Scoring Versatility Index

[steveshea]

Based on three new player tracking categories: Drives, Pull Up Shots and Catch and Shoot (now available at http://stats.nba.com/playerTracking.html), we define a players Scoring Versatility Index (SVI). It measures how balanced across the three scoring categories a player's scoring has been. For more details, see http://www.basketballanalyticsbook.com/ ... index-svi/.

Here are the top 20 so far this season:

[Crow]

I'd be interested to see how efficient these "versatile" guys are for each type shot. There is good and bad versatility. Some can truly do it all, others just think they can.

If one grouped the league into 3-5 bins based on versatility score and looked at the shooting efficiency of the bins which are on average more efficient with their share of the shots? Do good teams (on offense or overall) tend to have more highly versatile guys, the same or less than average? Do teams with more do better in the playoffs than those who are not so high? How many do you need?
If you characterize starting lineups as a 5 man string of above average versatile players or specialists in one area how are the lineup types distributed? What is the most common and best performing mix?

[Mike G]

Rubio is equally bad on all types of shot, and LeBron is equally good on all. They're equally versatile?

[steveshea]

SVI is only measuring versatility in total points. It does not attempt to measure efficiency. I absolutely believe efficiency is important. I believe there is value in versatility, but do not misinterpret SVI. I do not believe that a higher SVI necessarily means a better offensive player. SVI is just one small piece of the puzzle.

I think Crow got it right, some guys are versatile and some think they are. In fact, it seems to me that the players on the top 20 from the better teams tend to be the players that are actually efficient scorers in all three categories. If we can get a measure that combines efficiency and versatility, we could run a study in this direction (which I believe is along the lines of what Crow was thinking).

It we want to look at versatility and efficiency, how best to do that is an interesting question. Of course, we can put a measure of efficiency at each of the scoring categories next to the percentage of points scored in those categories. But I would prefer to have one number that incorporates efficiency and versatility-an Efficient Scoring Versatility Index (ESVI) perhaps. I am open to suggestions.

[Mike G]

SVI is only measuring versatility in total points. It does not attempt to measure efficiency.

Right. So if LeBron is averaging 27 pts, that's about 9 from each shot 'type'.
If Rubio is averaging 9 pts, that's 3 of each. Disregarding efficiency, they're still not equally Versatile in the normal sense of the word. LeBron is proficient at all 3, and Rubio is proficient nowhere.

Why not just use Points of each type, rather than % of a player's points?

[steveshea]

SVI is only measuring versatility in total points. It does not attempt to measure efficiency.

Right. So if LeBron is averaging 27 pts, that's about 9 from each shot 'type'.
If Rubio is averaging 9 pts, that's 3 of each. Disregarding efficiency, they're still not equally Versatile in the normal sense of the word. LeBron is proficient at all 3, and Rubio is proficient nowhere.

Why not just use Points of each type, rather than % of a player's points?

I think this is a third component. For a given player, we can look at their total production, how efficient they are, and how versatile they are. I guess the term versatility could imply some level of efficiency or total production. In the sense that if I say player A is more versatile than player B, we wouldn't expect player A to score 9 points a game and player B to score 20. Maybe I should use the term "balance" in scoring instead.

Anyways, I am interested in finding measures that incorporate all three characteristics (versatility, efficiency and quantity) in scoring. As the formula for SVI is currently constructed, I don't know that plugging in points as opposed to % of points works. That doesn't mean we can't find a different formula where we can use points. We know that players can score in quantity and not be efficient, so I don't think this will take care of the efficiency concern.

Thanks for the suggestions. I'll keep working on it.

[Bobbofitos]

Why does versatility matter?

[steveshea]

Why does versatility matter?

I don't know yet. Does versatility matter? We need the appropriate data, and then an intelligent metric on the data before we can run a serious study. RIght now, the data available to the public is limited, and the metric design is still in week 1. But I'm getting good feeback on the boards.

I suspect that, for example, a lineup of players that can only score while driving would not be so successful in today's NBA.

[Bobbofitos]

Why does versatility matter?

I don't know yet. Does versatility matter? We need the appropriate data, and then an intelligent metric on the data before we can run a serious study. RIght now, the data available to the public is limited, and the metric design is still in week 1. But I'm getting good feeback on the boards.

I suspect that, for example, a lineup of players that can only score while driving would not be so successful in today's NBA.

Would like to see polarized versatility lineups (try formatting 5man lineups with expected scoring type vs actual) to see how efficiency goes up or down (or, unchanged!) given versatility. Would gain some insight into lineup construction.

Also, is this broken down by shots attempted or made in each of these cats? I feel as though that changes versatility somewhat.

[steveshea]

Would like to see polarized versatility lineups (try formatting 5man lineups with expected scoring type vs actual) to see how efficiency goes up or down (or, unchanged!) given versatility. Would gain some insight into lineup construction.

Also, is this broken down by shots attempted or made in each of these cats? I feel as though that changes versatility somewhat.

It is points, although it is easy to fit the formula to attempts. The problem is that the drives data on nba.com is incomplete. They provide total # of drives, but they are including drives that don't result in that individual's shot.

I am planning to look at versatility's impact on lineup performance. I know player versatility and I know total lineup production, but I don't know how versatile a player was with one lineup as opposed to another. Also, I don't know the collective versatility of the lineup. Is it reasonable to assume that a player's versatility remains fairly constant across different lineups?

[Crow]

I'd think it might be possible these days to check how versatile a player was with one lineup as opposed to another using nbawowy.com but I haven't drilled down and check the shot information directly.

Some players' versatility will remain fairly constant across different lineups while others might not because of a choice or need to change behavior in the absence and presence of certain other players or a coaching assignment to do so or at least try to. Checking relative versatility in different lineups would be useful for understanding what is happening and what the other shot distribution options might be for particular players and the lineup as a whole given their talents and the possible desire to show a versatile, harder to predict and handle offense to the defense. So much information is possible to produce if decision-makers want to see and use it.

[steveshea]

I'd think it might be possible these days to check how versatile a player was with one lineup as opposed to another using nbawowy.com but I haven't drilled down and check the shot information directly.

Thanks, I'll look into what I can do with nbawowy.com

Also, we have posted a new blog on the Scoring Versatility Index (SVI). See http://www.basketballanalyticsbook.com/ ... ex-svi-ii/. There, we plot SVI against EFG%. No surprise that LeBron was the most efficient of the most versatile. Felton and Farmar rate high in versatility, but very low in efficiency. Jennings hasn't been versatile or efficient.

We also discuss some of the limitations at this point (in available data and the current construction of the metric).

[Bobbofitos]

Also, I don't know the collective versatility of the lineup. Is it reasonable to assume that a player's versatility remains fairly constant across different lineups?

I would assume this is not the case, but I suppose that's why it's worth looking into

[steveshea]

Also, I don't know the collective versatility of the lineup. Is it reasonable to assume that a player's versatility remains fairly constant across different lineups?

I would assume this is not the case, but I suppose that's why it's worth looking into

I agree. Unfortunately, it doesn't appear that the data (the breakdown of an individual's pull up shots, catch and shoots, and drives according to lineup) is publicly available.

[Crow]

Yeah, I guess so. But there is distance right? While part of versatility is how you go to the shot, there is also versatility based on distance of shot too and I am actually more interested in that. The two versions are somewhat close- i.e. drives are often short shots and more pull=ups are probably mid-range while threes are probably more often catch n shoot (I'd think).

Maybe one needs several publicly available databases or the coming next year vanatgesports offering. The idea of volunteer built additions and bridges between databases sounds nice but I am not super hopeful about them.


I will read the new article a bit later.