“All right, what do we think? What do we know? What can we prove?” –And the Band Played On
Loving it! So as a GM would you agree with the following strategy. Find a superstar player, position doesn’t matter but age does. Now try and fill your roster “around” the player is the order of Center, Point Guard, Power Forward, Small Forward, Shooting guard. Also try and add depth to your roster in the following order. Would such a strategy work as a general heuristic? Last but not least hope the league isn’t super awesome your year (sorry Utah!).
This is not the first time something like this has been asked and it made me think of writing down a simple algorithm of how I would build and run an NBA team for success. So what follows is the “Build me a Winner Algorithm” for the NBA take 1.
This article uses Wins Produced and WP48 [Wins Produced per 48 minutes] to evaluate player’s performance.* This measure uses three key components to evaluate a player:
- The player’s per minute box score statistics
- The player’s team’s per minute box score statistics
- The average performance at the player’s position (PG, SG, SF, PF or C)
A full explanation can be found here. To give a general scale, an average player has a WP48 score of 0.100. The very best players in the league usually have a WP48 over 0.300. To put this in perspective; an average player who plays a full season at 40 minutes a game would generate around 6.83 wins for their team. In contrast, a player posting a 0.300 WP48 would generate about 20.5 wins at 40 minutes a game over an 82 game season.
Let’s review first: what do we think? What do we know? What can we prove?
- Player value in the NBA is skewed towards points and not possessions stats. Over time this leads to a weak correlation between wins and money spent.Teams are in fact using the wrong stats to evaluate players. There are thus market inefficiencies to exploit. Stealing from my article on the Short Supply of tall People.
- Basketball more than any other sport is a sport about marginal value.
- Wins are a direct result of the marginal absolute productivity of the players on the court as measured in point differential (margin of victory).
- Wins produced uses regression to build a causal model for wins based on the statistics available in the standard boxscore.
- There are multiple factors and contributions (let’s call this player productivity) that go into scoring a point and the boxscore reflects a significant portion of these factors.
- Wins are a function of Point differential
- Point differential is a function of Player Productivity as measure in the boxscore stats and actually it’s a function of marginal player productivity (i.e. how much better your player’s on the court are than your opponent’s )
- Wins can thus be modeled as a function marginal player productivity
- Wins Produced uses regression to build that model and can be shown through correlation to be successful.
- The Short Supply of Tall people. Big Men (F/C) are on average more productive than everyone else . They in fact account for 50% of all productivity. This makes it harder for a center to be better than the average and thus accumulate wins in our model but this is not out of step with the reality of the situation. Teams also have a lot more at risk with their big men. It’s also much easier for a team to accumulate negative value at center and power forward because there is much more at risk.
- The Short Supply of Ball Handlers. Average Center and Point Guards have over time been much more valuable to teams than any of the other positions. Over the last 5 years the difference between an average center and a replacement level one is 4 more wins than the same at shooting guard (and 2 at Point Guard). The short supply of tall people is really not a surprise however the short supply of ball handlers is.
- Peaks for the players are skewing older over time (in fact the data is deceptive because it includes active players who may have not hit their peaks yet). In fact if I look at players born since 1970.You see the most players (32 out of 178) hitting their peaks at 28. What does this mean? It means that if you’re a GM signing a guy coming off his rookie contract (say 24 or 25), You can reasonably expect equal or improved performance over the course of a 5 year contract (thereby justifying a % increase from the base Year). However if your big Free Agent signee is 29 or over? You’re probably out of luck .
- The best individual seasons generally come from players playing with the team that drafted them (or the #$$%@%@ Lakers)
- The draft is not a place for quick fixes. Impact rookies are a rare breed. There have been 330 rookies selected in the top 10 since 1977and less than 15% of these rookies – who were generally considered “hot prospects” – have made substantial impact (>8 wins) his rookie season (and only 13 of the 33 players chosen with the first pick). If we look at the top 25 draft picks ever, the average pick of the top 25 is 12.24. Only 3 were the top pick (Magic, Robinson & Shaq) and only seven were in the top 3 picks (and none at number 2). This, and the fact that 8 of the top 25 were picked at 20 or later, strongly suggests the league in general is not very skilled at pinpointing incoming talent. The probability of getting no value or negative value from a draft pick fluctuates around 30% , however Impact players (Superstars & All time greats) are coming into the league at an increasing rate. This would help explain the fact that the quality of basketball seems to be at it’s highest levels in recent years (see here).
- The number 1 pick can and has been flubbed massively. 10 of the 30 #1 Picks fall in the second half of the rankings. The list includes some old WoW friends:
- Mark Aguirre
- Allen Iverson
- Kenyon Martin
- Glenn Robinson
- Kwame Brown
- Joe Barry Carroll
- Joe Smith
- Kent Benson
- Michael Olowokandi
- Andrea Bargnani
- Drafting Players under 20 is an extremely dangerous game. Only 13 of the top 200 players were aged 19 or younger at the end of their first NBA season
- Dwight Howard
- Tracy McGrady
- Kevin Garnett
- LeBron James
- Andris Biedrins
- Luol Deng
- Tyson Chandler
- Josh Smith
- Rashard Lewis
- Chris Bosh
- Kobe Bryant
- Cliff Robinson
- Andrew Bynum
- The Half baked notion that what wins in the regular season is not necessarily what gets you the trophy. The difference? Minute allocation & how wins produced are affected by that allocation. We continuously hear terms like playoff rotation & playoff minutes thrown around come playoff time. The half baked notion tells us that a good deep team filled with average and above average players will get you in the playoffs but to get far in the playoffs you need your wins to be concentrated in your Top 6.
- In the Regular Season:
- Your starting five account for 82% percent of your wins.
- Your second unit is important over the course of an 82 game regular season accounting for 18% of your wins
- After that everybody else is statistically meaningless.
- In the Playoffs :
- Your starting five account for 94% percent of your wins in the playoffs.
- Only the first guy of your bench matters accounting for 5% of your wins
- After that everybody else is statistically meaningless.
- In the Regular Season:
Build me a Winner
So Based on this knowledge, let’s try to summarize what my management philosophy would be as an NBA GM.
- A wins produced model (such as Prof. Berri’s Wins Produced or my own Wins over replacement Player (WORP)) gives a team a statistical edge over other teams in building a roster by properly identifying a player win contribution with a high level of correlation. My team would be built around just such a model (probably with enhancements for individual defense which as a gm I could design and pay somebody to data enter) and identifying underrated/underpriced players that are available.
- I’d use picks rather than free agents to keep my team successful both on the court and in the bottom line (See San Antonio and Oklahoma City). The draft is, the best source of cheap labor there is. When dealing with draft picks it is important to remember that you are getting a low cost player for four and not one year and any evaluation of draft picks should go beyond the rookie year. Of the top 100 picks half (and 122 of the top 200 picks) were taken after pick 9 suggesting there is always value in the later part of the draft. The draft is a high stakes lottery but it’s a rigged game for the owners. Salaries are fixed at a discount and the risk of utter failure is relatively low (30%). So we build thru the draft and not thru free-agency.
- Scorers are common and overpaid. I’d flip scorers for picks, picks and more picks see the previous point on draft picks.
- I’d prefer later picks in volume over high end lottery picks. Stars can and are had late in the draft. Picks would be used exclusively on high risk/high reward guys. You don’t play the lottery to win third prize. Second rounders would be used mostly on Euro guys that I could stash and see if they’re any good.
- Hate overpriced free agents but love minimum salary level players. We know that the talent identification algorithm for NBA teams is broken. That means good talent must be available out there and I’d spend money to find it. So I’d buy one (or possibly two) D-league teams to stash, test and develop talent. I’d look into buying a Euroleague team as well. I’d also be the the king of 10 day contracts and call ups. Roster spots 10 and up would be used at least 75% of the time for auditions and talent evaluation.
- Given that the top 6 is what matters for the playoffs. I’d trade three .100 WP48 guys for one .250 WP48 guy in a hummingbird’s heartbeat.
- Big Men and Ball handlers (Centers and Point Guards) are more scarce resources than shooters. I’d pay for skilled labor at those positions and always focus on depth there. This is the central idea behind my own Wins over replacement Player (WORP).
As I said this is a first take on this algorithm and I will probably revisit this over and over in the future. Please provide feedback and comments . As always this is only my opinion and I am the first to admit it could be flawed.You never know, If I get to be an NBA GM, I might hire you 🙂