Wednesday, July 17, 2013

Counters for daaayyys: Tanking is a Bad Strategy in the NBA

Is it inappropriate to use this picture in this context? Yes, almost certainly. 

The National Basketball Association is made up of 30 franchises, and the ultimate goal of every franchise is to win the NBA championship. The word "ultimate" is important, because it is readily apparent that NBA teams do not always act in the best interests of winning in the short-term. 

Occasionally, franchises appear to forgo fielding competitive teams in the short-run. This strategy, (colloquially referred to as "tanking") is employed because the NBA rewards failure. There are only three forms of player acquisition; via the draft, via trade or via free agency. The order of the draft is determined as an approximate inverse function of the winning percentage of each franchise from the previous season. Basically, if your team sucks, you are more likely to receive first dibs on draftees. If you land a top pick, you have the option of drafting from a bigger, and presumably, more talented pool of players. This incentive is quite appealing to many franchises, as some moderation of this "tanking" strategy is employed each year.

As far as I can tell, the community of NBA fans disagree on the validity of tanking. Some argue that it is an effective (or the only) way to acquire top-tier talent, while others claim that the strategy is almost universally foolish. The specifics of their arguments are laid out in detail in the aforementioned links, so I will just summarize their points below:

  1. You need superstars to compete in the NBA (both sides concur). Therefore, you need to acquire a superstar. The vast majority of players, including most superstars, enter the NBA via the draft. Therefore, maximizing your odds of landing a top-pick maximizes your odds of landing a superstar.
  2. Yes, players can be acquired via trade or free agency, but this is not viable because:
    1. Superstar players are very expensive and require loads of cap-space to sign
    2. Superstars almost always command maximum salaries, so teams with additional cap-space cannot offer a truly competitive price to incentivise these players.
      1. Therefore, they will decide based on other factors; ie: who they play with, geographical preferences
    3. Superstars are rarely traded, and command an exorbitantly high-price whenever they are traded.
  3. Drafting players allows teams to control their rights for the first four years of their career at a fixed price. Therefore, if these players produce at superstar levels during the duration of the contract, they offer significant value to franchises.
  1. The draft order is determined in part by winning percentage, but is also subject to a lottery. For example, the maximum probability of securing the top-pick is a mere 25%. This lottery introduces too much randomness and chance, thus rendering the tanking strategy to be too costly. 
  2. Productive players can be acquired via trade and free agency. Arguments to the contrary are based on assumptions that are pathos-oriented and not actualized in actual evidence.
    1. The draft is not the only way to acquire a superstar.

The "tanking" debate has waged on this offseason. Most scouting reports suggest that the upcoming NBA draft class will be loaded with future NBA superstars. Armed with these reports, the pro-tanking crowd has come out of the woodwork, and have laid siege to the anti-tanking community. The Raptors Republic community embodies one such battlefield. Recently, seemingly ubiquitous user "Tim W." wrote an article in favor of tanking, and presented counter-arguments to the foundational arguments of the anti-tanking crowd. I disagreed with many of his counter-arguments, so I've picked up a few shopping bags, and...nevermind. Here are my counter-counter-arguments. 

Faulty Assumptions
First, Tim offers some caveats and parameters about the way the NBA operates. I'd like to think of them as flawed assumptions because many of them are riddled with faulty logic.

"The NBA landscape is not equal; some franchises have inherent advantages"

This is partially true. Franchises are not homogeneous. However, how distinct each franchises is, as well as how much as how appealing these distinctions are, is unclear. Tim outlines two examples of inherent advantages; historical appeal, geographical advantage. 

He doesn't elaborate, but the assumption with historical appeal is that some franchises have "name-brand value" because they have been successful. Is this true? Perhaps. However, the supposed appeal appears to be inconsistent. Does Chicago have historical appeal? After all, the most celebrated player of all time played there, and the franchise won 6 championships. What about San Antonio? They have won 5 championships in the last 15 years. That's some impressive history. Yet, these franchises lack so called "historical appeal". Certainly, the idea of historical appeal is up for debate. 

Geographical advantage is far more tangible, at least in comparison to historical appeal. Cities are distinct; they offer a unique mixes of various civic factors, including climate, population distribution, unique attractions and most importantly, unique taxation rates. This point of taxation cannot be overlooked. Teams in Texas have the distinct advantage of having no state income tax, as compared to a state like New York, which has a 8.97% state tax rate on income. 

However, can we really generally quantify the summation of all these factors? I doubt it. Individual preferences are variable and unique. Tim presents the Heat, Knicks, Nets and Clippers are franchises that benefit from their geographical locations. This is presumably just a reflection of Tim's personal preferences for temperate weather and large cities (?). Each player has their own individual set of preferences. A player could be passionate about craft beers, and favor cities like Portland or Milwaukee. Or, a player may value tax rates and pick cities in Texas or Florida (zero state income tax). Who the hell knows? To each his own.

"Finding elite players outside of the top-five in the draft has become nearly impossible"

Tim presents a number of statistical studies which support his claim. Few players drafted outside the top-five picks become all-stars. Many Players in the top-five picks have become all-stars. To Tim, these statistics are "overwhelming". But what do these stats really mean?

I understand why the argument is overwhelming Tim. In his mind, being named to the all-star team, or being named to the all-NBA team is an accurate indication of productivity. Is this true? 

Being named to the all-star game is not always reflective of proficiency, or merit. Neither is being named to an All-NBA team

Tim is most likely mis-attributing the validity of these teams. In his mind, the logic is probably as follows:
  1. Lebron is an All-Star/All-NBA player
  2. Lebron is really good
  3. Therefore, players on the All-Star/All-NBA rosters are really good
  4. Carmelo is an All-Star/All-NBA player
  5. Carmelo is really good.
Either that, or he believes that the factors determining an All-Star/All-NBA title is accurate and reflective of productivity. Follow the following links to disspell the validity of All-Stars, and the All-NBA team

Two things. First, elite is determined by All-NBA team nomination, which is flawed. Second, no fucking shit. If you have (or believe to have) drafted an elite player, why the fuck would you give him up? Obviously elite players are likely to play for the team that drafted them; their teams had no reason to move them, and probably moved heaven and hell to keep them around. 

But sure, I see Tim's point. His point is that elite players are not easily acquired because they aren't usually available any other way. However, is that true? We'll elaborate on this point later on (hint: it's not).

Okay, now that we've tackled his faulty assumptions, let's move on to his counter-arguments.

There are numerous "studies" done that prove tanking doesn't work

Erm, sure. Let's ignore the shoddy grammar for a second. He espouses that these studies (1, 2, 3), are faulty because they don't account for non-statistical factors like historical and geographical advantages. I've already addressed those claims. 

What really irks me is that he doesn't actually address the aforementioned studies. He simply inserts his assumptions into the problem, and magically discounts the articulate arguments put forward by the studies. Wait, I think I can do that too.

Newton: "my model of gravity suggests that the Earth revolves around the Sun"

Me: "Nah, you're wrong because I believe that the Earth is the center of the universe and the Earth does not move, therefore the Earth does not revolve around the Sun"

Newton: " win?"

Come the fuck on, Tim. At least make an effort. 

Also, he specifically speaks to the Lakers and the Timberwolves (which is not addressed in any of the three studies he quoted). He suggests that the Lakers would never need to tank because free agents always come to them. That's just a silly claim; the Lakers routinely correctly identify productive players and acquire them because they have the pieces to do so. The opposite end of the spectrum is the Timberwolves, which fails to succeed because they are not good at identifying productive players and-often mis-manage their assets.

What if players don't pan out like expected before the draft?

Again, horrible grammar. He brushes this comment aside with the idea that you'll lose some, you'll win some. It doesn't mean that you shouldn't play. However, is that true? You need to know how often you will succeed versus how often you'll fail. This concern is very real; your high draft pick does not guarantee you a productive player. That risk is a cost, and a significant factor against the draft.

The player they draft could end up being a bust

His argument is that the higher the draft pick, the lower odds of landing a shitbag. He offers this hilarious study of top-five draft picks as evidence. Let's address that study.

The article's author readily admits that his "rating" statistic is overly simplistic. It is as follows

Rating = PPG + APG + RPG

This formula leaves out too many important factors, such as steals, blocks, turnovers, shooting efficiency, minutes played, etc etc. It also assumes that one PPG is equal to one APG and one RPG. Let's not even go there. 

How could you possibly determine anything with such a flawed instrument? Let's use reductio absurdum once more.

Me: "I dipped a tea-spoon into this pool of water and it was full, therefore I concluded that the pool is probably full"

Newton:" got me again!"

Also, let's address the self-fulfilling nature of that study. Teams obviously value their top picks, and would probably see fit to afford them the opportunities to perform. Therefore, when given more minutes, and more touches, static factors like PPG, APG and RPG should all increase. 

Teams need to win to attract free agents

So, wait. Earlier, he suggests that franchises like the Lakers and Celtics have an advantage because their franchises have been so successful, historically. He even substantiates his claim by suggesting that the Lakers don't need to tank because they can rest on their lorals. Now, winning doesn't matter? What?

Fine, let's put that double-standard aside. He suggests that winning doesn't attract free agents, otherwise the Spurs wouldn't have built through the draft. 

Did the Spurs build through the draft? Well, they won the lottery and scored Tim Duncan. Other players, such as Ginobili, Parker and Splitter have all come via the draft. They're an example of a successful "tank-job".

No they aren't.

Building through the draft is not the same as "tanking". The Spurs do two things well; they are excellent at identifying proficient players, and they manage their assets well. The year they landed Duncan, their superstar (David Robinson) was injured. That wasn't tanking. They also got really lucky and won the lottery that year. They correctly identified that Parker, Ginobili and Splitter were likely to become productive players, something other franchises failed to see. They drafted them with very late draft picks (late-late first round, two second-rounders).

"Oh, but they didn't sign any free agents, nyaaahhh"

Maybe they realized that free agency is more expensive than drafting? Drafting a player locks him into a rookie-scale contract. It also grants you an exclusive window to negotiate extension with the player. Free agency is subject to market conditions, and is often more expensive. Would you pay 24 million for Landry Fields when you can have Danny Green for the league minimum (the Raptors would)?

He then goes on to state "top-free agents usually want to play on a top NBA destination (fallacy), AND play on a contender." What?? So winning does matter with respect to acquiring players? What the fuck?

They have a better chance trying to trade for an elite player

Tim doesn't offer any real evidence to disprove the claim. He points to the Nets-Utah trade, where the Nets acquired Deron Williams in exchange for a bunch of draft picks and young players. He argues that top players command a ransom to acquire (except with Harden; how's Jeremy Lamb working out for you, OKC?), and thus it's a risky proposition. Also, players might get hurt or leave. 

Well, suddenly Tim's risk adverse.

The whole proposition of tanking is RISKY. VERY, VERY RISKY.

It's a matter of how much risk. With tanking, you need to A) win the lottery and score a "high draft pick", B) pick the right player. If you do all that correctly, you now face the same risks as having a "traded" player, in that C) the player needs to stay healthy, D) the player might leave. 

You know how you make a trade work? You need assets that appeal to your trading partner. What assets are appealing? Productive players, draft picks, cap space/expiring contracts, bargain contracts, and other elite players. 

Sure, trading for an elite player is risky, but it's a question of how risky, and specifically, how risky is it compared to tanking for players?

Creating a winning culture is important

Yeah, I agree with Tim here. This is stupid. However, winning games is not stupid. Win percentage positively impacts attendance and product purchases. Home playoff games are pure revenue for teams (players are not paid for playoff games). Teams do like making money, after all.

If you tank, the player will just end up leaving

So, you shouldn't tank when you already have an elite player? How can you even tank with an elite player, anyway? Teams have NEVER tanked with any of the following players on their roster; Lebron, Carmelo, Bosh, Paul and Howard.

Also, does anyone actually use this argument against tanking?

No teams that have tanked have won a title

This isn't even a valid argument to begin with, so I won't defend the credibility of the statement. However, I take issues with Tim's argument against the claim.

"Tanking isn't a strategy to build a championship team. It's simply a strategy to acquire an elite player"

So teams should pay a tonne of money to field a shitty team, forgo the revenue associated with winning, damage their brand credibility all for a wink from the luck gods to gift them an elite player, which isn't nearly enough to win a championship.

Come on.

Let's just move on. This is really draining.

I want to watch the Raptors in the playoffs/The Raptors aren't in a position to tank

So, Tim says that the Raptors probably won't be good enough to make the playoffs, yet he believes that DeRozan, Gay and Lowry all hold significant value. So...what?

First off, how much value could each player realistically hold? 

The Raptors acquired Lowry in exchange for a mid-first round draft pick when Lowry was signed to a very team-friendly contract. Last year he was very productive, but got hurt, and that team-friendly deal is quickly coming to an end. Given the hype around next year's draft, do you think a Lowry trade could break even for our beloved Raptors? I doubt it. I'd rather try to sign him to another team-friendly extension while his value is low (24 mill over 4 years, fourth year team option?). 

And DeRozan. He's polarizing. His talents are obvious; he can score via the jumpshot from inside the three-point arc, he can slash, he's great in transition and he can draw fouls. However, he's not very efficient. Neither his eFG% nor TS% has ever been above average for a shooting guard. He takes a lot of shots and is pretty average at everything else. 

Factor in that DeRozan is owed 40 million over the next four years and what do you get? Nothing, you get nothing in return for DeRozan. The market for inefficient wings who do little else other than take shots has shriveled up. After all, Colangelo is no longer a GM.

Then there's Rudy Gay. Sigh. At least Rudy Gay's contract is basically expiring after this season. He also shoots a lot for someone who isn't very efficient at it. However, fear not, Raptors fans. He's not exactly the same as DeRozan. Gay offers the same rate of inefficiency, but also throws in a whole bunch of turnovers! Now isn't that worth 20 million dolars a year? 

And again, consider what we gave up to acquire Gay. We gave up Ed Davis and Jose Calderon. They were very productive, but the Grizzlies clearly didn't agree. They turned Calderon into Tayshaun Prince and nailed Ed Davis to the bench. Do you really think Gay's stock has really skyrocketed during the half-season he spent with the Raptors? 

Secondly, believe it or not, this Raptors team is not that bad.

We have some good players. When healthy, Kyle Lowry is very productive (career .157 WP48, .190 last year). Amir Johnson is fantastic (career .206 WP48, .216 last year). Jonas Valanciunas was solid last season (.116 WP48) and will likely improve next year. Landry Fields is also very productive (.190 WP48 for his career, .154 last year). 

Look, I'm not saying the Raptors are going to take down the Heat and win the championship, but they have good pieces. They should challenge for a playoff spot. Projection systems have them winning 49 games next season

I know what you're thinking. "49 games? With this shitty bunch of players? No way". Ya-weh. We didn't really add any players of note this offseason, but we did manage to shed Andrea Bargnani. That alone will be worth 5+ wins.


Okay, so maybe tanking is a pretty silly idea. However, without tanking, how will teams acquire elite players? Well, do you need one concerted strategy?

Players can be acquired via the draft, free agency and via trade. You need cap space for free agency acquisitions, appealing assets for trading and picks for the draft. 

Of course, the common thought is that the above is all well and good if you want to acquire the DJ Augustin's of the world, but we need elite talent to contend for a championship. 

To solve this problem, we can turn to Michael Lewis's best-selling novel, Moneyball. Lewis chronicled the management practices of the Oakland Athletics, and how they were able to field such competitive teams for so little cost. The moral of the story was that the Athletics under Billy Beane and Paul DePodesta were able to correctly identify and acquire undervalued assets. 

At the time of publication (2002), baseball was still largely oblivious to sabermetrics, and more or less governed and operated based on the traditional stats (W, WHIP, RBI, R, AVG, etc). The stats told part of the story, but left out many key factors. The pervasive groupthink that governed baseball teams significantly undervalued on-base percentage. It turns out, if you get on base more often, you have a better shot at scoring runs. How 'bout it?

The Athletics leveraged their analytics advantage into numerous playoff appearances, and notably, the longest winning streak in MLB history. 

So how can the Raptors acquire undervalued assets? Well first, we need an accurate and objective method of quantifying a player's worth.

They could start with the Wins Produce model, created by Professor Dave Berri. Wins Produced is a catch-all boxscore statistic, which if you must, is akin to PER. The methodology is a tad complicated, and a full explanation can be found here, however the basic idea is as follows;
  1. Berri et al ran regression *complicated fancy math* and figured out how much each event in a basketball boxscore was worth, relative to points
    1. ie: a defensive rebound is worth .034 points
  2. These point values are known as weights. These weights are attached to each event, voila, you are able to condense the boxscore into one number
  3. Here's the kicker; the Wins Produced model explains 95% of wins on the team level. If you want a real-life example, I turn your attention to Wins Produced disciple Josh Weil, who made quite a shiny dollar betting on the Wins Produced model

So what makes Wins Produced more special than other advanced stats such as PER? Wins Produced is the best at explaining wins. After all, that's what we're here for, right?

Now, although I am very confident about this model, it alone is not enough. Wins Produced tells you the "what" and "how much" each player did, but does not tell you "how" he did it. What the model tells you is who is productive, and who isn't. You need to look at breakdowns (ie: shooting percentages, per 48 numbers, fouls, turnovers etc) to gain a better scope as to how a player contributes. 

Nevertheless, via Wins Produced, the Raptors can objectively determine a player's productivity. Armed with this knowledge, maybe the Raptors can exploit market inefficiencies like the Oakland Athletics.

Let's look at last season's leaders in Wins Produced:

Some names are obvious. Lebron, Durant and Paul are clearly three of the best players in the league. How were they acquired? Lebron via free agency, Durant via the draft, and Paul via trade. 

What about the rest of the names? 

I composed a breakdown of the top 60 players by Wins Produced last season. You can find the list here: This is what I found:
Out of the top 60, 55% came via the draft. That's got to be a vote for the tankers, yeah? Well, not really. Here's a further breakdown:
Wow. Only 8.3% of players (5) were drafted onto their teams with a top 5 draft pick! It certainly looks like you can get elite players by way out something other than a top 5 pick. Who knew?

Look, to be honest, this doesn't really say much, one way or the other. But it should at least dispell the notion that tanking is the only way to get a top player. That's certainly not the case, not by any objective measurements, at least. 

1 comment:

  1. That was a great article. The whole "geographical advantage" aka "toronto is too cold yo" grinds my gears.