Oh, sweet NBA youngling, will thee maketh the “Jump?” Part 2

“What’s done cannot be undone” — An unexpectedly long and detailed guide examining the metrics which might shine a light on the future of your favorite young NBA players.

In Part One, we looked at the success rate of NBA Rookies of the Year to go on to become All-Stars (or better) during their career. Between1980–1999, the success rate was 85%. From 2000–2018 the success rate dropped to 67%, perhaps due to the influx of one and done college players.

“Double doubles, toil and trouble” —What better way to liven up a dead article on statistics than to add quotes from MacBeth? (definitely a Celtics fan)

In Part Two, we take a look at the best statistics to use to evaluate player performances. Because so much of this article are lists of players with specific metrics to see if a stat is useful or not, it’s even tough to insert a few jokes and bad puns (except, of course my 2019 MVP winner — something wicked this way comes). There’s a lot of interesting stuff for stat geeks, but I will understand if many of you die of boredom (alas, poor Gutbloom, I knew him well).

And you would be forgiven, should you skip to the TL;DR at the end.

First, I wanted to examine the Third Year Player Rankings and the idea that this is the key season when most players make the jump.

Third Year Player Rankings

Check out Brandon Anderson’s article on the best third year players.

This group is clustered together more tightly than any group we’ve observed so far — too young to reveal a Tier 1 player, but too good to include the bottom two tiers.

Tier 2 (All-NBA, multiple All-Star, top 15 NBA-Player): Joel Embiid

Tier 3 (All-Star, top 30 NBA Player): None (but Siakam and Brogdon were in the conversation this year)

Tier 4 (NBA starter/key role player): Everyone else (although I think Sabonis, Murray, Brown and Ingram are the only players who could become All-Star level players).

Now look at the ages of his players.

1st Team: Embiid (25), Brogdon (26), Hield (26), Siakam (25), Sabonis (22)

2nd Team: Murray (22), Satoransky (27), VanVleet (25), Beasley (22), Zubac (22)

Honorable Mention: Brown (22), Ingram (21), Bogdanovic (26), Saric (25), Hernangomez (23), Finney-Smith (25), LeVert (24), Bertans (26)

The four best players were over 25. And the biggest third year jumps were made by Siakam, Hield and Brogdon.

12 out of the 17 players listed are 23 and over. Of the 4 young players on the All-Junior teams, only one of them is a starter (Murray); the others are reserves who have well-defined roles on their team and average less than 25 minutes per game.

That leaves only two players (22 and under) who are or were starters and have played at least 30 minutes per game: Brown and Ingram.

The dominance of the older players at this stage of their NBA careers shows a pretty strong correlation with having the physical and emotional maturity that comes by playing multiple years in college or in Europe before getting drafted. D’Angelo Russell’s third season made it seem like he would never be anything more than an average player, but look at the jump he made this year.

Young guys not named Kobe, LeBron, and Zion who are top 8 overall picks (this year’s class of Murray, Brown and Ingram) may be the guys who need “the milk of human kindness” (and a fourth year) to make the jump.

Regardless of where we rank these players, three things are clear:

1. The guys who really made the jump (Embiid, Siakam and Brogdon) are going to play at or near an All-Star level for the next few years.
2. There is a very low probability that 25 and over players are going to make the jump to the next level — it’s already going to be difficult for many of them to make the transition into starter roles.
3. The 22 and younger players are the ones that still have a chance to make the jump, as their bodies are still catching up to the other players in the league.

Statistical Methods to Explore

Question #1: Which metrics really tell us the most about player performance?

To make comparisons, we need to look at players who play the same position. I chose four point guards from the 2016–2017 season: Isaiah Thomas (2nd Team All-NBA), John Wall (3rd Team All-NBA), Tony Parker and Jrue Holiday to highlight the strengths and weaknesses of the most popular traditional and advanced statistics.

Traditional Stats

Thomas: 28.9 ppg, 2.7 rpg, 5.9 apg, 0.9 steals, 2.8 turnovers, 2.1 ast/to

Wall: 23.1 ppg, 4.2 rpg, 10.7 apg, 2.0 steals, 4.1 turnvovers, 2.6 ast/to

Parker: 10.1 ppg, 1.8 rpg, 3.5 apg, 0.5 steals, 1.1 turnovers, 3.2 ast/to

Holiday: 15.4 ppg, 3.9 rpg, 7.3 apg, 1.5 steals, 2.9 turnovers, 2.5 ast/to

With traditional box score stats, good players on bad teams can have inflated numbers from playing extensive garbage time minutes (On the other hand, there is also the danger of ignoring the performance of potentially great players on bad teams as in the case of Booker.).

In addition, great players on good teams might sacrifice stats to win, or play less because the team has a big lead. Finally, there are cases where limited players might have huge variations in their performance depending on whether they are in the right system or the wrong system.

In the bad old days for the Clippers, almost every player was terrible while playing for Donald Sterling, but then had a huge jump when he left the team. The reverse applies to players coached by Brad Stevens in Boston — they flourished with the Celtics, signed big contracts with other teams or were sent away in trades for better assets and then generally regressed when they played for other teams.

On the surface, most people think Thomas is the best player, but he’s basically a great scorer. John Wall was better in every other statistic, including steals, which are far more important to winning games than people realize.

Steals, blocks (when possession changes) and turnovers are vital because they have the potential to create 4- or even 5- point swings in a game. In the footnotes, I did a deeper dive, trying to derive a relative point value for each major statistic¹. Using my primitive calculations, the player per game rankings are as follows:

1. Wall 50.83
2. Thomas: 43.33
3. Holiday 39.13
4. Parker 20.16

The Value Point System (VPS) is another interesting, though flawed system. The definition is here. It places more emphasis on efficiency, passing and defense, and increases the penalty on players for missing shots, committing fouls and turning over the ball. Using VPS, we get the following rankings:

1. Wall 1.60
2. Holiday 1.53
3. Thomas 1.44
4. Parker 1.32

After doing the VPS grading on most of the top NBA Sophomores, I put this system aside because it overly rewards players who play defense and don’t take too many shots. Here are some rankings for their rookie and sophomore seasons:

2018: Ben Simmons (1.82), Jarret Allen (1.65), Derrick White (1.55), Bam Adebayo (1.53), Lonzo Ball (1.53), John Collins (1.53), Josh Hart (1.43), Monte Morris (1.72), Derrick White (1.51), Jayson Tatum (1.34), Bogdan Bogdanovic (1.25), Kyle Kuzma (1.20), Donovan Mitchell (1.14), De’Aaron Fox (1.13)

2019: Jarret Allen (1.89), Ben Simmons (1.81), Bam Adebayo (1.73), Monte Morris (1.72), De’Aaron Fox (1.53), Derrick White (1.51), John Collins (1.49), Jayson Tatum (1.39), Lonzo Ball (1.39), Bogdan Bogdanovic (1.25), Kyle Kuzma (1.19), Josh Hart (1.18), Donovan Mitchell (1.16)

Box Score (+/-) Rating: Parker (+5.9), Thomas (+4.7), Wall (+3.8), Holiday (+0.6)

Parker’s top +/- rating is solely the result of playing for the 61-win Spurs. What manner of madness is this?

“It is a tale

Told by an idiot full of sound and fury,

Signifying nothing.”

These numbers are full of noise, with huge disagreements over its importance (just read any social media battle concerning Lonzo Ball). There are too many cases where a team’s best scorer can also have the worst +/- rating, or a guy who scores almost no points can have the best +/- rating.

Advanced Stats

PER Rating: Thomas (26.59), Wall (23.28), Holiday (17.19), Parker (13.10)

John Hollinger’s famous PER rating can identify players that get lots of points and rebounds, but is totally useless when it comes to defensive players and glue guys. According to this stat, the “average” NBA player has a 15.0 rating, but fails to accurately rate the effect of a player like Marcus Smart, who has been the heart and soul of the Celtics, but has never come close to sniffing a PER of 14.0.

Defensive Rating: Parker (101.9), Holiday (103.8), Wall (108.4), Thomas (109.5).

In terms of day to day hoops reality, Parker was never a good defender. Holiday (voted 1st team All-Defense in 2018), received votes for the All-Defense team in 2017. Wall received the 7th most points among guards for the 2017 All-Defensive team. And Thomas, at 5' 9" was a huge defensive liability.

This is a sorry sight… but these statistical aberrations are explained by looking at each player’s team.

San Antonio (61–21), had the #1 defense in the league (102.8), with two players on the All-Defensive team: Kawhi Leonard (1st Team) and Danny Green (2nd Team).

New Orleans (34–48), was #9 in defense (106.7) and had Anthony Davis (2nd Team All-Defense).

Washington (49–33) was #20 in defense (108.7) and had no other players receiving votes.

Boston (53–29), was #14 in defense (108.0), and had four players receiving votes for the All-Defense team (Avery Bradley, Marcus Smart, Jae Crowder, plus what I assume was a single practical joke vote for Thomas).

Net Rating, according to NBA.com/Stats²: Parker (+6.2), Thomas (+5.1), Wall (+4.1), Holiday (0.7).

Parker was still starting for the Spurs, but only averaging 25.2 mpg. For context that same year, four guards on Spurs had a better Net Rating (Green +8.2, Mills +11.6, Simmons +9.7, Ginobli +10.1).

5-time All-Star John Wall was named to the All-NBA 3rd team in 2017.

Jrue Holiday, had a 0.7 Net Rating with New Orleans.

Thomas was the perfect fit in Boston’s system, as he was surrounded and hidden by their defensive stars, while being given total control of the offense. He had the greatest season of his career, and was in the MVP conversation.

Fair is foul, and foul is fair… unless it is James Harden getting free throws.

Net Ratings are limited and can be misleading, as good players on bad teams can look bad, bad players on good teams can look good, and players who have the right coach and teammates can look like superstars.

On/Off Net Rating, per Basketball Reference: Wall (+7.5), Holiday (+5.6), Thomas (+4.7), Parker (-3.4)

One fell swoop… on the court of hoops.

Suddenly, we have a completely different picture of each player’s value to his team.

On/Off ratings tell us what happens to a team when a particular player is either playing or sitting on the bench. It removes a lot of the noise created by the team’s overall level, as a great player on a good team raises his team’s play, while a great player on a bad team keeps the team afloat while he’s on the court, only to collapse when he goes to the bench.

While not perfect, On/Off Ratings may be the quickest way to separate the wheat from the chaff. After looking at On/Off Net Ratings, we can do a deeper dive in those ratings to see how much a player affects his team by looking at offensive ratings (ORtg) and defensive ratings (Drtg), or if those effects show up his net Offensive Rating (aka box score +/- ratings). When we do this, each player’s identity becomes apparent:

Wall was good (+3.8) when he was on court, and the team (-3.7) was bad when he was off court. His On/Off ORtg was excellent (+9.6), while only being a small drag defensively with an On/Off DRtg of (+2.1).

Holiday’s +/- rating was slightly positive (+0.6) when he was on the court, but his team was bad (-5.0) when he was off the court. But his affect on defense (-5.0 for opponent ORtg) was huge.

Thomas was good (+4.7) on court, and the team was neutral (0.0) when he was off court. His On/Off ORtg was enormous (+14.4), but he was horrific defensively, with an On/Off DRtg of (+9.7).

Parker had the best +/- rating (+5.9), but the mighty Spurs were outstanding (+9.3) when he was off the court. His On/Off ORtg was a small drop (-0.4), and a larger drag defensively, with an On/Off DRtg of (+2.9).

I think a good argument can be made for On/Off ratings to give us a better indication of a player’s overall value. It also allows us to get deeper insight by examining the offensive and defensive On/Off ratings in those cases where there are anomalies in the Net On/Off ratings.

As good as Thomas was in that special season with the Celtics, I think it’s fair to say that his success was a special combination of his offensive skills and the perfect coach and system. On the other hand, Wall (5 x All-Star) and Holiday (1 x All-Star with his former team) could be dropped into other teams and be effective. And Parker was an aging player four years removed from his last All-Star appearance.

Now before you start complaining about how I cherry picked those players from that particular season (which I did because of the massive discrepancies in almost every other metric), let’s look at other groups of players to see if On/Off Ratings give us a better insight into a player’s performance than the other metrics.

Question #2: How well do On/Off ratings correspond with the greatest players in the game?

Here’s a list of the MVPs³ over the last 15 years. Look at those massive On/Off Net Ratings. Since 2003, the average On/Off rating for an MVP is +13.3. LeBron and Curry have averaged almost +17.9.

2003 Duncan (+14.7)

2004 Garnett (+20.7)

2005 Nash (+14.9)

2006 Nash (+9.1)

2007 Nowitzki (+12.5)

2008 Kobe (+7.0)

2009 LeBron (+21.2)

2010 LeBron (+16.8)

2011 Rose (+2.7)

2012 LeBron (+14.0)

2013 LeBron (+15.3)

2014 Durant (+6.0)

2015 Curry (+17.8)

2016 Curry (+22.2)

2017 Westbrook (+12.5)

2018 Harden (+5.4)

2019 Danny Green (+18.1)⁴

With only a few exceptions⁵, the MVP has a massive On/Off Net rating.

Question #3: Do On/Off Ratings provide insights into valuable players who play good defense?

Here are the members of the 2018 NBA All-Defensive Team, along with their PER rating and Net On/Off Rating in parentheses:

Rudy Gobert (DPOY), 20.71 PER (+8.1)

Dejounte Murray, 15.09 PER (+5.6)

Jimmy Butler, 23.70 PER (+13.2)

Victor Oladipo, 23.21 PER (+14.2)

Draymond Green, 16.10 PER (+8.1)

Jrue Holiday, 17.87 PER (+12.5)

Al Horford, 17.63 PER (+8.0)

Anthony Davis, 28.98 PER (+10.8)

Joel Embiid, 22.95 PER (+12.7)

Robert Covington, 12.98 PER (+14.8)

Players that weren’t big scorers (PER Ratings under 18) still made a big impact. In fact, the only outlier of the group was Robert Covington, who will probably never sniff an All-Star game. He was a key to the 76ers’ regular season success, but then fell apart in the playoffs (-13.1 On/Off Rating).

Question #4: Do On/Off Ratings provide insights into big scorers who play poor defense?

I know a part of this has to do with the quality of a team, but here are the On-Off Ratings for some of the players with the highest PER in 2018 who had a Net On/Off DRtg that dragged down his team.

LeBron James, 28.65 PER (+2.4)

Kyrie Irving, 25.03 PER (+3.1)

Enes Kanter, 24.02 PER (-2.3)

Demarcus Cousins, 22.66 PER (-1.0)

Jonas Valanciunas, 22.52 PER (-3.8)

Kevin Love, 22.44 PER (-0.8)

DeMar DeRozan, 21.10 PER (-2.7)

Lou Williams, 20.29 PER (+2.3)

Julius Randle, 19.96 PER (+0.3)

Aside from Kanter, these are starters or sixth men who put up big numbers in big minutes. What stands out the most is how “bad” LeBron James was last year, especially when we look at his year over year progression of On/Off Ratings.

Question #5: How well does this metric correspond with year over year progression?

Here’s a list of past Rookies of the Year and how their On/Off Court ratings have changed over time:

2005 Okafor: year 1 (-0.7); year 2 (0.0); year 3 (+4.0); year 4 (+0.7); year 5 (+6.0); year 6 (-1.4)

2006 Paul: year 1 (-0.6); year 2 (+1.3); year 3 (+9.0); year 4 (+19.6); year 5 (+5.7); year 6 (+13.9)

2007 Roy: year 1 (-0.8); year 2 (+6.6); year 3 (+9.2); year 4 (+3.3); year 5 (+1.7); year 6 (+5.5)

2008 Durant: year 1 (-8.7); year 2 (-8.6); year 3 (+16.5); year 4 (-0.6); year 5 (+1.6); year 6 (+7.2)

2009 Rose: year 1 (-8.2); year 2 (+4.4); year 3 (+2.7); year 4 (+2.0); year 5 (-7.0); year 6 (+0.9)

2010 Evans: year 1 (-6.2); year 2 (+1.1); year 3 (-2.7); year 4 (-0.2); year 5 (-4.3); year 6 (+4.8)

2011 Griffin: year 1 (+5.4); year 2 (+18.8); year 3 (+1.2); year 4 (+8.4); year 5 (+13.2); year 6 (+3.9)

2012 Irving: year 1 (+4.7); year 2 (+1.0); year 3 (-4.3); year 4 (+9.2); year 5 (-0.2); year 6 (+7.2)

2013 Lillard: year 1 (+9.1); year 2 (+6.5); year 3 (+2.3); year 4 (+2.0); year 5 (+4.8); year 6 (+9.3)

2014 Carter-Williams: year 1 (+3.6); year 2 (-0.3); year 3 (+0.6); year 4 (+0.2); year 5 (-3.9); year 6 (-26.7)

2015 Wiggins: year 1 (-1.8); year 2 (+6.3); year 3 (+2.7); year 4 (+3.8); year 5 (+1.7)

2016 Towns: year 1 (+1.8); year 2 (+1.5); year 3 (+13.5); year 4 (+5.0)

2017 Brogdan: year 1 (+4.8); year 2 (-2.0); year 3 (+4.9)

Here are the progressions of five players who are sure bets to make the Hall of Fame. Years in bold represent MVP years.

LeBron James: year 1 (+2.1); year 2 (+8.8); year 3 (+10.4); year 4 (+8.8); year 5 (+10.9); year 6 (+21.2); year 7 (+16.8); year 8 (+9.0); year 9 (+14.0), year 10 (+15.3); year 11 (+6.8); year 12 (+16.8); year 13 (+16.5); year 14 (+17.0); year 15 (+2.4); year 16 (+8.1)

Stephen Curry: year 1 (+2.1); year 2 (+4.5); year 3 (+11.0); year 4 (+2.7); year 5 (+15.1); year 6 (+17.8); year 7 (+22.2); year 8 (+14.0); year 9 (+13.0); year 10 (+16.1)

James Harden: year 1 (+3.8); year 2 (+2.8); year 3 (+8.0); year 4 (-3.5); year 5 (+7.6); year 6 (+8.8); year 7 (+7.6); year 8 (+3.0); year 9 (+5.4); year 10 (+4.8)

Russell Westbrook: year 1 (+3.8); year 2 (+3.2); year 3 (-8.5); year 4 (+2.5); year 5 (+3.5); year 6 (+1.8); year 7 (+6.6); year 8 (+12.9); year 9 (+12.5); year 10 (+12.0); year 11 (+6.1)

Giannis Antetokounmpo⁶: year 1 (+7.0); year 2 (-0.6); year 3 (+3.9); year 4 (+2.9); year 5 (+9.4); year 6 (+9.7)

Based on the first three years for all of these players, the following patterns emerge. While I can’t say these correlations are set in statistical stone, and there are sure to be exceptions to the rule (like Covington). It seems like there are far fewer outlier results than with the stats I disregarded.

1. Any player who registered an On/Off Rating over (+6.5) in one of their first three years became an All-Star.
2. Any player who registered an On/Off Rating over (+9.0) in one of their first three years made an All-NBA team.
3. Only five players earned an On/Off Rating over (+13.0) in one of their first five seasons (LeBron, Paul, Durant, Griffin, Towns).
4. Any player who had a positive On/Off Rating in each their first three seasons became a franchise level player.
5. 10 of the 13 players who achieved All-Star or All-NBA level On-Off Ratings made the jump within their first three years (2 as rookies, 3 as sophomores and 5 as juniors). Irving made the jump in his fourth year (at age 22). Westbrook had a huge Net On/Off ORtg (+7.5) in his fourth year, but the team was so good it pulled down the Net On/Off Ratings for every core player (Westbrook, Durant, Ibaka, Harden). He finally registered a big jump in his 7th year, when Durant was hurt for most of the year.
6. In Rose’s MVP season, he had an elite Net On/Off ORtg (+8.8), which was offset by the excellence of the rest of the team (+6.1), which won the most games in the league that season.

Here are a few additional notes about the progressions of non-All-Star ROY players:

1. Okafor’s best Net On/Off Rating was a slightly positive Net On/Off Rating (+0.6) offset by a bad performance by the rest of the team (-5.4).
2. Wiggins’ best Net On/Off Rating was a negative Net On/Off ORtg (-1.6) that was offset by a horrific performance by the rest of the team (-7.9).
3. Brogdon’s year three Net On/Off Rating (+4.9) was a huge Net Off/On ORtg (+11.2) offset by his team’s fantastic play (+6.3). Given the fact that Brogdon had the #1 defensive rating in the NBA amongst all guards (min. 20 games and 20 minutes played), and joined the elite 50/40/90 shooting club, it’s safe to say this guy will be an All-Star very soon.

TL;DR: On/Off Ratings seemed to do the best job of answering the questions I posed.

And in the small number of outlier performances , it was fairly easy to see where a great player’s effect was either exaggerated or minimized by the strength of his teammates.

In cases where a player made a big difference on offense (Thomas), or defense (Holiday), it was easy to compare their On-Court ratings with team ratings for the rest of the league to see if they were indeed playing at an elite level.

For example, with Isaiah Thomas, the Celtics ORtg was 116.7 when he was on the court. That year, the Golden State Warriors had the #1 ORtg of 114.8.

On the other hand, when Jrue Holiday was Off-Court, New Orleans dropped from the #9 defense to #24.

In Part 3, we will finally analyze the NBA Sophomores, to determine who will make the jump. I promise it will be much shorter than this installment which seemed to take on a life of its own.

Out, damn stat! out, I say!

Thanks for hanging in there.

¹Using the league median of 1.07 points per offensive possession, each steal is worth at the minimum 2.14 points, and each turnover is worth a minimum of -2.14 points.

A defensive rebound stops an opponent’s possession and starts an offensive possession, giving it the value of 2.14, while an offensive rebound gives a team an extra possession for another 1.07 points. The median NBA defensive rebounding % is 77.1%, so we can assign a rebound the value of 1.89 points.

Assists are worth over 2 points, as they result in both two-point and three-point field goals. This year, the median team made 11.3 three points shots out of 41.35 field goals made (27.3%), so we could assume that an assist is worth 2.27 points.

From these numbers, we could roughly calculate each player’s game value in terms of median points per possession. I didn’t do all the extra detail work to figure out the value of blocks, as it’s really hard to know how often it results in a turnover. Obviously, by using median values there could be dramatic changes, depending on the efficiency of each player and his teammates):

1. Wall 50.83
2. Thomas: 43.33
3. Holiday 39.13
4. Parker 20.16

²For some reason, there are glaring differences between the defensive and offensive player ratings on NBA.com, the “Splits” offensive and defensive ratings on Basketball Reference, and “player efficiency” on TeamRankings.com. I used the ratings provided by NBA.com, but the Off/On Ratings by Basketball reference because I didn’t find anything like it on on a team basis on NBA.com.

³MVP votes are obviously skewed by a lot of factors that have nothing to do with how much a player impacts his team. Sometimes, it’s just the best player on one of the best regular season teams. Sometimes, the best player misses too many games (Curry in 2018). Sometimes, it’s a makeup vote for a past injustice (Kobe in 2016 was a +12.6, then +17.6 in the playoffs). Sometimes, a young player gets overlooked (Finals MVP Wade in 2006 was a +15.2). And sometimes it’s just that feeling that “we can’t keep giving it to the same guy every time” (nods to MJ and LeBron).

⁴Yes, this is a joke.

⁵There was a strong backlash vote against LeBron (+9.0) in 2011 and Curry (+13.0) was injured for too many games in 2018.

In 2008, Kobe had a solid Net On/Off ORtg (+9.0) but the team was good without him (+2.0). Chris Paul, the second overall vote getter, was close in Net On/Off ORtg (+8.1), but his team’s weakness (-0.9) raised his On/Off Rating (+9.0).

In 2018, Harden had a solid Net On/Off ORtg of 10.5, but the team was still really good without him (+5.1).

⁶In his fourth season, Giannis made a jump in efficiency, registered his first positive Net On/Off ORtg (+0.9) and got his team into the playoffs. In year five, he had a better Net On/Off ORtg (+2.8), and his lousy team (-6.6) pushed up his Net On/Off Rating to 9.4. This year, Giannis exploded with a Net On/Off ORtg of +12.5. His team was so good (+2.8), his net rating dropped.

More NBA Stuff…

Talk about opposite directions: West sticking around for the playoffs, and Magic running out the exit at Staples.