevin Durant and James Harden. Look where the past two MVP’s are: 2.22 and 2.19, ahead of Damian Lillard, who was in the bottom of the league’s turnover rate. But James Harden is below 2.0, meaning that for every two baskets he assists on, he’ll cost his team more than one possession. When I see that, combined with his defensive woes, I no longer think of Harden as an efficiency or statistics guru’s dream. He may be a great scorer because of where his shots come from on the floor, but I’m not drawn to him overall to be a positive contributor.
What about those point guards who are eligible for an extension? Kemba Walker at 2.56, Brandon Knight at 2.02 and Ricky Rubio at 1.73. Walker has an incredibly impressive rate, highlighted by a severe cutdown on turnovers and a very safe system under new coach Steve Clifford. Knight is just at the league going rate for a point guard, so it’s not surprising that he didn’t get a suspension. But Rubio… Rubio is still heralded as a great one. Rajon Rondo, another pass-first point guard who was omitted from this list because he missed last season due to injury, has a much greater ASTrate:TOV% statistic than Rubio. Here’s the comparison from the last three years.
In order to get to the next level as a credible pass-first point guard, Rubio will have to get above 45% on his assist percentage. Rondo’s lowest rate in the last three seasons came without Pierce, Allen or Garnett – and he cut down on his turnovers that season as well. Rubio actually increased his turnovers and decreased his assist rate last season, despite seeing Kevin Love play 59 more games in ’13-’14 than in ’12-’13.
Two other factors need to be accounted for in this debate. First is the shooting percentage of the team. Quite simply (and factually), teams that make more of their shots will yield players with higher assist rates. Who is to say that Rubio didn’t set up his teammates just as well as Rondo and his teammates simply missed those opportunities? Well, in the Rubio and Rondo debate it makes matters even worse for Rubio. Rondo’s Celtics were 29th in the league in field goal percentage; the Wolves were 23rd.
But it raises a strong debate. The league average field goal percentage in 2013-2014 was 45.4%. Miami was at above 50% as a team, while the lowest two teams were only two percentage-points below the average. Only two teams were above 48 percent: San Antonio and Miami, and only thirteen teams above the league average. For the most part, I’ve found that with the exception of the few teams that are incredibly high in the league in terms of shooting, the differences in field goal percentage of the team is not enough to skew the data of a point guard’s impact.
The other factor that needs to be accounted for is the team’s system of sharing the ball. Some teams and offensive systems rely on a guard to distribute the ball to others. So I added on a new column to the ASSTrate/TOV% graph with that player’s percent of their team’s assists per 48 minutes.
This table tries to lay the responsibility within each team’s system to create assists for others. Some players’ numbers will be high if they’re the only one expected to create for others. For example, the four lowest on the list are Parker, Chalmers, Lin and Iguodala. Chalmers, Lin and Iguodala have someone else on this list with a higher percentage, meaning they aren’t the main creator on their teams. Parker’s numbers are indicative of the extra passing system the Spurs run – assists aren’t expected to come just from Parker, they come from the offense.
The players highest in this list have the ball in their hands the most on their team: Chris Paul, John Wall and Stephen Curry dominated possessions frequently with the ball in their hands. It speaks both to the coaching system and the lack of other playmakers on the team, which leads to how that system is built. Some teams may have an incredibly high amount of their assist burden concentrated on two players. For example, Durant and Westbrook combined for 56.8% of their team’s assists per 48 minutes last season, meaning that while their turnover rates may be high, the system is designed around them having the ball, so those turnovers are at least coming within the confines of where the usage should be. Curry and Iguodala combined for 54.5%, the next highest teammate tandem on the list. The highest tandem in the league: Phoenix’s Eric Bledsoe and Goran Dragic at 59.5%.
But the biggest value that the % of team assists statistic shows in this table comes in comparing it with the ASSTrate/TOV% value. When the A:TO rate is high and the % of team assists is low, it means that a player is exceptional at creating for others and maximizing each possession. Tony Parker is the prime candidate here: he assists on 31.7% of buckets while he’s on the floor despite not being a prime creator in percentage of the offense, but he takes care of the ball while doing so. Damian Lillard and Mike Conley also fall into this category, mainly because they take care of the ball. Their assist totals aren’t as indicative of success because the offense gets assists from elsewhere.
Conversely, a player with a low A:TO rate and a high % of teams assists gets gaudy numbers in terms of assists because their system relies on them, not because they make tons of great plays. The numbers are driven by two things: high turnover rates (like Ricky Rubio and Kendall Marshall) or domination of the ball within their offense (like Michael Carter-Williams and James Harden).
A player with high outputs in each column is heavily leaned upon in their offense and do pretty well within it. Obviously Chris Paul is the leader here; Kyrie Irving and Kyle Lowry get high marks here as well. A player with low outputs in both columns means they likely aren’t the primary option within their offense as a creator and their turnovers are to the detriment of their team more than their assists are a positive. This includes players like Jeremy Lin and Mario Chalmers: Deron Williams may even fall into this category.
So what the hell does this all mean? It means that we need to look at more than an assist to turnover ratio to see the true value of a point guard. Figure out how much they’re being asked to do, both in terms of scoring and creating for others. Then see, based on that usage, how they protect the ball. Hopefully stats like these can serve as indicators as for who does the most good for their team.