# Nylon Calculus: Measuring extra scoring opportunities

Occasionally, an NBA broadcast will start their commentary with keys to the game, basically what a team must do in order to win. They’re often vague and obvious to the point of being unintentionally funny — crash the boards and take care of the ball. The reason is simple, to limit opposing scoring opportunities while maximizing your own. Stats like rebounding and turnover rates, as well as second-chance points and points off turnovers help show which teams excel in those areas of the game. But in this post, I created a couple metrics to account for all additional scoring opportunities and points from them.

It would be understandable to expect the first step for creating such a metric to involve offensive boards and turnovers. However, while my method was by no means perfect, I worried that particular way was a mistake. How big of a mistake depends on what related stats are considered. Offensive rebound and turnover percentages would be a simple way to calculate extra scoring opportunities, but rebounding percentages are calculated by all available individual ones while turnover ones are measured per 100 plays. They’re not meant to be used together.

The results from considering raw offensive rebounds and turnovers weren’t too far off from what I used. The issue here is that while team totals for each season include team turnovers, team rebounds are left out. That means missed shots that land out of bounds after a defender blocked it, or loose ball fouls drawn by a player on the shooter’s team would be left out because those situations need to credit a rebound, but it’s rarely to a player.

The way I measured extra scoring opportunities started with neither turnovers nor rebounds, but shot possessions. Again, it wasn’t a perfect route, but it avoided the team rebound concerns. The formula for net scoring opportunities per 100 possessions is below:

(Team Shot Possessions + Extra Free Throw Possessions) – (Opponent Shot Possessions + Extra Free Throw Possessions) / (Estimated Possessions) * 100

Estimated possessions were according to the formula used by Basketball-Reference, which tries to factor in team rebounds rather than leaving them out entirely like at NBA.com. Shot possessions were the sum of field goal attempts and free throw possessions. The latter statistic was explained in my last post, but it’s any two or three-shot foul where the last attempt can be rebounded by the opponent. That filters out fouls like flagrants and clear paths. Including turnovers in the formula threw the metric all out of whack, in case that idea was floating around.

What does net shot possessions look like on paper? Below were four examples, the best and worst from this season and the best and worst since 1997. I also included how they ranked in rebounding and turnover rates during their seasons.

Andre Drummond made a up a decent portion of the Pistons’ marks, leading the league in offensive and total rebounding rate this season, while multiple wings logged major minutes and averaged under two turnovers per 100 possessions. Detroit wasn’t as impressive as the 1998 Nets, but they were one of the best teams since 1997 by defensive rebounding and taking care of the ball, relative to league averages, ranking third and 14th out of 622 teams. As for the bottom in net scoring opportunities, the 2008 Suns and 2017 Nets got there after barely avoiding finishing last place in three different, essential categories. That was a common theme with the Seven Seconds or Less Suns, but ridiculously efficient scoring and never fouling balanced things out and then some. The Nets’ shooting was merely average, falling to the bottom of the lottery by limited scoring opportunities while giving up extra efficient ones.

But we’re still missing one more way to pile up points in not exactly organic ways. There are offensive rebounds and turnovers, but also trips to the free throw line left out of free throw possessions. The majority came from and-1s and technical fouls, but there’s always the rare flagrant, clear path, inbound, and away from play fouls. As you could guess from the formula earlier, I simply titled these extra free throw possessions.

Teams averaged two to five of them per 100 total possessions, but when compared with opposing totals, they’ll typically be cancelled out. Only five teams this season fell above or below a net of +0.5 or -0.5, no team since 1997 went above or below +1.5 or -1.5, and they didn’t alter the best and worst in net scoring opportunities of this season or the last couple decades. Still, they’re worth noting since these are indeed additional scoring chances.

Below is a dashboard looking at all net scoring opportunities before moving on to points:

Points were calculated by taking the team and opposing second chance points, points from turnovers, and made extra free throws. While made extra free throws were from play-by-play data, second chance points and points from turnovers are easily accessible, found at NBA.com. When combining all three statistics, I labeled these as net extra points.

When comparing net extra points with net scoring opportunities per 100 possessions, the best and worst teams this season remained the same with Detroit at +5.6 and Brooklyn at -7.0, but the rankings over the last 20 years changed quite a bit. This wasn’t surprising as no two teams score or defend the same off rebounds, turnovers, and technical free throws.

Below is a chart from Tableau comparing the net scoring opportunities and extra points per 100 possessions

The easiest way to explain the differences in the chart was by looking at turnovers, and how well teams scored and defended after they actually happened. That’s for multiple reasons. One was because they’re the juiciest of scoring opportunities — league averages were at their highest this season at 115.5 points per 100 possessions after turnovers. Another was while we can compare second-chance points, the absence of team rebounds on sites limits how they can be measured over several years. Accuracy on extra free throws is nice, but in most cases it wasn’t a huge deal.

So to get an idea of how large some differences were in turnovers, I touched on the 1998 Nets and 2008 Suns, which were the best and worst teams by net scoring opportunities with the 2008 Pistons and 2012 Bobcats, the best by net extra points. I also included some metrics to help explain the differences.

The Bobcats struggled by these and most other statistics from 2012, but it was 2013 that was the worst ever net rating off turnovers at -17.8. Phoenix’s efficiency off them covered their difference in turnover rates, except when playing Detroit. In their two matchups over the 2008 season, the Pistons went 2-0 and +19 and +21 in extra points. They were elite in transition defense, fifth-best after coughing the ball up since 1997 when adjusting for league averages, and 11th by net rating. That closed the gap in net scoring opportunities between them and the Nets, who forced 302 more turnovers than they committed, doubling Detroit. New Jersey’s net points off them per 100 possessions of +4.3, along with being about even in second chance scoring, were underwhelming compared to how many opportunities they gave themselves. By extra points per 100 possessions, the Nets finished 1998 third to Chicago and Boston, and 36th since 1997.

In the Tableau dashboard at the bottom of this post, you can find net scoring opportunities and net extra points, but I also included several other statistics to accommodate them. Some of them might seem minor. For example, did true shooting margin or the percentage of turnovers including steals really need to be included? They were, because while I think net scoring opportunities does a better job of showing the value of rebounding and taking care of the ball than other simple methods, it isn’t perfect and neither are second chance points and points off turnovers.

Below is why to take the two stats I put together with some caution:

## Rewarding or penalizing shooting

The formula for net scoring opportunities is supposed to value rebounding and turnovers (and extra free throws, sort of), but all the formula does is pit a team’s shot and total possessions versus their opponents. Made shots take away offensive rebound opportunities and forcing misses creates more for opponents. Because of this, it’s possible teams that shot exceptionally well or very often forced others into inefficient attempts were penalized, but it could also only apply to extreme situations.

The Golden State Warriors from 2016 was an example. They were +7.3 in true shooting percentage compared to their opponents, the highest mark since 1997. When looking at teams of the last two seasons, they were second-worst by net scoring opportunities. If we look at how they ranked in the two rebounding and turnover statistics, 2016 didn’t totally add up.

The 1998 and 1999 Spurs were a similar example, holding teams to the two lowest marks in true shooting, relative to league-averages, while third and tenth in net scoring opportunities. But in their first year with Tim Duncan, they ranked bottom five in both turnover metrics and their championship squad was 14th to 21st in both those and rebounding rates. Of course, they made up for it in Spursy ways, driving down true shooting marks by limiting 3-point attempts and not fouling.

Changing the formula to the differential in offensive rebounds and turnovers instead of shot possessions didn’t help the Warriors or Spurs much, the former team going to seventh over the last two seasons and the latter to fourth and eighth. Most good shooting teams or ones with great defense have leaks somewhere, maximizing an area of the game while punting others.

## Rebound Clusters

Offensive rebounds extend possessions, making it possible for more than one to happen at a time, and creating a cluster. While multiple offensive rebounds make a possession likely to end in a basket, it’s still one and a maximum of four points for all that effort. In the long run, having those offensive rebounds spread out across multiple possessions creates a higher ceiling on second chance points.

Rebound clusters can’t be controlled because every realistic opportunity should be pursued, but a team with less offensive rebounds in a box score could actually have more possessions including one, like 10 offensive rebounds over eight separate possessions versus nine over nine possessions. This makes it possible for a great offensive rebounding team to have their numbers padded, ranking highly in net scoring opportunities, but lower by net extra points and second chance points.

One team I suspect to have clustered, and as a result inflated their offensive rebounding was the 1998 Nets. Without adjusting for league averages, they have the highest offensive rebounding percentage since 1997, at 36.7. A lot of those went to Jayson Willliams, who led the league in individual rate at 20.5 and in percentage of Nets offensive boards he was responsible for when on the floor at 53 percent. But Williams was a poor finisher around the rim, shooting just 54 percent when league averages each year are in the low-60s, and he took a lot of shots from that area of the floor. The Nets in general struggled in the restricted area, making them a prime candidate to grab offensive rebounds, miss put-back attempts, grab their own board again, and hope to at least draw a foul.

However, the late-90s was an era of inefficient basketball. Even if the Nets clustered, it wouldn’t be surprising if most teams from the 90s did given the high offensive rebounding rates and low scoring efficiency.

There isn’t an efficient way to look at clusters anyway. They’re easily distinguishable in play-by-play data by covering half a page with events, but to log them over a full season would require something like exact possession totals and everything that happens in one to have a distinct possession number attached to it. Unintentionally, I think Senthil Natarajan’s breakdown of the best offensive rebounders provided some candidates of clustering, specifically when looking at players grabbing their own missed shots or those who had low points per play after offensive rebounds, like Greg Monroe and Jusuf Nurkic.

## Turnover quality

It’s possible a team looks bad by net extra points compared to net scoring opportunities because of clustered offensive rebounds. One thing that could also make results unusual are not just scoring and defending after turnovers, but the types of ones forced and committed. This is because while they’re great opportunities for the opposition to score, that’s only if they’re “live”, which I characterized as turnovers featuring steals. Dead-ball ones were bad passes or lost dribbles that went out of bounds, or time-related such as shot clock violations or three seconds in the lane.

Though my percentage of turnovers being live and dead-ball differed from Mike Beuoy’s (I just divided steals by all opposing turnovers, including shot clock violations) the most important part of his blog post on them was that teams score off of dead-ball ones even less efficiently than after made baskets, .986 points per possession versus .998, from 2011 to 2015. Meanwhile, live-ball turnovers averaged 1.19 points per possession. This makes sense given no transition opportunity after when the play is dead. At least on made baskets, there’s still a sliver of hope as the scoring team sets up their defense.

Like clustered offensive rebounds, the quality of a turnover doesn’t appear to be controlled other than by players known for steals, so some combo of length, speed, and green light to gamble. That makes sense for this season’s Warriors as they had 61.7 percent of their forced ones including steals, the most since 1997. Both Beuoy’s metrics (1.33 points per possession off live-ball turnovers) and mine (124.7 per 100 possessions off all) had the Warriors as the deadliest team off turnovers this season. When comparing each of our results over each season, some rankings were shuffled, but the clear best and worst teams were the same. The 2017 Warriors look great, the 2014 Clippers even better, and the 2006 Spurs were a chore to score on even in transition.

Next: Nylon Calculus -- In defense of radar charts

Below is a look at the percentage of turnovers being live and efficiency off them, though I adjusted for league averages since both of those stats have changed as the NBA went through different eras. Offense was kind of a jumble, but the Spurs clearly stuck out on defense.

In general, turnover quality should be something to be aware of, but despite the difference in efficiency in dead-ball and live-ball, points off turnovers should also remain valuable as a statistic. Forcing any should be good since the opposition is guaranteed zero points and provides an opportunity to string together scoring runs. But if extra points or possessions looks off when compared to the other, this is something to remember.

Overall, great shooting can cover up flaws in rebounding and taking care of the ball. For those who score or defend in scrappier ways, hopefully it shows in extra scoring opportunities and points.