clock menu more-arrow no yes

Filed under:

Austin Rivers, Bradley Beal, and the problem of comparing players who fill different roles

New, comments

Coming into the 2011-2012 college basketball season, no incoming freshman guard had received more media attention than Austin Rivers. In Thursday's NBA draft, fellow freshman guard Bradley Beal was drafted ahead of Rivers. Beal was also highly regarded coming out of high school, so it isn't particularly surprising that he was drafted before Rivers.

As a statgeek, I should probably prefer Beal to Rivers. Focusing just on the scoring contributions of these two players, Beal was clearly more efficient. Both players had similar usage ratings, meaning they had the ball at the end of possessions about the same fraction of the time while on the court. Rivers' used 25% of the possessions while on the court, while Beal used 23%. Beal's offensive rating, no matter how you chose to measure it, was substantially better than Rivers' rating. For simplicity, we can look at how efficiently the two players scored, using true shooting percentage TS%. Beal's was 0.575 and Rivers' was 0.538. Beal was a more efficient scorer than Rivers. Case closed. Both guys are being drafted primarily for their potential as scorers, so Beal should go first. The NBA got it right.

Or did they? I am not going to argue that the NBA got this evaluation wrong. I am just going to argue that things are a bit more complicated than what you get from a cursory look at advanced metrics, and that from my perspective Beal and Rivers have statistical records that suggest more or less similar capabilities as scorers. Follow along after the jump.

I believe in the power of measuring things. Measurement is great. In this context, it allows us to compare Beal and Rivers, and a really simple tool like true shooting percentage tells us Beal is the better scorer. Other measures such as effective field goal percentage agree. Beal's eFG is 0.525, and Rivers' is 0.506.

But I also believe that measurement is hard, and projection from those measurements is even harder. The goal is projection when it comes to the NBA draft. We want to try to figure out which player will be better in the future. One significant factor that led to Beal's better TS% was his better free throw shooting. Rivers shot 66%, while Beal shot 77%. Austin Rivers is just too good of a shooter to expect that his free throw shooting won't improve in the future. He could very easily close this particular gap before next season. Give Rivers 75% from the line (a shooting percentage that he should be able to achieve at some point), and his TS% climbs to 0.556. The gap between these two players remains, but it has gotten smaller.

The free throw issue is easy to determine, but the next one is a little bit harder. When we want to conclude and predict things based on measurements, it is very important that we understand what we have measured. In this case, TS% and eFG% reflect more than just a player's abilities. They also reflect the way in which a player is used within a particular offensive system. When using these numbers to project things, we need to remember this. Ideally, we would correct eFG% and TS% for system dependent stuff, but this is probably not possible. Alternatively, it is a good idea to at least try to understand the system dependent stuff, and try to guess at how large of an effect these things will have on the statistics of Beal and Rivers.

The Florida and Duke offenses are similar in many ways. Both rely heavily on ball screens, and both result in a lot of three point attempts. Last season, 45% of Florida's attempts were from three point range, while 39% of Duke's shots were three point attempts. But Beal and Rivers' roles in these offenses were quite different.

The first place where these differences show up is in the shot distribution of each player. Using play-by-play data, I like to track where different players get their shots. Shots fall into three categories: shots at the rim (layups and dunks), two point jump shots, and three point jump shots. Generally, your best bet to score efficiently is to get to the rim. Differences in shot attempts and FG% at the rim don't really account for the differences between Rivers and Beal. Rivers took 31% of his attempts at the rim, and Beal took 34% of his field goal attempts at the rim. (Note that because of a weird naming inconsistency, Beal's stats are split up into two separate lines in the link.) This is not very much of a difference, considering the small samples involved. Both players converted over 60% (Rivers is at 62%, while Beal is at 65%) of these shots, which is a pretty typical value.

Where the differences start to show up is when we look at the breakdown of shots that each player took as two point jump shots and as three point jump shots. Beal took a much higher percentage of his shots from three point range. He took 46% from three point range, and 19% as two point jump shots. Rivers took two point jump shots more frequently, and a lower percentage of his shots are from three point range. 29% of Rivers' shots were two point jump shots and 40% were from three. Again, both players converted very similar percentages of each shot type (Rivers' percentages were a little bit higher), but three point shots are worth more than two point shots.

The difference in eFG% between Rivers and Beal is primarily due to this shot distribution difference. If Rivers had the shot distribution of Beal, so heavy in efficient threes and light on mid-range jump shots, his eFG% would have been 0.527. Recall that Beal's eFG% was 0.525.

But now we are faced with another tricky issue. It doesn't seem right to just wave the magic wand and grant Rivers the same highly efficient shot distribution of Beal. Some of this difference is due to each player's role in their respective offense, which they don't really control. But some of it could also be due to shot selection. When we look at shot location distributions, getting to the rim is clearly a skill, and Beal took a somewhat higher fraction of his shots at the rim. But how much is two point jump shot vs. three point jump shot distribution a skill, and how much is related to the team's offensive style and the player's role?

I can't answer this question, but I can point to an oddity in the statistical record of Austin Rivers that will provide a hint. Generally, a very high percentage of made three point shots are assisted. This shouldn't be much of a surprise; so many of the shots players take from three come in catch-and-shoot situations. It is very common for 90% of a player's made three point shots to come off of an assist. 90% of Beal's made three point shots were assisted. Sometimes, players who handle the ball a lot will have a lower number, but it is quite rare for the assist rate on three point shots for a particularly player to be much lower than 70-80%.

For Austin Rivers, 41% of his made three point shots were assisted. I have not systematically searched through the data, but this is the lowest number of any player that I am aware of for the 2011-2012 season. Looking further, 0% of River's two point attempts were assisted! I don't like to use exclamation marks - it is my belief that we are each given four or five to use in our lifetime, and when they are gone we die - but I think I will use one there. Additionally we find that 13% of Rivers' made baskets at the rim were assisted. For Beal, 32% of his made two point jump shots were assisted and 38% of his made baskets at the rim came off of assists.

Looking through these data, few of Rivers' points came on catch-and-shoot opportunities. 14% of Rivers' points came on catch-and-shoot threes, whereas 31% of Beal's points were off of catch-and-shoot threes. Rivers had to create much more of his own offense than Beal did; this reflects the differences in their roles. From watching both Duke and Florida play, Rivers clearly played with the ball in his hands more than Beal did.

None of this is meant to diminish Beal, who is a really good player. One of the things that impressed me most about him was how well he adapted to what he was asked to do. He is a 6-4 guard, and he had a defensive rebounding percentage of 18%. His team played a very small lineup much of the time, and they needed him to get those rebounds.

But Rivers is a good player, too. Don't be fooled by his lower effective field goal percentage.