A few days ago David Biderman – of the Wall Street Journal – offered an examination of the history assists in the NBA (see The NBA’s Most Misleading Number). This article led me to think a bit more about assisted baskets and NBA history. And this is not just because I am quoted in the article.
Before I get to what I was quoted saying in this article, let me offer this interpretation of NBA history.
The National Basketball Association – via a merger of the National Basketball League and the Basketball Association of American — came into existence in 1949. If one looks back on that first season – via Basketball-Reference – one is struck by how little data was tracked. The numbers we have to track the productivity of a player from the 1949-50 season consists of points scored, shooting efficiency (from the field and the line), assists, and personal fouls. There is no data on rebounds, turnovers, steals, or blocked shots. So no data was tracked that would allow one to evaluated the impact a player had on gaining or maintaining possession of the ball.
Looking at the available data one wonders (at least I wonder) how it was decided to track these specific numbers. Certainly one suspects – as we think is true today – that scoring dominated player evaluation in 1949. So it’s understandable that people would record data on shooting efficiency and total points. And personal fouls had to be tracked because of the rules of the game.
But why track assists? If I was a reporter I might be inclined to track down someone familiar with the early history of the game. But I am not a reporter (and this is just a blog) so let me just offer some speculation based on simple economics.
Here are four basic propositions about the game of basketball in 1949.
1. Scoring dominates player evaluation. Again, this appears to be true today and given the data tracked in 1949 was likely true 50 years ago.
2. Without the three point line, shots closer to the basket – where shooting efficiencies are higher – are more valuable than shots further from the basket.
3. With a focus on scoring and getting shots closer to the basket, big men who could score inside are going to be valuable (think George Mikan).
4. Big men, though, lack the coordination to advance the ball across the court. In other words, big men need someone to get them the ball. Hence there is a need in basketball for little men or guards.
Guards, though, have an incentive problem. If players are rewarded for scoring, little men have less incentive to pass the ball. But if little men don’t pass the ball to the big men, the team will be less likely to win.
To solve this problem — I would argue — teams started tracking assists. In other words, teams essentially told guards that if they passed the ball to someone who scored, the guards would get rewarded for this behavior.
Now, one should note – especially in the era of the scoring point (see Isiah Thomas or Stephon Marbury, etc…) – guards still have an incentive to focus on scoring. But at least with assists tracked, guards have some incentive to pass the ball.
Let’s go back to the Biderman’s article. In this article he poses the following question:
Why are there more assists per field goals made now than 50 years ago?
Here is Biderman’s data:
Until the early 1970s, most teams were awarded assists on about half of the field goals they made in a given game. That number jumped to 60% by the end of that decade and has hovered around that level ever since. (Last season, the average team was given assists on 58.4% of their made field goals.)
David Biderman contacted me about six weeks ago to discuss this trend. As I have noted in the past, interviews with the media are like a test. Except you don’t always know the questions and you don’t get much of a chance to study. So my answer to Biderman’s question was essentially off the top of my head. And here is what Biderman reports I found on the top of my head:
Dave Berri, an associate professor of economics at Southern Utah University and lead author of “Wages of Wins,” a sports analysis book, says the uptick in assists is a function of players improving their shotmaking over the years. He says players in the 1950s and ’60s shot poorly, so more baskets came after offensive rebounds, where no assists are awarded.
It’s certainly true that players in the past shot very poorly. If we look back at the 1949-50 season we see the average field goal percentage was only 0.340. It was not until the 1959-60 season that this average surpassed the 0.400 mark. And it was not until the 1969-70 season that the league surpassed the 0.450 mark. With more missed shots it seems likely that in the 1950s and 1960s a made field goal was more likely to follow an offensive rebound, as opposed to a pass. Consequently fewer made shots had an assist.
To test this idea all we have to do is track how many shots followed an offensive rebound in the 1950s and 1960s. And then we can compare this result to what we see today.
Of course there are some problems with this approach. Offensive rebounds weren’t tracked until 1973-74. And play-by-play data is a very recent innovation. So it looks like we are out of luck.
Although we cannot do this study, I would love to see a different study of assists. Specifically, I would like to see what would happen if the NBA stopped tracking assists altogether. Well, that’s not really what I want. What I want is for the NBA to stop telling players about their assists. I suspect we would see passing in the game diminish. Without the assist in the box score, players should have less incentive to pass and therefore we should see less passing.
Unfortunately, I don’t think I am going to get that study either. I can say that the impact assists have on player evaluation appears minimal. At least, it’s not nearly the impact we see from scoring totals. So although I think assists are tracked to encourage passing, I think the incentive system in the NBA still encourages players to focus on scoring. And in that sense, I don’t think basketball has changed much in 50 years.
The WoW Journal Comments Policy
Our research on the NBA was summarized HERE.
Wins Produced, Win Score, and PAWSmin are also discussed in the following posts:
Finally, A Guide to Evaluating Models contains useful hints on how to interpret and evaluate statistical models.