How is this calculated? I thought it was just points but after my first game I can see it isn't. The last column is off %. Why is Browning (13pts) at 22.2% even though he scored less than Williams(15pts)? Thank you ahead of time for any input..

Robert Browning	Sr.	SG	100	23.0	13.0	3.0	44.4	22.2
Kirk Axford	Jr.	PG	100	20.0	8.0	2.0	33.3	16.7
Dewey Williams	Sr.	SF	100	24.0	15.0	3.0	83.3	16.7

fyi, they took the same number of shots, so its not shots per game. someone else has to chime in.

I believe it is turnovers + shots. And looking at your players, it is evident that I am completely wrong. 

Keep in mind that when a player is fouled on a missed shot, it doesn't go in the books as a FGA... but at a glance, that math doesn't track for Browning.  Maybe some combination of what tarvolon and I are getting at.  It's a good question.

It can't be that, car, because it all adds up to 100%. It could be percentage of shots, turnovers, and shooting fouls drawn out of total shots, turnovers, and shooting fouls drawn, but I don't see how that makes sense with the players listed. 

Looks to me like Browning has 9 shots, 1 TO, and 3 fouls drawn.(13 total)
Axford has 9 shots, 5 TOs, and 0 fouls drawn. (14 total)

But Browning has a higher Off%

off % is supposed to be calculated from the number of possessions that are run for each player. just going to think out loud to see how that looks...
browing - i count 13 in the pbp, this includes shots, fouls before the shot (which isn't in the box score), and turnovers. 9 shots, 1 turnover, 3 fouls before the shot

williams - i count 11 in the pbp. 6 shots, 1 turnover, 4 fouls drawn before the shot

axford - 9 shots, 1 foul drawn before the shot, 5 turnovers, clearly this one doesn't add up.

i don't think all the turnovers necessarily count, however. a turnover could have happened before the half court or before possession really was assigned to that player. axford had 4 balls stolen and 1 illegal screen. the illegal screen was definitely not a possession of his, the 4 balls stolen could go either way.

i started out thinking, well, turnovers - im pretty sure any player can get a turnover when a possession is assigned to a player. so really turnovers in general i went in skeptical of. some clearly count, but some definitely don't. i didn't check the single turnovers by the other 2 guys yet, i do have to run here though... hopefully someone can run with this.

anyway if axford needs to match williams, he has 10 with shots + fouls drawn (which now that i think about it, a foul drawn can happen off the ball - someone would need to check the pbp, im not sure if we always have that level of detail, similar to axford's steals). of his 5 turnovers, 1 doesn't count - but it probably counts for someone else! who? i'm not sure. whoever it takes to get the numbers right, i suppose :)

i am interesting in carrying this all the way through... OP, can you post the entire player % list if you get a chance before 2? it would be helpful if someone who was less retarded than i am being at math right now could say what whole number of shots would even make those %s listed feasible, and what the necessary number of possessions per player is. i really doubt there is more than 1 answer and ill figure it out later if nobody else does, i just have to go. i think we can figure this out, but its definitely through the PBP and there are definitely uncertainties, you can't just do simple math off the box score.

dividing the 22.2 by 16.7 yields 1.3293
to find the range, the smallest that could be without rounding is 22.15/16.75 = 1.3224
the largest is 22.25/16.65 = 1.3363

thats a pretty tight range. im figuring thats got to be one and a third, 1.3333, which would be the ratios of possessions on 16 and 12. its also possible possessions somehow are calculated weighted so its not only whole numbers, but its going to be easier to figure this out if it is whole numbers :)

somehow, browning would have to get up to 16, with the other two being at 12, which seems tough honestly. it could also be 12/9, but that does seem unreasonably tough, given that axford has 10 with shots + fouls drawn alone :) so maybe it is the weighted idea. but i think its better to try for the whole numbers. with 16/12, that implies 72 total possessions for that side, which seems in the realm of reasonable given the # of shots is pretty close to that.

we'd have to run through the entire pbp to see how many unaccounted for events there would be, like axelford's illegal screen turnover - clearly someone else had possession on the team - so thats 1 we can assign anywhere to make the numbers add up. its not clear to me if a steal can be a possession for another player - i suppose the steal could have been on the pass to that player - but im not sure if the engine works that way. there are also things like losing the ball out of bounds - do we consider that a possession for the player, or one that could have happened by any player on a possession that was already assigned to a player? i also feel like in some cases, steals could happen with no player possession yet - like on bringing the ball up the court. im not sure if this is only possible in press or not. it would be interesting to try to add them up though and see where the total lies, we could probably nail this down 100% in about 5 pbps.  unfortunately, looks like the OP didn't see my message or was unable to post the player %s in time, so we might have to throw away this example and start again with one of the world that is about to begin the season - get like 5 sets of data from 5 coaches - categorize the events by the type and by player - could sort all this out.

that probably really wouldn't be too bad - however, its obviously non trivial. i think by my numbers, it already shows that the order shown in the OPs example is logical, so if you guys just want to take what i did here and assume its true, thats cool with me, its not like any of this really matters. it might enhance our collective understanding of the sim engine, which is where my interest really comes in, to me its cool to know something like what kind of turnovers can happen when a given player has possession, if there are turnovers before a player gains possession, etc....

I believe that gillespie is accurately summarizing what is believed to be the way it works.

One key item is that it is NOT something calculated over a full game - it is a probability applied on each possession based on who is on the floor at the time and the ratio of the their % distro.  So if there are five guys on the floor with % of 10,10,10, 5 and 5 that would be the ratio of their distros.

A second key item for coaching is that it does not mean that players may not get touches that do not correspond to the ratio:

a. random variation
b. other sources of touches - an offensive rebound, a steal, etc

Posted by mamxet on 8/12/2014 4:37:00 PM (view original):
I believe that gillespie is accurately summarizing what is believed to be the way it works.

One key item is that it is NOT something calculated over a full game - it is a probability applied on each possession based on who is on the floor at the time and the ratio of the their % distro.  So if there are five guys on the floor with % of 10,10,10, 5 and 5 that would be the ratio of their distros.

A second key item for coaching is that it does not mean that players may not get touches that do not correspond to the ratio:

a. random variation
b. other sources of touches - an offensive rebound, a steal, etc

you are correct that distro is not calculated over a full game - it is only applied to players on the court. however, for offensive %s (which are just a cosmetic stat for the use), it is calculated over game play thus far in the season - so its a full game after 1 game, 10 full games after non conferences, etc...

agree fully, the cosmetic stats are calculated that way , at least I fully believe so.

for new coaches, distro is one of the top few non intuitive mechanisms in the engine

Turns out Off% is simpler than any of us expected.  (Player FGA + FTA/2) / (Team FGA + FTA/2)

Posted by llamanunts on 8/15/2014 12:29:00 PM (view original):
Turns out Off% is simpler than any of us expected.  (Player FGA + FTA/2) / (Team FGA + FTA/2)

that's depressingly simple, but it doesn't actually fit the numbers unless it's rounded down to the nearest whole number.

I'm getting the right answers when I apply the formula to my Bowie St. team - rounding to the tenth of a % seems correct.

if that's the formula, then Axford and Williams shouldn't have had the same Off% after one game, because one had 9.5 (FGA + FTA/2) and the other had 9 (FGA + FTA/2)