OK, so you want to use advanced statistics to your advantage on the individual level. You understand that offensive rating is to the individual what offensive efficiency is to the team. However, there's still the contradiction that Kevin Durant's offensive rating of 116.5 pales in comparison to Matt Lawrence's mark of 123.5. There's no stat that would convince someone with even a cursory knowledge of this sport that Missouri's three-point shooting specialist had more game than the second pick in the 2007 NBA Draft.
Indeed, the utility of offensive rating requires context, and a large chunk of that context is how involved a player is in his team's offense. This is often described as a player's "usage," or the percentage of his team's possessions that he personally used. Players are credited with using possessions the same way a team is. Most of the time a used possession involves taking a shot that isn't rebounded by the offense, or committing turnovers. Adjustments are made for grabbing offensive rebounds and dishing out assists. In the end, usage is a great approximation for a player's involvement in the offense, and understanding this is the key to using advanced analytics to distinguish the star players from the role players.
Superstars combine high efficiency and high usage. Outstanding role players provide high efficiency without high usage. Durant is an example of the former. He used 31.6% of Texas' possessions while he was on the floor. With five players on the floor, an average player will use 20% of his team's possessions. Hoops being a game where the team dynamic is so important to success, most players are clustered around the average usage. Roughly half of all players have a usage between 17% and 23%. Only a select few get over 25%, and only 16 players in the nation had a larger role in their team's offense than Durant did.
On the other hand, Lawrence used just 14.8% of his team's possessions. By making 44% of his three-point shots, he was very effective in his role, limited as it was. Lawrence can't beat a defender off the dribble and he's not effective in traffic. He's a spot-up shooter, and when he gets open, he's very accurate. He doesn't get open often, and since Lawrence doesn't bring much else to the offensive table, his overall impact on his team was much smaller than Durant's was.
Understanding usage not only helps our statistical comprehension of what a player brings to his team's table, it can also be a big help in looking to the future. Remember, a superstar combines efficiency with usage. And if we're looking for future superstars, we need to find players who have the potential to do that. Speaking to the usage aspect, role players don't usually become go-to guys from one year to the next, or at any point during their careers.
Using data from the 2005, 2006 and 2007 seasons, it can be illustrated just how rare that phenomenon is:
Comparing a player's usage from one year to the next can be done by using the values on the horizontal axis as the starting point. The blue lines represent the 25th and 75th percentile values for the following season. So half of all players at a particular usage last season will find their usage fall between the blue lines this season. The thin black lines represent the 5th and 95th percentile values, thus 90% of all players will find themselves between those lines.
Take a player who used 15% of his team's possessions last season. This season, he has a 50/50 chance of using between 15 and 19% of his team's possessions, remaining firmly in role-player range. He has less than a 1-in-20 chance of exceeding 22%, meaning that a lot of things have to come together for the role player to become a star over the summer. Lest you think that it's much less rare for a player to make that transition over a career, here's how the same graph looks for all regulars in 2005 who also played in 2007.
The difference between the first chart and the second chart is barely noticeable. The envelope of possibilities is slightly bigger in year three than year two, but the message remains: Once a player demonstrates himself to be a role player, it's unlikely he'll ever be a go-to guy and, therefore, a superstar. It's not quite a law in college basketball, but players who are not very involved in the offense tend to stay that way. Any major changes in a player's usage are usually the result of filling the hole left by a departing possession eater.
Take the case of Josh McRoberts. McRoberts was projected as a lottery pick heading into his sophomore season, even after a freshman year in which his usage rate was around 17%. During his first season, he was playing with two offensive superstars, J.J. Redick and Shelden Williams, who weren't around for McRoberts' sophomore campaign. So it was plausible that McRoberts would see a larger than normal increase in his usage. Let's take a look at what would be considered a normal increase.
Year 1: 17.1% Expectation
75th percentile 20.0
95th percentile 23.3
Even without knowing the potential that scouts saw in McRoberts, just knowing his situation, that his team lost two high-usage players from the previous season, we could reasonably expect him to push that 95th percentile mark. After all, this was a situation where a lot of extra possessions would have to fall on the returning players' shoulders. Even using 23% of his team's possessions, however, would not have given McRoberts enough scoring opportunities to be considered for national or even conference player of the year honors. In fact, Josh McRoberts ended up with a usage rate of 21.9%, and continued to receive criticism for not taking over games as the Duke offense struggled.
Like any aspect of statistical analysis, understanding usage is one piece of a big puzzle. It should be stressed that this tool strictly applies to a player's offensive contributions. Players do jump from being decoys to go-to guys in one season, and some even regress the other way. Those are the exceptions. By and large, a player's role on his team in one season is a good indicator of his role the following season. It's largely on that principle that the team previews are based.
Ken Pomeroy is an author of Basketball Prospectus.
You can contact Ken by clicking here or click here to see Ken's other articles.