Introducing Arena-Adjusted Assists

by Blake Murphy 6. January 2013 16:17

When Jose Calderon was recently credit for an assist on this play (the first in the video), it had some questioning the validity of assists based on favorable home-court scoring. There’s no way, in any sense of the definition, that should have been scored an assist for Calderon.

The definition, per the NBA’s statistics manual according to this Wall Street Journal article, is as follows:

The NBA statistician's manual says an assist should be "credited to a player tossing the last pass leading directly to a field goal, only if the player scoring the goal responds by demonstrating immediate reaction to the basket."

Really, the validity question should come up for a handful of other reasons with respect to assists, but the idea of a home-advantage is an interesting one, and one that Ken Pomeroy had previously tried to tackle", albeit it was about college basketball rather than the NBA.

Question: Do scorers favor home teams when giving out assists?

According to one former Grizzlies stat-hand, it’s a lot of subjectivity and there’s room for bias from the scorer’s position. That said, that same WSJ article linked earlier had comment from the NBA that all stats are reviewed, so perhaps this is a problem that has been ironed out since the late-‘90s.

Nonetheless, the topic got me curious. In a discussion with my Beyond the Boxscore colleague Bryan Grosnick, he mentioned that he had actually pulled data on this matter before. He was kind enough to send it to me and share his findings with me.

I repeat: All of the credit for the research and pulling the stats belongs to Bryan. I greatly appreciate him sharing it with me to communicate via Hoopdata.

Bryan’s methodology for answering the question was as follows:

“I found a way to quantify it out, by comparing road and home assists to field goals made for each team, during each season from '06-'07 to '11-’12.”

That is, it compares a team’s at-home A% with its road A%. (Note that this method, using Assist Percentage, should strip out factors like pace and teams playing better at home, since it’s just A/FG. Not perfect, but it’s a start.)


Question: Do scorers favor home teams when giving out assists?

It turns out that yes, scorers tend to favor the home team when it comes to giving out assists.

Answer: Scorers give home teams a boost of 2.7 percentage points, which makes for a 4.9% boost in Assist Percentage.

In other words, the Assist Percentage for all road teams in the sample was 56.1%, while for home teams it was 58.8% (a 2.7 percentage-point or 4.9% increase).

The data that Bryan gave me contains every team season in the sample, so there is a lot of room for me to build on this initial article with further analysis. My plan over the next few weeks is to re-create the assist leaderboards in each season using what I’m roughly calling an “Arena-Adjusted Assist” metric (AAA). Those types of articles should flow from this finding and this data set fairly easily, especially given the wide disparity between teams in particular seasons.

Some Interesting Findings *The top-five most favorable home scorers belong to: Nuggets, Clippers, Hawks, Lakers, Cavaliers *The five least favorable home scorers belong to: Heat, Suns, Kings, Knicks, Grizzlies *The single highest-inflated season was the 2008-09 Nuggets with a 13% inflation *The single most-deflated season was the 2006-07 Heat with a -8.7% deflation *LeBron’s quest to average a triple-double is actually hampered by Heat scorers *Steve Nash was probably undervalued in his MVP years

Please let me know via comment/email/twitter if you have any ideas for follow-up pieces based on this information. As mentioned, I plan to do a handful of articles based off of this data set in the next couple of weeks, and suggestions are always welcome. So stay tuned for more.

And a HUGE thanks to Bryan for all his hard work.

Follow Blake on Twitter.

Bookmark and Share DotnetKicks dotnetshoutout


Greg Monroe's Emergence as a Passer

by Blake Murphy 20. December 2012 22:15

When it comes to passing big men, it’s usually the Gasol brothers who get most of the love. After all, they rank second and third among players classified as forward, center, or forward/center by Basketball Reference in assists per game this year. They both also have a strong track record and reputation for being adept as passing, especially from the elbow for high-low feeds.

Joakim Noah is in first with 4.5 per game, the first time he’s eclipsed the rough “three per game is a good passing big man” plateau. Noah has been getting love all season for his phenomenal play, so I won’t expand except to say that, yes, he’s really improved and is a treat to watch and own in fantasy.

It’s the fourth name on the list that I want to talk about, because I think Greg Monroe may be the league’s best passer out of the low post. His 3.4 assists per game are solid and, without digging deeper, show a passing skill most big men don’t have. However, unlike the other names around him on the leader board, Monroe isn’t really an inside-outside big man in terms of the shots he takes.

To wit, look at his Hoopdata page. He actually leads the entire league in field goal attempts per game at the rim with 7.1 and also takes another 2.6 from three- to nine-feet out. In total, he takes more shots inside of 10 feet than any other player in the NBA except for Dwight Howard, Brook Lopez and Nikola Pekovic, and they’re all razor-close.

Monroe proceeds to take just 2.5 shots per game outside of 10 feet, meaning that roughly three quarters of his touches come from in the paint or short out on the baseline. So he’s spending a good amount of time down there.

Of his 3.4 assists, 1.9 of them lead to baskets at the rim. These could be high-low feeds, quick dishes under the basket or passes to back-door cutters when he is facing up. Nearly one per game lead to threes, as well, which are likely passes out of the post and/or passes out of double teams.

It’s difficult, given the current statistics available, to tell where assists come from and which type lead to shots from where. It’s possible that Monroe simply gets a lot of assists when he’s outside the paint and then takes his shots inside, but that doesn’t seem to fly with conventional wisdom and what comparable players do.

For example, Noah’s 4.3 assists are spread more normally across the floor in percentage terms than are Monroe’s, which are almost exclusively at the rim or threes, the best “type” of shots a player can get for their teammates.

One final note is that, as a team, the Pistons shoot the sixth worst percentage in the league on shots at the rim (although that is including Monroe, who is a well below-average finisher himself). They are also roughly a league-average team at hitting threes. So it seems it’s possible Monroe is missing out on assists at the rim that an average team may hit, though it’s difficult to tell exactly.

However, just for confirmation that what I’ve witnessed in a handful of games and see in the stats is backed up by a “Pistons regular,” I asked Dan Feldman of Piston Powered to provide his thoughts, which ran counter to my thinking:

“Monroe works from the high post more than his shot selection would indicate. He's been a hesitant mid-range shooter lately, which has frustrated many fans, so most of his scoring comes in the low post. But most of his assists have come from the high post, from where he hits cutting teammates. I suspect his passing is similar to Noah's and Marc Gasol's.

Has Monroe evolved as a low-post passer? Yes. But was a very good passer at Georgetown, too. As a rookie, the Pistons didn't ask him to do it all. Last year, they did a little. This year, more than he's improved, I just think they're finally letting him loose.”

As Monroe continues to work to refine his game in his third season, some may look at the drop in PER and the slight dip in rebounding and suggest he’s faltered or stagnated. But in at least this one area, Monroe has continued to show growth, or at least unleash a skill that already existed below the surface.

Follow Blake on Twitter.

Bookmark and Share DotnetKicks dotnetshoutout

DeAndre Jordan and the Crunch Time Conundrum

by Blake Murphy 26. November 2012 19:01

DeAndre Jordan is a worse free throw shooter than some players are three-point shooters.

With a 44% career free throw rate, Jordan is a huge liability late in close games, enough so that coach Vinny del Negro has to be wary of playing him in such situations.

In fact, Jordan averages 26.5 minutes per game but has played just 4.5 minutes per fourth quarter, well below what you’d expect for a starter and one of a team’s core players. He’s been protected on the offensive end such that he’s actually yet to take a free throw in the fourth quarter, and has only been afforded 0.5 field goal attempts per fourth quarter. Basically, when Jordan plays in the fourth, he is there for defense only.

I was interested to see if Jordan (a) is in a unique situation with limited fourth quarter minutes and (b) should be a better free throw shooter given his shooting rates elsewhere.

Is this unique?

For the first point, I used’s Advanced Stats Tool to find players in “crunch time” (last five minutes, ahead or behind five points) who had the lowest usage rate.

We are, of course, dealing with very small samples at this point in the year, so Jordan is one of many players with a miniscule usage rate in small crunch time minutes. He’s certainly not unique in this regard, but it warrants further and more detailed study controlling for FT%, position, etc.

Should he be better?

For the second part, I had planned to compare Jordan’s shooting percentages from different distances and come up with a sort of “expected FT%.” However, Jordan shoots so exclusively inside of five feet that the exercise would have been pointless.

Just how limited is Jordan’s range? Well, he’s taken just 23 shots outside of five feet this season, and just five outside of nine feet. He’s a 65% shooter within five feet and shot 66% from there last year (when he took just 32 shots outside of five feet and just 10 outside of nine feet). It makes sense for him to stay there since he’s so effective, and the samples are too small to know if he’d be any good from elsewhere (anecdotally, I doubt he would).

Are they right to bench him?

So the question for the Clippers becomes whether Jordan’s defense (and other offensive tools, such as screen settings, providing the threat of post offense, etc) is worth enough to keep him on the floor despite his free throw shooting.

The Clippers are actually better offensively with Jordan on the court, by 5.5 points per 100 possessions, but this is almost certainly due to him spending all of his time with strong teammates. Defensively, the team rebounds and blocks slightly better with Jordan on the floor but allows a significantly higher eFG%. The net effect is that the Clippers have been 7.1 points per 100 possessions worse on defense with Jordan off the floor.

Overall, the Clippers have been +4.3 points per 100 possessions with Jordan on and +5.9 with him off. He’s a slight net loss if we don’t control for his teammates, which is nearly impossible since he’s played 332 of his 344 minutes with Chris Paul.

Basically, you can’t fault Vinny for gluing him to the bench when the game is on the line. He hasn’t proven enough that he’s a better option than Turiaf, Hollins, or a smaller lineup, and until he can more reliably hit the freebies or more definitively prove himself a defensive presence, there’s not enough evidence to cry foul on del Negro.

And even if you could cry foul, Jordan would likely miss the free throws.

Bookmark and Share DotnetKicks dotnetshoutout

Defense and the "Fifth Factor"

by Blake Murphy 25. November 2012 14:13

On a recent Raptors Republic podcast, a show I host each week, one of the panellists suggested that he thought the Raptors’ decline in defense may have been due to them playing at a faster pace this season.

No, the analysis wasn’t based on a faulty mathematical basis – he was not mistaking, as many did during the D’Antoni-era in Phoenix, that a faster pace led to more points and more points were indicative of a bad defensive unit. Instead, he was suggesting that, even for an efficiency stat like defensive rating, pace could still have an impact.

It’s a fair point. Basically, if a team was successful slowing the game down and forcing teams to operate in the half court more, speeding up the game could expose things like poor transition defense, less capable individual defenders, and more.

So, taking his comment and running with it, I pulled some defensive data from Basketball Reference for the past few seasons, dating back to 2007-08. I was first interested to see if teams who played at a faster pace tended to have a worse Defensive Rating. The relationship between pace and defensive rating is shown in the graph below – the relationship had a correlation of 0.29 and an R2 of 9%. The p-value was insignificant, meaning that, even though the relationship isn’t strong, it still exists, and faster-paced teams do tend to be of poorer quality on defense.

I found this to be a bit interesting, as my initial assumption was that, while my panelist’s argument made sense intuitively, pace is normally not discussed as a factor in defensive rating. In fact, the “four factors” that are most commonly discussed are Opponent’s Effective FG% (eFG%), Turnover Percentage (TOV%), Defensive Rebound Rate (DRB%) and Free Throws per Field Goal Attempts (FT/FGA). The “Four Factors” of defense had individual correlations with Defensive Rating as follows.

You can see from the table that an opponent’s eFG% has by far the strongest correlation with Defensive Rating, which makes obvious sense. Adding pace to our “Four Factors” to give us a “Five Factors” model adds very little to our understanding of what makes a good defense, so it seems a waste.

However, is it possible that pace has an impact on eFG%? This would be helpful, since a team looking to improve its Defensive Rating can’t really just say “get more stops” to improve defense. Instead, if pace has an obvious impact on eFG%, teams can then look to slow things down to improve their defense, along with forcing more turnovers, fouling less, and doing a better job on the boards (essentially, I wanted to know if we can replace “get more stops” with a more practical suggestion for teams).

Like with pace and Defensive Rating, we see a somewhat weak but statistically significant relationship. Thus, we can safely suggest to coaches that slowing down the pace should have a small positive impact on defensive rating.

Thus, our “Four Factors” should now be: *Defensive Rebound Rate – crash your own boards harder, box out more, leak out less *Turnover Rate – force more turnovers *Free Throw Rate – send teams to the line less *eFG% - get more stops, which can be aided by slowing down the pace

Bookmark and Share DotnetKicks dotnetshoutout

The Rare Decline of Tyreke Evans

by Blake Murphy 21. November 2012 07:13

Tyreke Evans is putting himself among some unique company. In this case, I don’t mean that as a compliment.

Evans currently has the unique distinction of being one of a small group of players in the three-point era to have failed to match his rookie performance in his next three seasons.

That is, Evans looks to be one of a rare group of players to peak as a rookie.

To determine this, I used Basketball-Reference’s Player Season Finder to find all qualified rookies (in terms of minutes) in the three-point era. I then pulled the second, third and fourth seasons for each of these players to determine who else had failed to meet the bar they set as a rookie.

To avoid including great players who may have fluctuated around a strong level of play, I took out all players who had a Win Shares Per 48 (WS/48) of greater than 0.100 in subsequent seasons (this is a somewhat random cut off, as all cut offs are, but 0.100 roughly splits the league at 150 players). In essence, I eliminated any players who remained “starting quality” beyond their rookie years to find players who began as capable rookies but failed to remain effective.

For Evans, he appeared to be a superstar in the making as a rookie. He was just the fourth player ever to average 20-5-5 as a rookie, joining Michael Jordan, LeBron James and Oscar Robertson. He had an 18.2 Player Efficiency Rating and was worth 5.4 wins according to Basketball-Reference. In his sophomore year, though, he fell off a cliff, coming in at just 1.6 wins, a 14.4 PER, and declining efficiency across the board. He had a slight rebound to 2.7 wins and a 16.4 PER last year, though his rates didn’t match his rookie season. And so far this year, his WS/48 is nearly identical to his terrible sophomore year, with his PER at a career low 13.9.

So how rare is it for a player to appear a potential star as a rookie, only to become a hardly usable player in the three years thereafter?

I found eight such players, and the list isn’t pretty.

Sam Mitchell - a 26-year old rookie, Mitchell basically settled in as who he would be right away, though he would have small spikes in Year 6 and Year 9.

Manute Bol - had his highest WS/48 as a rookie and only remained an effective player for his first three seasons, primarily since his terrible offense negated much of his shot-blocking prowess.

Ron Mercer - using basic stats, he was an adequate player for some time, but advanced stats show that his usefulness was based almost entirely on volume. He would never post a league-average PER and his rookie season was by far his best in terms of total contributions (WS/48).

Courtney Lee - has remained on the periphery of usefulness thanks to some decent defensive acumen and floor spacing, but his rookie season was his best thus far. In his fifth year, the Celtics mark his fourth team and, while he’s off to a very rough start, there’s certainly evidence he can stick as an outside-shooting defensive rotation player.

Popeye Jones - had a great rookie season and didn’t drop off too much in subsequent years, though his fifth season (with the Raptors, of course) was an injury-shortened disaster. He bounced back afterwards and essentially produced at the same level in Years 1-4 and Years 6-9, so he probably doesn’t fit here.

Herb Williams - had a spike after the fourth year with a strong fifth and sixth season, but once again fell off until an odd resurgence in small-minute efficiency with the Knicks at age 34. If he were a guard, he’d be a nice glimmer of hope for Kings fans, but it’s not clear if a guard has the same potential for a bounce back.

Danny Vranes - great as a rookie, good as a sophomore, okay as a junior, and on his way out of the league shortly after. He fits the narrative of a rookie-peaker best, enjoying a steady decline thereafter.

Tyreke Evans - his fourth year isn’t done so maybe it’s not fair to include him yet. But a few notes on how Evans has changed since his rookie year, courtesy of his Hoopdata page:

*In Evans worst seasons, he’s shot more from 16-23 feet. Not a surprise here, as he is a terrible shooter from this range, hovering around 30%. It’s probably in his best interest to give up the long-two pull-up (less than 20% of these shots are assisted).

*Evans hovers around 60% at the rim but, not surprisingly, got there much less often in his two down years. This has also led to steadily declining free throw rates.

*He’s decided to try and get himself going more this year, posting a career-low assist rate of 17%. He has also used his penetrating ability to create for others less, assisting on only half as many threes as usual.

So not only is Evans in unique company as a player who peaked as a rookie, but he also appears to be trending in the wrong direction. This isn’t a matter of shots just not falling, but an apparently declining skill set and an eroding basketball IQ.

There’s certainly still time for Evans to turn things around, but teams should be wary of tendering him a sizable restricted free agent contract this summer unless he shows a turnaround between now and then.

Bookmark and Share DotnetKicks dotnetshoutout


Archive | FeedSubscribe | Log in