From Mr. Partnow http://nyloncalculus.com/2015/12/21/nylon-calculus-101-possessions/ "Per Play" offensive rebounds start a new event while "Per possession" they continue the possession. As for why numbers might be different in various places, counting possessions can be tricky: http://nyloncalculus.com/2015/12/21/nylon-calculus-101-possessions/ Sounds like the key is that a play is easier to count as a rebound resets the play, whereas an offensive rebound can extend the "possession" further even with a miss, meaning that teams that are excellent offensive rebounders will have inflated points per possession but not points per play.
I get this for late game situations, but a lot of times when calling timeouts it is because the offense is sputtering and the other team is on a run. This is where you can't go by the numbers because there are clearly points in the game where the offense is not working and not functioning up to it's usual PPP.
I think that's where a good coach comes into play - because you may lose some short-term efficiency but you gain long-term efficiency. You have to have a feel for when you think your team can recover and when they can't. Besides, if you use half-court efficiency as a benchmark, you don't seem to be losing that much efficiency after a timeout, but you do remove the chance of benefiting from the increased efficiency of transition plays or plays where the defense cannot fully set.
Just FYI, another wrinkle in play vs. possession which is necessary to understand these stats @SethPartnow So teams that get lots of ORBD might have inflated points per possessions as misses don't hurt, but not points per play? His reply: "inflated" isn't the right word, because they scored those points, but the gap between per play and per possession will be greater
measuring "Per Play" is silly. It sounds like a new play. It's usually not. A new play would mean the team resets back out front after a rebound. They usually don't. It's most often a quick shot, if not a put-back attempt. Further, if you run a play and the defense knocks the ball out of bounds, why isn't that a new play. An inbounds play is actually a new play. I think "Per Play" would make more sense to people if it were called "Per 24 Shot Clock." That's really what it measures. Not new plays. personally, I'm only interested in Per Possession stats, like coaches. All your offense (and the impact of off rebounds) is measured in that.
Time outs are often called when a team can't score any points. When we miss alot of shots, and the other team is scoring a timeout is called. So I would say in this case, our chance to create points greatly increases.
Per possession does inflate production somewhat for teams that are good at getting offensive rebounds, but I agree with you in general that possession has less noise and is probably a more robust statistic.
It doesn't inflate. I don't know what you mean by that. Seth Partnow also said inflate is the wrong word. You're using a definition most people don't use. Possessions has historically always been the measure coaches use, because it captures everything involved while you are on offense. A per 24-second clock measure is not measuring plays. That's too much trying to be like football. The wording is wrong. A "put-back" is not a "play." It's a result of a possession that included an offensive rebound. An inbounds after the defense knocks the ball out is actually a called, set play, but "Per Play" stats says it isn't. It's just arbitrary. Per Play stats are stupid.
Lets say that you miss three shots but get the offensive rebound, and have a one minute long possession or so and end up with 2 points. That's weighted the same as a team that makes three shots over the course of a minute with 2 points each. But those aren't apples and apples in terms of efficiency. On the same token, per-play penalizes the former team but not the later, so you have one team that has 0.67 points per play, and the other that has 2 points per play. Obviously these are extreme and arbitrary examples, but the truth is probably in between - a team that gets a lot of offensive rebounds also has to miss a lot of shots, and while you should reward that production, it's not the same level of efficiency as a team that doesn't miss. If you look at the website, you'll see that most formulas correct in one way or another for offensive rebounding, but that and free throw rate seem to be the biggest "noise" variables when using those metrics. "Inflate" may be the wrong word but it's clear that possession efficiency is more generous towards strong rebounding teams.
I agree though that per-play stats have too much noise. Per possession is probably the best way to measure efficiency on a team or set level.
That's the point. Measure the success of the offense until the offense loses possession. The opposing team can't score while you have the ball. So, how many points do you score before the other team takes possession of the ball.
That's a fair point. But, this may just be silly semantics, but let's say that you have a team like Cleveland vs. a team like Golden State. Cleveland averages about 80 possessions a game, and they shoot 45%, but they collect 20 offensive rebounds a game. As a result, they have 100 plays and score on 45% of them, let's just say 90 points. So about 1.1 points per possession but 0.9 points per play. Golden State averages about 100 possessions a game, and they shoot 50%, and they collect 10 offensive rebounds a game. As a result, they have 110 plays resulting in 110 points, so again 1.1 points per possession but 1.0 points per play. There's not a huge difference there and I didn't include 3's but hypothetically the points per possession masks some of the mistakes Cleveland is making but points per play penalizes them despite getting the offensive rebound. Neither stat is without limitations for measuring efficiency, but I'd agree with you that per possession makes more sense.
