Saturday, June 30, 2018

Article archive

Since the internet has been subsumed by the ubiquitous "load more" button everywhere, it's difficult to find certain articles. I've decided to make an archive of my work to make that easier; at the very least I'll use it frequently when I can't remember where I posted something or what my method was years ago. Just a warning that this won't be updated religiously. But I will post things chronologically in each section with the most recent at the top, and the recommended section will just have a few of my favorite pieces I think people should at least try first.

Recommended:

How Will We Remember Kevin Garnett? along with the conclusion piece Remembering Kevin Garnett

Roy Hibbert and the Search for Truth in Defensive Stats

Closure on 3-point Team Defense

Usage Versus Efficiency

Reinventing PER

Week in review archive:

NBA Awards Picks Weeks 25 and 26: So Long, and Thanks for All the Fish. Game 82 MIN-DEN, defender positions, Westbrook's dismissed triple-double average, Jokic's rebounding, Andre Ingram's debut, 6MOTY, MIP, all-rookie, ROY, COTY, OPOY, all-defense, DPOY, all-NBA, and MVP.

Week 24: LeBron’s defense, 76ers, and Raptors playoff rotation. LeBron's defense is a liability, origins of an MVP with how team results are misleading, play-by-play data inaccuracy, Embiid's broken face, post-ups and spacing, Markelle Fultz rebooted, Toronto's playoff rotations, Jerami Grant versus Carmelo Anthony, and LeBron James' almost-there triple doubles.

Week 23: Spurs, Raptors, and Reader Questions. The fall of San Antonio, leaders in offensive fouls, Kyrie Irving and the injury bug, Toronto treated as the 30th team, Grizzlies lose by 61 points, all-NBA guard spots, +/- for hedging versus switching, what it means to be good or bad versus the best teams, and how to quantify tanking.

Subject archive:

Who Deserved the 2018 Finals MVP?




Wednesday, May 21, 2014

Homecourt Advantage in the Playoffs

I've tackled homecourt advantage before as a topic, specifically looking at travel distance and rest. An interesting discovery, and one that's not too surprising after the fact, is that true homecourt advantage is smaller than most assume it is because more away teams play on the end of a back-to-back. The typical value people use is +3.2 for the home team, while if you take back-to-back games into account it drops to +2.8. That may not seem like much, but in the competitive NBA (and in the betting market) those margins are important.

One would assume then that the playoffs would actually have a smaller homecourt advantage than the regular season because there are no back-to-back games. Crowds may seem louder in the playoffs, but the NBA statistics world is one of skepticism -- we're not prone to giving credit for good will and exuberance. However, home teams are vastly better in the playoffs, and it's fairly consistent season after season.


Regular season
Regular season (no back-to-back’s)
Playoffs
Homecourt advantage (points per game)
3.25
2.80
4.32
Seasons 2000 to 2013

(Technical note: you can't simply compare the points scored by home teams in the playoffs to the away teams because teams don't always play an even amount of games. The better teams are most likely to play more games on their own floor. To correct for this, I added a "virtual game" to every series with an uneven amount of games. The virtual game is just an average of the real games played at one location. For example, for the Spurs-Blazers series, the virtual game is the average of games 3 and 4, so that now there are three games at San Antonio and three at Portland.)

As a check, a team rating system was built via linear regression for seasons 2010 through 2013 with the playoffs included. Homecourt advantage was estimated at 2.89, while in the playoffs it worth over an additional point at 4.02 overall. Over the same period using the simple method as in the above table, homecourt advantage was 3.94.

How did this season rate on its own? So far homecourt advantage has been roughly a full point less in the playoffs than the regular season at around 1.8. That's the worst differential by far since 2000 -- and given the crazy first round, it's not surprising. No season compares to this one yet, but there was another outlier season but in the other direction. In 2008, home teams were better by an average of 8 points. That's an enormous advantage; it's like the difference between a 42 win team and a 62 win team.

Crazy fluctuations notwithstanding, over multiple seasons homecourt advantage appears to be much higher in the playoffs than the regular season. There is something to play for. And if there's a winner in all this, it might be the fans for providing their home team the support when every game matters.

Monday, May 19, 2014

Western Conference Finals: Spurs and Thunder

The discussion on these finals is mostly how the Thunder will fare without Ibaka and how the athletic Thunder are the antidote to San Antonio's precision passing and movement. It's hard to say what exactly the Thunder will do without Ibaka on the team since he only missed 3 games in the past 3 seasons. But we could also have the same finals we've seen the past couple years, barring Indiana making it past Miami. However, Chicago and Utah met two times in a row, and no one downplays the brilliance of the '98 finals. This should be an enjoyable series, and the loss of Ibaka makes this a fresh matchup with some unpredictability.

-This is an opportunity for Oklahoma City to play small with Durant as the nominal PF and try to blow San Antonio out of the water. The Thunder can go small without too much concern because the second big man for the Spurs is usually either Splitter or Diaw -- you can live with them shooting more often.
-I'm not one for calling out a coach, but I'm afraid Scott Brooks will respond to the loss of Ibaka by playing Perry Jones and Caron Butler heavy minutes. San Antonio would clobber them.
-In two games without Ibaka in 2013, Nick Collison started one and played heavier minutes; Jones started the other game. Perhaps Brooks wanted Collison matched up with Scola in one game and was fine with Jones in the other since it's easier to hide someone on Hickson.
-My own basic team strength metric has San Antonio with a +1.5 point advantage. This includes the playoff games so far. However, when you vary the strength metric by opposition strength, the Thunder are better versus elite teams.
-Tony Parker is also a bit limited, but with a long rest period he could be completely fine. This is what to watch for in the first game.
-There's a rust versus rest argument here, which has been overshadowed by the Ibaka news. The results are unclear with respect to rust and rest, but from my own analysis rust is not harmful.
-The good thing about doing a post right before tip-off: there's news Collison will start. Brooks did start him last season, so it's not a complete surprise. And as Pelton noted, the Collison-Durant pair destroys opponents.

Prediction: San Antonio in seven.

Team................
Team rating (point diff.)
Strength variance
Team rating versus +3 team
Team rating versus +6 team
Oklahoma City Thunder
6.73
0.39
8.49
10.25
Los Angeles Clippers
7.32
0.08
8.15
8.97
San Antonio Spurs
8.24
-0.49
7.35
6.46
Houston Rockets
5.00
0.05
5.75
6.50
Golden State Warriors
4.97
-0.32
4.61
4.24
Dallas Mavericks
3.21
0.18
4.33
5.45
Miami Heat
4.04
-0.20
4.02
4.00
Portland Trail Blazers
3.96
-0.27
3.72
3.49
Indiana Pacers
3.45
-0.11
3.72
3.99
Phoenix Suns
3.05
-0.03
3.54
4.03
Minnesota Timberwolves
3.03
-0.22
2.97
2.91
Toronto Raptors
2.11
0.08
2.93
3.75
Memphis Grizzlies
2.12
-0.02
2.64
3.16
Chicago Bulls
0.83
0.10
1.73
2.62
New York Knicks
-1.37
0.51
0.75
2.87
Washington Wizards
0.82
-0.28
0.56
0.30
Brooklyn Nets
-1.13
0.06
-0.36
0.42
Atlanta Hawks
-0.57
-0.26
-0.77
-0.97
Denver Nuggets
-1.43
-0.14
-1.27
-1.10
New Orleans Pelicans
-2.03
-0.06
-1.60
-1.18
Sacramento Kings
-2.10
-0.11
-1.82
-1.55
Detroit Pistons
-4.11
0.56
-1.83
0.45
Charlotte Bobcats
-0.94
-0.53
-1.93
-2.92
Cleveland Cavaliers
-3.83
-0.10
-3.54
-3.25
Boston Celtics
-4.90
0.03
-4.23
-3.57
Orlando Magic
-5.85
0.24
-4.54
-3.23
Los Angeles Lakers
-5.29
-0.45
-6.06
-6.83
Utah Jazz
-6.22
-0.39
-6.80
-7.38
Milwaukee Bucks
-8.43
0.32
-6.89
-5.34
Philadelphia 76ers
-10.66
-0.39
-11.25
-11.84

Sunday, May 18, 2014

Eastern Conference Finals: Hibbert and Bynum's Freaky Friday

A couple months ago when Indiana was flounder and the basketball world was awaiting the playoffs, we all worried the eastern conference would be a pointless slog. However, Indiana righted the ship, and they defeated a good Washington team with relative ease. Hibbert, in particular, has made a great comeback. It was seemingly an overnight transformation from a staggering zombie to the old defensive whiz we had known from earlier in the season -- he even rediscovered his offensive game.

But what caused his slump? What's odd is that he's done this before. After the large contract extension last season, he struggled with his shooting touch. Before the all-star game he averaged only 10.0 points per game on a ghastly 41.4 FG%, and after his scoring increased to 15.7 points on a 50.8 FG%. All his other stats had remained stable, and there were reports a nagging wrist injury destroyed his offensive game for the first half of the season. So by the all-star break last season, he was being called a massive disappointment on his large contract, and jump ahead a year to this season's all-star break he was a centerpiece defensive player who looked like a lock for Defensive Player of the Year. But after the all-star break, not only did his scoring and shooting percentages drop, everything did.  What happened?

Hibbert's all-star game splits are pretty astounding, but there's another factor here: the Bynum trade. Although Bynum barely played, his time on the team had a suspicious negative effect. Indiana played worse for multiple reasons -- it's rarely just one cause, even though everyone likes to posit their own theory -- and one of them was Evan Turner's quiet disaster on the court, but you can't blame Turner for Hibbert. What's strange is that Hibbert improved immediately after Bynum was released from the team in the playoffs.

The results are striking. His scoring decreased, and it was from both angles: his shooting efficiency plummeted to pathetic levels and he shot less often. Unfortunately, even though he assisted less often, his turnovers slightly increased. His rebounding fell off too, and it was in both areas. He even blocked less shots.

 Bynum
PTS/75 poss
TS%
Usage
TOV/75 poss
OREB%
DREB%
Blk/75 poss
Ast%
GmSc/75 poss
With
15.9
53.4
20.2
2.3
11.3
16.5
3.2
6.9
13.4
Without
11.9
45.2
18.2
2.5
7.7
12.0
2.6
4.8
7.0

Game score is a basic summation of box score stats invented by Hollinger as a single-game PER statistic. It was nearly chopped in half, which is, frankly, astounding. The graph below shows a 5-game rolling average; it shows how well he does in five game chunks (weighted by recency, so the most recent game is more important.) It's not perfect, but Hibbert's slide is apparent, and he hit a low point, obviously, versus Atlanta, and then had a meteoric rise to close the series and then help defeat Washington. It appears Hibbert progressively got worse as the season went on before hitting rock bottom in the playoffs, but it's a curious coincidence Hibbert improved considerably once Bynum was released.


.There are likelier causes to Hibbert's swoon. For instance, he put on extra muscle weight in the off-season, and it may have been too much for his 7' 2" frame to get through the season. When his game sank and his foot speed diminished, he also lost his confidence -- that's a possibility. Or, of course, when Bynum was signed he switched bodies with Roy Hibbert a la Freaky Friday, and Bynum's lack of effort led to Hibbert's awful second half of the season, while Hibbert was sidelined in Bynum's body. Just tell everyone the numbers support this theory.

Series notes:
-Miami is the magic elixir for Indiana's offense. Last season their offensive rating was 104.3, and against Miami in the playoffs they scored 107.2 points per 100 possessions. Since Miami is an above average defense, that's a fairly significant increase.
-LeBron's been playing historically good offensively in the playoffs, but he shouldn't be ignored because of the lowly eastern conference; they have been playing good defenses. (The Nets have had a good defense since January.)
-Hibbert, as highlighted above, was terrible during the second half of the season, but he did destroy Greg Oden in a late season game, and Oden has barely played since then. That's quite remarkable, and a sad twist in a strange tale of two centers.
-Last playoffs Wade couldn't crack a 50 TS%, and now he's hanging around 55 TS%. It's a bit astounding LeBron won a title with Wade being unable to shoot well for most of the playoffs; he gets little credit for that.
-Paul George, meanwhile, is averaging 22/9 with a 57 TS% and great defense. The fickle NBA audience has turned on him already, but he's lived up to the hype.
-At the beginning of the season I criticized Miami heavily for going after Beasley because I felt that even Beasley at his best is the wrong fit. He had a good start to the season, but then faded and lost a spot in the rotation due to poor defense. And when replaced by virtually everyone else on the team, the team is better. Sorry for being right, Miami.
-Homecourt advantage is tossed around a lot when discussing series outcomes, but I've seen very little detailing the important of homecourt advantage in the playoffs. People probably assume the same kind of strength as in the regular season but without any stats backing up that claim; however, I do think most people actually think homecourt advantage is greater in the playoffs. The complicating factors are that better teams play more often on homecourt and there's no back-to-back games in the playoffs -- since more away teams play on the end of a back-to-back, this reduces the strength of HCA. But looking at the past three playoffs (2011 to 2013), and adding a "virtual" game to any series with an odd number of games so that teams have the same number of home and away games, homecourt advantage is +3.84. That's much greater than the typical value assumed (3.2) -- and without back-to-back games, it drops to a mere 2.8. Given how strong the crowds are in the playoffs, this shouldn't be a surprise.
-Full season results are more predictive than post all-star break results for teams -- which does not bode well for Miami. If Indiana is back, they should not be lightly handled.
-Miami beat Brooklyn by an average of 5.4 points per game. Since Brooklyn was a +1.4 point differential team after January first (when their rotation settled after dealing with injuries) in a weak conference, this suggests Miami is a +6 strength team. Miami also beat Charlotte by 10 a game, and since Charlotte was a -1 SRS team that suggests a strength of +9. By reasonable estimates in the playoffs, Miami has been a (roughly) +7.5 team. The assertion that they were coasting in the regular season and would shift a gear in the playoffs is at the very least plausible. Having Wade for every game, of course, is a tremendous help. The only problem now is a recovered Indiana team that matches up well.
-Indiana took game 1, but the game was an offspring of the past few games we've seen from them. Hibbert and West bully their way inside, while Miami makes runs when Indiana goes to their bench. This should be another close series. 

Thursday, May 15, 2014

2014 Defensive Awards

As the playoffs progress, let us not forget the regular season, where few people cared about the Wizards and Durant was king. For this awards installment, it's all about defense, which remains the trickiest area to judge. A few advanced stats are useful, mostly plus/minus varieties, but the "eye test" is also very important. As such, video is available for every major player discussed. Since stats.NBA.com has every play accessible and viewable, I hope it's used more to discuss players, especially to an audience unfamiliar with them.

First-team All Defense

G - Ricky Rubio

Rubio makes first team for having the size to smother point guards or cover shooting guards. He picks up a high number of steals, but he's not a dangerous gambler. In the first clip below, you can see him guarding Chris Paul through picks, quickly recovering to stop Crawford, and then poking the ball out of his hands causing a turnover. The second clip shows a rare Chris Paul bad pass, where Rubio intercepts a pass to Griffin very easily with his wingspan.



G - Andre Iguodala

He's the league's premier wing defender in his prime, casting a large shadow on defense. Since he's changed teams a lot lately, we can see the outline of his impact. In 2012, Philly was 3rd in defensive efficiency, and when Iguodala left they fell to 15th. Denver improved from 20th to 11th. When Denver lost him after a single season, they dropped back down to 21st. And Golden State improved from 14th to 3rd. While there are other factors here like Bogut's health and coaching, Iguodala fares well by other metrics. He leads the league in pure adjusted plus-mins (RAPM) and is rated 8th by ESPN's RPM (RPM has a bias for big men, however, by incorporating height.) In the below clip you can watch him keep up with an aggressive LeBron and force a tough shot (which LeBron makes somehow.)



F - Paul George

Paul George is the top wing defender on the best (regular season) defensive team in the league. I'm not quite sure he tall he is anymore -- the offseason growth reports may not have the best sources -- but he's big enough to pester almost anyone on the perimeter and quick enough to cover multiple positions. You can watch him force Joe Johnson into a tough shot below.



F - Taj Gibson

The Bulls have consistently played better defense with Taj on the court for years. He's strong, long-armed, and executes Thibodeau's system well. Pure RAPM has loved him for years. In the clip below, he ends up guarding the uber-athletic John Wall, who can't get past Taj and launches an off-balance jump shot that's blocked.



C - Joakim Noah

He's the defensive player of the year, so this isn't a shocking pick. However, his plus/minus numbers have been lukewarm for a while, and it's unclear why this is. Perhaps RAPM is overrating Boozer's defense and not giving enough credit to Noah. For instance, Chicago's defense has been better or neutral when the same four-man lineups have Noah instead of Taj Gibson, who regularly posts a high RAPM. Nonetheless, Noah's an intense defender and the central defender on the league's second best defense, playing huge minutes and doing crazy things like being the only guy who defends in an all-star game, as shown below.



Second team

G - Patrick Beverley

The playoffs showed Beverley's peskiness, pressing Lillard and hounding him for every second despite coming off an injury and falling ill. He's probably the most annoying defender in the league, the Dennis Rodman of point guards. His low minutes total hurts him, but there weren't too many worthy guard candidates. He's a good enough thief to steal one from Chris Paul, as seen below.



G - Jimmy Butler

Butler's a tall, tenacious defender who's a perfect fit for Thibodeau's defensive system. He rates fairly well by most defensive metrics, but he's probably far from his peak. However, as with Beverley, the competition wasn't intense, and he's certainly better than, say, Chris Paul, who's not always engaged on defense. Butler guards LeBron fullcourt in the clip below, staying with him until the end, where he blocks LeBron's shot.


F - Kawhi Leonard

Like Butler, Kawhi has a great environment in which to learn defense, and he got an accelerated course in defense guarding LeBron in the finals. He wasn't a destructive defender like a Tony Allen, so given that he was under 2000 minutes Kawhi might be a surprising pick, but he only played low minutes because his team wanted everyone fresh for the playoffs. In the clip, he stays with Durant on the perimeter, taking away a three, and recovers on a drive to nearly steal the ball and bother his shot.



F - Tim Duncan

Old man Duncan remains a top-notch defender even as he approaches his 40's. His secret is being able to contest and block shots without even leaving the ground, often getting on his toes rather than jumping. This allows him to be nearly immune to shot fakes, keep his positioning, and be ready for a rebound. In the clip below, he comes over to stop a Durant shot at the rim at the exact right time and blocks the ball so well they call it a jump ball. He does this by barely jumping.


C - Dwight Howard

No one likes Howard anymore, but he's recovered from his major injuries and is back to being an ideal center on defense, guarding the rim, rebounding at a high level, preventing easy post-up plays, and being quick enough to navigate switches and guard smaller players. For instance, the intimidating Pekovic was destroyed on a post-up in the first clip. Howard's as strong as anyone, and he has a monstrous wingspan. The second clip shows an example of how Howard doesn't spike every blocked shot into the stands anymore; he's now directing them to his teammates. According to NBAWOWY.com, when Howard blocks a shot, his team recovers the ball 64.8% of the time, which is significantly above average (the average is usually in the upper 50's.) For comparison, San Antonio team only rebounded 57.0% of Duncan's blocks, and last season it was still under Howard's mark: 63.9%.




Player..............
Minutes
Blk%
Stl%
DRB%
Team Drtg
Iso/post PPP
RPI RAPM
RPM
Ricky Rubio
2638
0.3
3.6
11.9
-0.4
0.65
1.61
2.45
Andre Iguodala
2040
0.7
2.3
12.6
-4.1
0.81
4.05
4.74
Paul George
2898
0.6
2.7
17.7
-6.3
0.78
2.29
2.70
Taj Gibson
2531
3.7
0.9
17.1
-5.0
0.83
2.60
3.59
Joahkim Noah
2820
3.3
1.9
24.5
-5.0
0.78
1.38
3.88
Patrick Beverley
1751
1.0
2.2
7.4
-0.1
0.58
0.33
1.30
Jimmy Butler
2591
1.1
2.6
10.5
-5.0
0.72
1.58
1.23
Kawhi Leonard
1923
1.9
3.0
18.9
-4.2
0.69
1.07
1.98
Tim Duncan
2158
4.6
1.0
28.3
-4.2
0.80
2.47
5.25
Dwight Howard
2356
4.0
1.2
27.9
-0.1
0.74
2.97
4.91
*Iso/post PPP is the points per play allowed via Synergy. For C's and PF's, post PPP is used.
**"Pure" (RPI) RAPM was taken from this site.
***Team Drtg is a team's points allowed per 100 possessions adjusted for strength of schedule.


Defensive Player of the Year

With injuries to key defenders and the collapse of Indiana in the second-half, an opening emerged. Critics of the modern game like to tout the virtues of past legends, but they fail to give credit to the obstacles players face today -- playing against the best players from several countries, detailed video scouting rendering any weakness you have visible and exploitable, and a quick, athletic league with defensive rules that hurt slow help defenders. If you're going to be an elite defensive player in the modern NBA, it's not by accident.

3rd place

Dwight Howard

Howard's underrated now, and a multiple-time DPOTY winner who's back to looking like himself is not a strange pick for third place. I can imagine some people arguing for DeAndre Jordan, but that's silly: he still made too many mistakes, and his gaudy block totals aren't important if he's constantly being faked out of his shoes.


2nd place

Andre Iguodala

If he had played more minutes, I would have argued him for 1st place. I detailed his exploits above. He holds down the fort on a defense that collapses without him. While today big men are usually more valuable on defense, a smart, athletic, and long wing defender can cause the same impact.


1st place

Joakim Noah

He won due to attrition. He played a huge amount of minutes for an outstanding defense and played center. That's the argument in a nutshell. He's versatile and does well at almost everything you'd want from a center on defense. This is also likely to be his best season ever. It's a fine choice for the award, though a bit underwhelming.

Finally, there are a number of honorable mentions:

Mike Conley should be a candidate for a long time, but wasn't as good this season. You can see how he uses his quickness to stay in front of Lillard in the clip below, and forces him into a two-point fallaway jumper.



Tony Allen is still a terror on defense, and when healthy he's the only wing defender who approaches Iguodala. This was evident in the first round of the playoffs when the Grizzlies nearly upset the Thunder. In the first clip, Allen denies entry to Durant and steals the ball. In the second, he fights through a screen and blocks his shot.



And for the last part of the Grizzlies appreciation chain, Marc Gasol, who won the DPOTY last season and was out of the running when he injured his knee. But he remains a great defender and was unfairly forgotten this season. He's not exactly athletic, but he has great timing and knows just when to slide over; an example is below.



For the final clip, here's the underrated Amir Johnson destroying a Miami Heat pick and roll and saving the ball for his team.




Hibbert is the most notable exclusion, but Indiana actually defended much better with Mahinmi on the court and Hibbert had problems guarding quick players out of the paint or, really, the perimeter in general. The Atlanta series exposed these issues. Ibaka's a very good defender, but his blocked shot totals don't make him into a candidate -- he may not even be the best frontcourt defender on the team. Sefolosha regressed, though he's still pretty capable. Splitter was very effective when he played, but it was far too infrequent. Nene had the same issue -- and with him on the court, Washington punched above their weight and knocked out Chicago. And lastly, Draymond Geen may sneak onto a list someday, as the lineups with both Iguodala and Green dismantled other teams on defense.