Saturday, April 19, 2014

2014 Playoff Preview: Round 1, Part 1

As the internet is flooded with previews for the playoffs, I thought I'd keep my predictions concise and include whatever information I found to be unique.

East

Indiana Pacers (1) versus the Atlanta Hawks (8)
-While much has been written about team chemistry during their second-half implosion, CJ Watson was injured for most of the last third of the season, and the results were pretty drastic: before 3/5/14, they had a point differential of +7.7 as the best in the league, and after that -4.7, which includes four games he played in where they won three. At the beginning of the season I highlighted how a competent backup point guard would help their woeful bench, and it's proven true here, as without him the offense falls apart when they go to the reserves. They also need his outside shooting.
-The Evan Turner trade was a couple weeks before that, but they faced a weak schedule before losing Watson and had single digit wins against teams like Boston. The stat geeks were derided for not believing in Turner's potential, with all that silly talk about pace inflating stats and inefficient shooting. But we've seen who's been more accurate long-term.
-Remember, Indiana did beat OKC recently (and yes, CJ Watson played.)
-Despite Indiana's strength and Atlanta's lukewarm season, this is a poor matchup for the Pacers. Every major rotation player besides Elton Brand can shoot a three, and they'll force Hibbert out to the perimeter where he's not quick enough to contain pick and roll's and step-back three's -- or simply not used to it.
-The Hawks were treading water above 0.500 for a while, but Millsap's injury was too much to handle, as losing Horford for the season destroyed their effective depth. With Millsap, it's a good team.
-This is a test of faith of sorts for NBA fans -- do you believe in full season results or momentum? Who are the Pacers now? If these Pacers are what their numbers suggest over a full season, they should have few problems extinguishing the Hawks; however, if they have truly slipped this could be a hard-fought battle with an opponent who will put their big men into uncomfortable positions on defense.

Prediction: Indiana in five games.

Toronto Raptors (3) versus the Brooklyn Nets (6)
-After the Rudy Gay trade, the Raptors have been a +4.4 team, fifth in the league.
-Once Kirilenko came back from injury, the Nets were +1.4 in point differential the rest of the season, much better than their total season results. However, during the same time span the Raptors, naturally, were significantly better.
-Note that Brook Lopez went down for the season just a few days before Kirilenko came back. Once he was removed from the team and the Nets had more versatile weapons, their odd lineups with all 6' 7" guys plus a center were very effective. It's also more evidence that center scoring is overrated, at least when the center is not a good passer/distributor.
-Toronto has been a deceptively strong team with only one all-star, especially since that all-star is probably only the third best player on the team after Lowry and Amir Johnson.
-Most full season team ranking systems are harsh on the Nets due to their early struggles and their incessant resting, including that last game of the season where Jason Collins played 23% of his total season minutes.
-Weighing by projected starter minutes and ESPN's real plus/minus, the Nets are actually a stronger team, but this all depends on how much they'll use their lowly rated young guys, Plumlee and Teletovic.
-An injury to a key Brooklyn player is not a remote possibility.
-You have to really distort the numbers to make Brooklyn look significantly better. If it's close, Toronto's homecourt advantage is the trump card.

Prediction: Toronto in seven games.

West

Oklahoma City Thunder (2) versus the Memphis Grizzlies (7)
-Along with the Clippers, I found that OKC played a lot better as its competition increased.
-An ESPN video had some new metric about "synergy" that found Durant and Westbrook had negative synergy, which runs counters to previous two-man teams with titles like LeBron-Wade and Kobe-Pau. I'm not exactly sure what this metric is, but it's probably close to going to the player pair plus/minus page here and comparing a pair to what the team's overall plus/minus is. For example, Durant/Westbrook rated a +7 there, which is nearly identical to the team's adjusted rating on basketball-reference. LeBron/Wade were a +7 too (though LeBron/Bosh were +9), but the team's adjusted overall rating was only +4.7, suggesting a positive "synergy."
-After Marc Gasol came back, the team posted a +3.5 point differential. This is a pretty scary 7-seed.
-The Grizzlies will swarm Durant well, but I don't expect it will be effective during these playoffs with Westbrook back to draw attention and help set up their role players, and with a better Durant.
-Prince is going to see major minutes, and in theory his length is useful against Durant ... but he's long in the tooth and hurts their already anemic offense. This is a team that needs a major punch in its offensive attack.

Prediction: Oklahoma City in five games.

Los Angeles Clippers (3) versus the Golden State Warriors (6)
-Bogut going down with an injury reduces their chances, but if they play Draymond Green heavy minutes they can still be an effective team.
-The Warriors, in fact, are an elite team if they don't let the bench get in the way, as Marreese Speights and Harrison Barnes are two of the worst rotation pieces for good playoff teams. Fortunately, they have Steve Blake to replace Jordan Crawford and, before him, the general void of not having a real backup point guard.
-Mark Jackson's strange insistence on playing bench units separately and not staggering key ball-handlers Curry and Iguodala, or even David Lee, limited their firepower in the regular season, but in the playoffs this is less of an issue. They could be better in the playoffs.
-Chris Paul is a fantastic playoff player and deserves a better reputation.
-Going back to the synergy metric, Paul/Griffin were +11 versus a +7.3 rating, for what it's worth.
-With Redick back and Glen Davis replacing some of the awful big man bench minutes, the Clippers are fully rounded and are much better than most think.
-Using my new variable team strength metric, I found the Clippers rated best in the entire league versus playoff quality competition. Meaning, their regular season point differential underrated how good they were against better opponents (or, conversely, how much they let off the throttle versus inferior competition.)
-With Bogut out, Jermaine O'Neal and Draymond Green are the best options for Griffin. But O'Neal is old and will probably get in foul trouble, and Green is too small -- Griffin underwhelms against bigger defenders. Griffin may destroy poor David Lee.

Prediction: Los Angeles in five games.

Thursday, April 17, 2014

Variable Team Strength Ratings

At the beginning of the season when previewing teams I introduced something I titled "V-SRS," which was a tweak of basketball-reference's team rating metric, SRS, that adjusted a team's SRS based on how strong their opponent was. Basically, it tests the idea that some teams run up the score on weak teams or others are stronger versus the best teams in the league. With the playoffs coming, I thought this was the perfect time to try for an entire set of rankings based on "V-strength."

Quick description: the team rating (point diff.) column is just a basic team strength metric adjusted for the schedule, including back-to-back games. It's very similar to SRS. Strength variance is the degree to which a team's strength rating varies based on how good the opponent is. I've provided two columns to illustrate this: a team's rating versus a +3 team, which is a typical playoff team and one near 50 wins, and what a team's rating would be versus a +6 team, which is a strong contender as most title-winners are near +6 or over. Since the metric needs to be tested over multiple years still, I sorted teams by their strength versus +3 teams because it's a more conservative estimate of how good they are against competitive playoff teams.

Team
Team rating (point diff.)
Strength variance
Team rating versus +3 team
Team rating versus +6 team
Los Angeles Clippers
7.27
0.15
7.90
8.53
Oklahoma City Thunder
6.60
0.31
7.68
8.76
San Antonio Spurs
8.01
-0.44
6.87
5.72
Houston Rockets
5.08
-0.02
5.20
5.31
Golden State Warriors
5.11
-0.17
4.78
4.45
Portland Trail Blazers
4.37
-0.24
3.83
3.28
Miami Heat
4.14
-0.18
3.77
3.41
Dallas Mavericks
2.93
0.13
3.49
4.05
Indiana Pacers
3.66
-0.15
3.37
3.09
Phoenix Suns
3.05
-0.04
3.11
3.16
Toronto Raptors
2.53
0.09
2.95
3.37
Minnesota Timberwolves
3.03
-0.19
2.62
2.21
Memphis Grizzlies
2.15
-0.09
2.05
1.95
Chicago Bulls
1.13
0.08
1.53
1.92
New York Knicks
-1.37
0.47
0.22
1.81
Washington Wizards
0.49
-0.45
-0.70
-1.88
Brooklyn Nets
-1.57
0.09
-1.13
-0.70
Atlanta Hawks
-0.84
-0.31
-1.62
-2.40
Denver Nuggets
-1.43
-0.15
-1.71
-1.98
New Orleans Pelicans
-2.03
-0.05
-2.00
-1.97
Detroit Pistons
-4.11
0.57
-2.23
-0.36
Sacramento Kings
-2.09
-0.10
-2.23
-2.38
Charlotte Bobcats
-0.92
-0.63
-2.67
-4.41
Cleveland Cavaliers
-3.83
-0.10
-3.97
-4.12
Boston Celtics
-4.90
0.03
-4.66
-4.42
Orlando Magic
-5.86
0.24
-4.97
-4.08
Los Angeles Lakers
-5.30
-0.43
-6.43
-7.56
Utah Jazz
-6.23
-0.38
-7.20
-8.17
Milwaukee Bucks
-8.42
0.31
-7.34
-6.26
Philadelphia 76ers
-10.66
-0.37
-11.62
-12.57

*Note: a positive strength variance means your team's rating increases as the opponent gets better, and negative means your team rating's gets worse as the opponent gets better

When the media cite records of, say, the Spurs versus playoff teams, they're essentially doing a weak version of variable team strength. However, they do not adjust for schedule or the point differential of those games. Thus, the variable team strength metric, something I'm calling "V-strength" for now, more accurately reflects how good these teams are versus the top competition without completely disregarding all the other games and with an objective statistical adjustment. 

One point here is that, yes, the Spurs were better against weak teams, though they rested their top guys so heavily that once the starters play more minutes the effect could be washed out. Also, I believe people are sleeping on the Clippers. They appear to play well against top competition, they're healthy at the right time, and they're being underrated because they're a new team on the big stage. The Mavericks and Bulls are two I would have thought of before crunching the numbers, as Dirk plays well in all situations and the Bulls have a top-notch defense and coaching to grind out wins versus better teams. The Heat, however, look worse, but perhaps with more playing time for Wade they'll punch at a higher weight class than their numbers indicate.

It's a new, untested metric, but it's good to quantify a natural question that's constantly arising -- how good are you against the best teams?

Monday, April 14, 2014

2014 NBA Awards and Voting Philosophy

This covers the "miscellaneous" awards. I'll cover the defensive awards in another article and the all-NBA/MVP in another.

NBA awards are inherently meaningless. Arguing over them is like starting a fist-fight with a ghost: no one will see it, and you look crazy. But we still invest a lot into awards because they signify something official -- Defensive Player of the Year winners are etched in stone, some players are known first for their awards, and there's a decent amount of media attention for even the less important awards. If we're going to keep running out these things, we need to be smarter about it.

First of all, as John Hollinger said, you cannot make up your own rules to vote. People have added the "must be on a winning team, preferably a contender" criterion to Most Valuable Player, and because people have been doing that for a while the momentum is strong enough to keep this mindset going. It doesn't matter to people that it's an incorrect voting method -- because that's how things were done in the past, that's how they'll vote in the future. You can point out even legends like Kareem and Garnett missed multiple playoffs in their prime, but it doesn't matter. How good your teammates are has an effect on your MVP chances. You should not create your own rules.

If anyone reading this has a vote (slim chance), please consider that we're voting for individual awards and don't be fooled by basic points per game stats. Even real plus/minus has some glaring errors, as advanced plus/minus has continually underrated Noah's defense, for example, and we should always take a step back and consider the context. With that said, most of the awards in reality have different meanings -- and I do hope we start to move away from these invented definitions.

Most Improved Player Award
(aka player with the biggest increase in points, rebounds, and/or minutes)

The Phoenix Suns.

Outside of giving the award to an entire team, there are a number of good candidates here, and it's hard to narrow it down. I hope it doesn't go to Andre Drummond or DeAndre Jordan, who are simply playing more minutes. Cousins, Anthony Davis, Lillard, Dragic, Griffin, and Paul George have all taken a giant step forward, and they weren't non-entities before. There are a few other role players worth mentioning, like Kendall Marshall, Gerald Green, Isaiah Thomas, and Lance Stephenson, but the most impressive improvements, in my opinion, have been from Anthony Davis and Dragic. People saying Davis shouldn't win the award because we expected big things from a number one pick are voting for a fictional award they created.

Also, people saying Drummond should win because guys have a harder time keeping up their numbers as their minutes increase need to take a close look at the Millsap doctrine. Starting more often and playing more seems to help a player's per minute rates, as possibly because they're more comfortable and have more time to get into a rhythm.

Real votes: Anthony Davis. Followed by: Dragic and Isaiah Thomas.

Anthony Davis has improved his ballhandling and scoring ability:


Coach of the Year
(aka most surprising team)

Gregg Popovich.

He's the best coach in the league, and I think he's doing the best job of coaching right now. It's that simple. Their defensive system is among the best, and so is their offensive system -- it's rare for a coach to juggle both. They're past 60 wins with homecourt advantage guaranteed now, and they've done that without anyone over 30 minutes. That's absurd. It's like Popovich made a weird bet with someone that he could get away with no one over 30 minutes.

Since the award is usually reserved for most surprising team, as apparently there's no possible way the credit for a team doing better than expected should go to the players, Hornacek may actually win this. I can't imagine anyone admitting he's a better coach, but, again, people make up their own rules.

Honorable mentions: Tom Thibodeau and Rick Carlisle.

The motion of Popovich's offense is beautiful:


Rookie of the Year
(aka rookie with the highest PPG+RPG+APG)

Nerlens Noel.

Simply by not participating in the worst rookie class since 2000, at least so far, Nerlens Noel is the default winner. Though for serious candidates, Oladipo and Carter-Williams are the typical stat-stuffing but inefficient scorers, but Trey Burke has helped Utah's offense from falling into a bottomless pit, Tim Hardaway Jr. has been one of New York's few bright spots and replaced some of JR Smith's shooting, Pero Antic is a poor man's Okur, Dieng is late to the party but is now threatening Pekovic's starting position, and Steven Adams has emerged as an alternative to Kendrick Perkins.

But there's one guy with a slightly better case than everyone: Mason Plumlee. When Brook Lopez went down as Brooklyn was already struggling, everyone wrote off the team, but they've been a very good team since the start of the new year and Plumlee, with Garnett playing low minutes, is part of the reason why. He's a great finisher near the rim, and he also protects it at the other end of the court. His brother blocked a huge dunk attempt from Anthony Davis in the all-star game, and Mason recently made a game winning block on LeBron James. He's rookie of the year for 2014 -- not the most inspiring class, but it might be a deep one.

Real votes: Mason Plumlee. Followed by: Deing and Trey Burke.

Plumlee wins the game with a block on LeBron:


Sixth Man of the Year
(aka the leading bench scorer)

The San Antonio Spurs.

Actually, there's nope now. Taj Gibson has become the leader for this award, despite most of his value coming from the defensive end. Last year I argued he deserved consideration, which was laughable to most people. With an improved post game, people are finally paying attention.

However, the Spurs have blitzed people with their bench mob lead by Manu Ginobili, who in limited minutes has been almost his old self. He shouldn't be ignored, but Taj edges him out due to a higher number of minutes -- he's first in the league among bench players in "pure" RAPM wins. Belinelli and Markieff Morris deserve honorable mention too. Then there's the Birdman rarely missing from the floor and leading some good defensive units for Miami.

Real votes: Taj Gibson. Followed by: Ginobili and Chris Andersen.

Manu's passing is electric:

Tuesday, April 1, 2014

A Response to the Value of Steals

An article on Nate Silver's new website FiveThirtyEight has been attracting a lot of attention, and it's apparently gone Hollywood and will have three sequels, one of which has debuted today (in Morris' defense, he did the same thing with his Rodman series.) The central thesis is basically saying that if a team had to choose which event happened, a point or a steal, you should choose the steal, but that's nothing new -- steals lead to a possession with a higher offensive rating and they deny the other team from scoring. It's the presentation of the results that bears scrutiny.

Today's followup article considered defense and gambling. My issue, however, was more with a residual analysis. Basically, the model output you usually see is the estimation of the average marginal value of a stat. But the value of a steal changes from player to player. There are players who are great at "picking your pocket" without sacrificing much on help defense like John Stockton, Blaylock, and Chris Paul. They have supremely fast hands, and get less of their steals from jumping passing lanes and going out of position on defense. By contrast, there are serial gamblers like DeMarcus Cousins, Westbrook, and Ellis who are not great defenders despite their gaudy steal totals. Cousins reaches too often instead of protecting the rim, and Westbrook and Ellis are too aggressive and can blow defensive schemes. Asserting that a player is good or great, as was done in the original article with Rubio, whose shooting has derailed the season, is dangerous. Not all steals are the same, and players collect them differently. It's an era with a wealth of video and stats, and we don't need to go back to this brand of basic analysis.

The benefits of the steal appear to outweigh the costs, on average, and this is nothing new. We do quibble on the exact value, and the now infamous article asserts that it's worth nine times as much as points. Continuing in backwards fashion, let's start with the end of the article concerning irreplacibility. This has been done before with a different method by his fellow FiveThirtyEight writer Neil Paine, finding that steals are average in responding to a change in role and a team's own steal rate also changes at an average rate. Though if the numbers in the FiveThirtyEight article are correct, that steals are almost perfectly irreplaceable, they're not entirely useful: a stat being irreplaceable doesn't necessarily mean it's valuable. DeAndre Jordan's FT% shooting is not replaceable in the sense that so few can match his awfulness, as only 1 other players in NBA history with at least 500 attempts have shot under 50%. After all, even if you can't replace the steal, it's not a huge loss since there's only a small correlation between team success and steals.

The body of the article concerned an empirical value of steals, and it opens by taking a swipe at the popular metric PER:
"The most famous attempt is John Hollinger’s player efficiency rating, which ostensibly includes steals in its calculation but values them about as much as two-point baskets."

Hollinger's metric is ancient in the NBA statistics movement, and it wasn't tested to provide the most powerful results. That's nothing new. The original model in the FiveThirtyEight piece appears to have used a regression model with points, rebounds, assists, blocks, and steals where the dependent variable (what they're trying to predict) is the effect on a team's point differential when a player is absent. It's termed "With-or-without you." This is actually even more basic than PER, which corrects for league averages, pace, and includes more stats, and without testing we don't know which one is more useful -- the article criticizes Hollinger for guessing statistic coefficients, but we don't see the predictive ability of "with-or-without you."

However, what's problematic is that we know little of what his data was actually like. I've actually used the same method before. In fact, my findings indicate that Rondo was overrated, even though he's a high steal rate player. I know that method's usefulness, but I also know how finicky and prone to random errors it is. It's essentially like raw plus/minus, probably a little better though, and is not the most robust method for testing the power of box score stats. I'm all for alternative means of testing from all angles, but if we have no way of testing and verifying the findings it's not practical and enlightening research.

As I had hoped from Nate Silver's website, I expected more transparency and, basically, more statistics -- we should see the results from the regression in the liner notes at the very least, to read over p-values, adjusted R's, F-statistics, degrees of freedom, etc. One important question I have is, what exactly do the data consist of? The article only devotes a single sentence to this issue:
"For this article, I’m using team game “with and without you” (WOWY) comparisons from all player seasons from 1986 to 2011 where a player missed and played at least 20 games."

Which players? Every player? What about the players who sometimes miss games because they're not very good and not needed? Was the line drawn at starters? How is it controlled for when a player misses the same set of games as another, or vice versa: a player goes down only when the superstar played that season?

The crux of the article, however, is on the "marginal" value of steals versus points. This is a pretty common tactic in modern stat-speak now, particularly with the Wins Produced clan and their "yay points!" thesis: deride the common folk for focusing too much on points per game. But with my background in engineering, I am hyper-vigilant when it comes to unit analysis. When I saw how steals were compared with points, I immediately saw a disconnect in juxtaposing steals and points. A player who averages one more steal a game than the average player is not the same situation as a player who averages one more point per game. There's a preferred method of comparing coefficients (at least in many circles) -- elasticity, an economics term for how a change affects the outcome (in this case, a team's point differential.) Typically, a median or mean value for the independent variable is chosen, and it's then increased by percentage points to illustrate how the change dictates the dependent variable.

In other words, we don't compare dissimilar units and instead look at a more realistic range. For the mean value, I simply used the season totals for 2014 and converted them to per 36 minute rates. This is in one sense the "average" player, and without a clear understanding of which players he used in the study it will be sufficient. The average scoring rate was 15.0 points per 36 minutes and 1.1 steals. Those aren't bad estimates, as doubling the rate gives you what are near typical leader results: 30 points and 2.2 steals. What happens when you apply the given coefficients to a ten percent increase? Well, points actually look more valuable. A ten-percent increase in the mean value of points per 36 minutes causes a greater positive effect on a team's scoring margin, per Morris' coefficients, than a ten-percent increase in steals per 36 minutes.


Coefficient
Mean value
Elasticity change at 10%
Points
1
15.0
1.50
Rebounds
1.7
6.4
1.08
Assists
2.2
3.3
0.72
Turnovers
5.4
2.2
1.18
Blocks
6.1
0.7
0.43
Steals
9.1
1.1
1.04
*Using median values from basketball-reference's summary page, 13.0 points and 1.0 steals, doesn't significantly change the results: 

Essentially, the main reason steals looked so valuable in the original article is that they are rarer, and there was a misapplication of unit analysis in comparing stats. Now, I'm not saying steals are unimportant and we should bow down to points. There are a lot of different effects interacting here, like how steals are an indication of awareness and athleticism. Morris cited a statistical plus/minus model for the veracity of steals, but that model and other similar ones have found the product of the usage (how often you shoot) and assist rate to be significant and important. That entangled with points to an extent, for just one example. But a lot of great players, like Olajuwon and David Robinson, have high steal rates, and it's true of some underrated players like Millsap and Thaddeus Young. There are certainly good reasons in valuing steals, but asserting that they are nine times valuable as points is misleading.