APBRmetrics

[drebelx]

Hi All. Recently I was introduced to the world of basketball. I'm mainly a fan of baseball and of the sabermetric revolution that has been occurring there. It's very exiting to see that basketball statistics are starting to be analyzed and improved.

I posted some of my thoughts about basketball on my friend's fantasy basketball forum on the subject of how to determine the best basketball players for a real life NBA team using and manipulating the stats that are given to us. I brought on the idea that Assists were not a good number to use at all for this. I caught a lot of trouble from them because of this.

Granted that I have not watched or played much basketball so far, my thinking was this... Assists are just like RBI's in baseball. RBI's are awarded to the hitter if they create a run. The trick to the RBI is that there must be runners on the bases to get those runs (beside the homerun). In the sabermetric world, a hit that generates an RBI is no more valuable than the hit that doesn't. The thinking goes so long as the batter did not create an out, it's a good thing

Getting an Assist seems like it is akin to the RBI.

A basketball player is awarded with an assist if the next guy they pass to makes a basket. What difference does it make that the guy they pass to scores? Shouldn't the thinking be more like in baseball where so long as the passer did not turn the ball over and keeps the possession of the ball on the team, it's a good thing.

One of the arguments I heard from the league was that it takes skill to find the player who is open and can take a shot. Since I haven't watched or played many games myself, I can't attest to this notion.

I wonder what the great minds of basketball think about assists?

[bballfan72031]

Hello drebelx. Though I'm not much of a poster here, I would like to say welcome to the board.


That is the exact same thought I had about assists. For example....

Sometimes Player A will make an easy pass (single), to Player B, who then makes an excellent pass (double--RBI) to Player C, who easily dunks it (single--run).

Other times Player A will make an excellent pass (double) to Player B, who then makes an easy pass (single--RBI) to Player C, who easily dunks it (single--run).

And sometimes Player C will just get the ball and juke everybody for the score (like a homerun).


Is that kind of what you're saying, drebelx?


Anyway, I'm just glad that someone else has the same thinking as I do.

[drebelx]

sportfan, thanks for the welcome.

You nailed my thought. That is exactly my point. If you can eliminate the notion that it takes skill to pass to someone who immediately takes a shot, a lot of things start to fall in place, in line with how baseball is studied. Possessions became the equivalent to At Bats. Passes become the equivalent to Hits. Enough passes (hits) will eventually lead to a basket (run) so long as you don't turnover or miss the basket (out).

It is also interesting to think that in baseball the batting order does not change 1 to 9, but in basketball we can think about it as if a hitter can bat 1st, then 3rd and then again 8th.

Also intersting is that a run (basket) ends the inning (team possession)

So does this idea of Assists being equivalent to RBI's have any merrit?

[Kevin Pelton]

I don't think the RBI comparison is particularly apt, though there certainly is an opportunity element to assists. The RBI is a method of measuring hitting, as are batting average and OPS etc. It's just not as good, because of the situational aspect.

Assists, on the other hand, do not measure a player's ability to score, but instead an entirely different skill, passing.

I've been racking my brain a bit, and I think this is the best baseball comparison I can make with assists ... in baseball, the goal is to score runs. There are two primary ways to do this -- by getting on base and by advancing runners. So we have two different metrics to measure this -- on-base percentage and slugging percentage.

In basketball, the goal is to score baskets. Though this can only be done by making a shot (just as runs are scored virtually only on hits), good passes improve your chance of scoring, as do good picks and floor spacing, etc. (things that can't be precisely measured currently).

So to me, assists and points are like on-base percentage and slugging percentage in that they are the two key components in creating points (or runs).

(DeanO would take me aside here to hammer home that offensive rebounds are offense too and completely measurable, but let's keep things simple for now.)

[Dan Rosenbaum]

Assists are nothing like RBIs. The reason that sabermetricians put very little value on RBIs is that the hits, walks, etc. that generate RBIs are already counted for the player generating the RBI. With a handful of exceptions, putting a lot of weight on RBIs would be double counting.

Assists, unlike RBIs, are not already counted as some other statistic. There is no single, double, triple, homer, or walk that is already counted for the player making the assist. So the comparison of assists to RBIs is completely off base.

Passes are also not really like hits. Four base hits necessarily generate a run (as long as there is no double play). Four passes do not necessarily get a team any closer to scoring than zero passes.

More importantly, assists capture an element of basketball that does not really have a baseball corollary - teamwork. Baseball is a series of repeated one on one contests with a smattering of teamwork once in awhile. Basketball is fundamentally a team game, and assists are one of the statistics that help us capture this teamwork.

[drebelx]

First I have to say it is an honor to enter a discussion with you Dan. It's also amazing that I'm discussing with the Admin as well.

There is an opportunistic situation for RBI. The batter can only get RBI's if there are men on base (besides the homerun). Manny Ramirez hits a double with a man on third and is awarded an RBI. Next time Manny Ramirez hits a double with noone on and does not get an RBI. To gain an RBI it depends on what the previous hitter have done, not Manny's skills as a hitter. RBI's are a poor measure of the hitter's skills and abilities. My thought with basketball is that Assists are a poor measure of skill because too much of it depends on the shooter's ability and situation. Luke Ridnour passes to Ray Allen, Allen takes a shot and misses. Ridnour does absolutely nothing different the next time he passes to Ray who then takes a shot at the same spot on the floor with the same defensive arrangement and makes the basket this time. Ridnour is awarded with an Assist. Does anyone else see the fallacy in this? Seems to me that the key is passing not assists.

I agree with the notion that in baseball the goal is to score runs and to do that you need to get on base. In basketball the goal is to make baskets and to do that you need to pass. Absolutely.

I don't think Points and Assists correlate very will with on base percentage and slugging percentage considering the former are absolute numbers that are counted and the latter are percentages. If anything something like eFG% could potentially correlate with slugging. A new number has to be created to be equivalent to on base percentage along the lines of passing and possession.

Teamwork in baseball is very different than in basketball, that is for sure. We should try to correlate somethings about the two sports anyway. In baseball the teamwork there is to not get out, give the next hitter a chance to do something. In basketball to me it seems that correlation there would be to not turn over the ball, give someone else on the team the chance to do some damage.

Am I off base thinking that an Assist is just a glorified pass?

[jeffpotts77]

Drebelx,

I think that assists are clearly somewhat dependant on the shooter as you illustrated with your Ray Allen example, but if the ability of the passer had nothing to do with the likelihood of the basket, then it would mean that if we replace Luke Ridnour with Steve Nash in the Sonics lineup, Nash's Assist rate would then drop down to Ridnour territory. I don't think any of us believe that would happen, so this suggests that there is some skill associated with finding someone in a position to score.

[Mike G]

You might try reading some recent threads, as I think we just went over this (without these analogies to baseball).

There are good passes, safe passes, great passes, useless passes, etc. Some lead to high-% shots, while others lead to shot-clock violations or turnovers. Just making generic "passes" doesn't produce anything.

If you make 10 useless passes just to get rid of the ball, you might get one assist, or none. If you make 10 effective passes to higher-efficiency scorers/situations, you might get 6-8 assists. Over the course of a season, effective passers get lots of assists (and relatively few turnovers) and are easily distinguished from ineffective passers.

[drebelx]

Jeff,

The Nash for Ridnour senario is interesting, but the idea of Nash getting more assists than Ridnour could easily be explained by the idea that Nash is a better passer in general than Ridnour. If I were to replace Nomar Garciaparra with Alex Rodriguez, Alex would totally get more RBI's than Nomar. The reason for that is because Alex has better hitting skills (OBP, SLG) than Nomar does and therefore is likely to generate RBI's more often, not becuase of some special ability to generate RBI's.

Mike,

The thing with passes is to have them correlated with hits. Being on base can be walks, singles, doubles, triples, homeruns, misc. That correlates with the subjective measurement of good passes, safe passes, great passes, useless passes, etc. On base percentage doesn't tell you anything about the quality of the hit, or what situation the hit was in. It only tells you that this guy gets on base at this rate. Passing need a metric like this, I think.

The idea that over the course of a season the good passers end up with more assists maybe true, but also true is that Assists typically end up with the player in the PG position. Assists paint a very inaccurate picture of the passing situation.

[Kevin Pelton]

[Dan Rosenbaum]

[drebelx]

I think some comparisons to baseball could be fruitful if given some thought and understanding the differences and making some corrections for those differences. You brought up the point that if 4 players come to the plate in baseball and do not get out, they score a run. This is because of the limits set up by baseball, there are only 4 bases. In basketball there are no bases, it is limitless, we can pass to our hearts content and never score till eventually the clock tells us to give the ball to the other team. A turnover. The thing that is trying to be avoided.

In baseball you get an out and you fail for the team. In basketball you turn over and you fail for the team. Only one person has possession of the ball as does one person in baseball hold the bat.

[S.K.]

The main reason that the RBI comparison doesn't work is that RBI% is a fluke. In other words, there aren't players out there who will consistently get more RBI than they should, given the men on base and their own hitting ability. One year Player X might drive in an unusual amount of guys given his situation, but the next year he'll revert to normal and Player Y will be the flavor of the month. Getting RBI is not a skill, it's simply a product of hitting ability and opportunity.

Assist%, on the other hand, has been shown to be a consistent ability. How much of that is a product of usage is debatable, but we can certainly agree that if you moved someone like Nash off the ball to SF he would still have a better assist rate than average.
_________________
No books - no articles - no website.
Just opinions.

Ill-informed opinions.

[drebelx]

bchaikin in another post on this forum brought up Brevin Knight.

[FrontRange]

I hestitate to reply bcs it is really a silly discussion to compare assists (which in-accurately measure an individual skill) to RBI (which measure a circumstance), but . . .

Brevin Knight has always had good assist rates, above 10/48 minutes every year except last year. This year is slightly better than other good years but he would have projected to top 10/5 in most over his other years on per 48 minutes basis

You can certainly have system/position assists, but no matter what system/position you play Mt. Mutombo in, he will never get alot of assists - he doesn't have the passing, dribbling, pentration skills.