I like using 100. Because I like thinking of it in percentages. Seems way less complicated to me.

One thing rarely mentioned about figuring distribution (maybe because it’s already obvious to most), if I have 2 players both set at 10 distribution, and one of them is playing 20 mins a game and the other is playing 10 mins per game, the offensive output % will be 2 times higher for the player getting the 20 mins.

I take this into consideration frequently when I have backups that can score against a poor defender.

I use ratios.

That way the distribution stays balanced as subs filter in and out.

Posted by iamjoeyd on 2/9/2021 4:24:00 PM (view original):
I use ratios.

That way the distribution stays balanced as subs filter in and out.

You mean a distribution value with a denominator of the percentage of minutes played?

Posted by craigaltonw on 2/8/2021 8:25:00 PM (view original):
One thing rarely mentioned about figuring distribution (maybe because it’s already obvious to most), if I have 2 players both set at 10 distribution, and one of them is playing 20 mins a game and the other is playing 10 mins per game, the offensive output % will be 2 times higher for the player getting the 20 mins.

I take this into consideration frequently when I have backups that can score against a poor defender.

the first paragraph is definitely correct, in the sense that a player at 10 distro will, generally speaking, shoot twice as much as a player at 10 distro who plays half the minutes (generally because maybe the guy playing 10m is playing with low distro teammates while the 20m guy is playing with starters - that sort of thing will cause a significant skew)

i am struggling with the second part though. you take it into consideration how? boosting the scoring of a backup facing a poor defender is often a solid strategy, but i'm struggling to see how that can be related to the phenomenon in your 1st paragraph, in a productive way. generally speaking, the fact that a player playing half the minutes shoots (roughly) half as much is a feature, not a bug - but really it has very little to do with the analysis around distribution, as best i can tell.

what really drives the main line analysis around distribution is scoring efficiency (in a broad sense) - how many marginal (additional) points will your team get for each marginal possession given to a player? in the main line of distro analysis (scoring efficiency - before factoring in 'side' considerations), following standard game theory analysis, you achieve equilibrium / optimal play when the expected marginal value of the marginal possession for each player is equivalent (if the marginal part is confusing, just ignore it - things like tips in and put backs and base turnovers that a point guard will commit no matter how many possessions they have, are out of scope of the marginal possession from distro, but you don't have to worry about that as a newer coach). if this concept seems weird, just think of it like, the more scoring efficiency a player has, the more distro he should get - but that this analysis, the main line part, frankly has absolutely nothing to do with minutes played or total scoring volume over a game. its all about scoring efficiency per possession.

put simply - if two players are going to score equally efficiently, because of their ratings, their matchups, any of that stuff - but one plays more minutes - their distro should match. from there, its completely appropriate and normal for the guy who plays more minutes to get proportionally more shots. but beyond just being normal - in the main line of distro analysis - it is improper or sub-optimal to attempt to compensate to offset that effect. this is a direct consequence of the optimization condition stated in the above paragraph. if you were attempting to leverage this reality (the 10m guy shooting half) to adjust for something else, like the 40% scoring penalty, that would be a different story.

you can only hope for proper or optimal play if you adjust off of scoring quality for just cause, of which there are a bunch but they mostly share a similar flavor - its things like, needing to boost or diminish certain scorers to get more or less of a certain type of scoring, or needing to boost or diminish certain scorers to steer clear of the 40% scoring penalty - or anticipating a particular game planning strategy from your opponent. this is true regardless of whether you are considering distro for base team planning (opponent agnostic) or distro for a specific opponent (note on opponent-specifics like the quality of defenders you'll face - i would view those as resulting in an adjusted scoring quality assessment of your players, which in turn drives the opponent-specific distro - so something like, my backup is facing a weak defender, i consider part of that same central main line of distro analysis, on scoring quality - its just that you only do that in the opponent-specific portion of the scoring quality assessment. newer coaches should note that the opponent-specific part of distro planning usually should only result in small deviations from the base team setup, the opponent-agnostic one).

To me, the question of whether you set distribution to 100, or something else, comes down to how you want to think about your offensive approach. Whether you want to think about distribution as total percentage of plays run, or as something else. If you set to 100, and you are trying to get to a straight, granular % of plays run, you also need to think about how often a player is on the court, which may be what Craig is getting at; so if you’re trying to exploit an expected bench matchup, you need to think not only about that individual players distribution, but the distribution of the players who will be on the court with him.

Personally, I find it simpler, and much better, to approach it the other way. Using the primary (10), secondary (8), tertiary (7), other (0-5) approach, I can think about distribution mostly as shots per 20 minutes under normal circumstances, understanding that my opponent (or bad sims) might try to do some things to disrupt that sometimes. To each their own of course, but this works much better for my brain.

interesting shoe, i see what you mean. here's the gillispie system for actually doing distro - its WAY simpler in practice, than in theory - i know my theory sounds complex but the implementation is as simple as any of yours, IMO:

step 1: i start the year and set distro for my players in 1 simple pass, based purely on their relative scoring ability (consistent with how distro actually works). i typically start with my best player and put him at 12, but that is arbitrary, anything from 8 to 20 or so is probably pretty comparable. this is almost always a per scoring player, and then i go down in order of quality, setting distro for the rest of the per scoring players.

the way to set the next player is to think of the ratio. if player A is the guard with 12 distro, and player B is on the court with him, what's the shot ratio? should player A be taking double the shots? then B would be 6. 50% more? then B would be 8. (no minutes, lineup, rotation considerations here! very straight forward - scoring quality - based on ratings and iq and nothing else).

then i go through the bigs doing almost the same thing. i compare the best big to the best guard and set his mark (taking into account, at this singular moment, the reality that bigs with equal distro tend to shoot a bit less than guards), but then i go through and compare all the bigs to that primary big. if my best big is 10, then same thing - for each other big - while the 2 players are on the court together (otherwise their relative distro is MEANINGLESS), how much more should the better big get shots?


step 2: i then consider where step 1's final distro would leave me. am i too heavy on 3s? too light on big men offense? is my top scorer too high where he's a risk of exceeding 40% of scoring while on the floor, on too often of a basis? is my top backup scorer so high that he might have the same problem? there's a few other things along these lines, but really that's plenty. if any of those things are a concern, i tweak - just a little, usually - to cover.


this completes the process for team setup. game planning, i just go through and kinda softly reconsider step 1 and 2 in light of my opponent. over the course of the year as i learn and players grow, i mostly do simple tweaks on the settings to boost players a bit who are getting better.

anyway, i think if folks try this process, they'll find a couple things - 1) its actually pretty damn simple - the considerations in step 1 are as straight forward in most cases as comparing 1 guard's ath/spd/lp/per/bh to another's, or a bigs ath/lp/whatever to another's. 2) this method lines up extremely closely with the actual functioning of distribution, what distro really is to the sim engine, which makes it unlike pretty much any other method described in this thread.

One of the interesting questions for me is how much should one adjust distro during the season. I - very roughly - start with a distro process akin to what gil describes - but he's a better coach, so all I will claim is that it is similar to gil approach

Now, players then play - and some score very efficiently and others do badly. But, I then say to myself, is the sample size worth my attention? A few games are no basis for adjusting. And any small sample will involve opponents who may or may not be representative. A guy who does great as a backup in say five OOC games against weak opponents may not be very good at all in conference play against tough opponents.....especially if you move him into the starting slot.

yet, I cannot resist looking at results and thinking about adjusting distro. My compromise is to look at performance and consider adjusting distro whenever I get grades - so that is four times a season. I resist taking small samples too seriously. But, if a guy is shooting 40% playing the five after 12 games that at least raises the question whether his distro is too high - or something else needs an adjustment

Posted by fd343ny on 2/10/2021 9:44:00 PM (view original):
One of the interesting questions for me is how much should one adjust distro during the season. I - very roughly - start with a distro process akin to what gil describes - but he's a better coach, so all I will claim is that it is similar to gil approach

Now, players then play - and some score very efficiently and others do badly. But, I then say to myself, is the sample size worth my attention? A few games are no basis for adjusting. And any small sample will involve opponents who may or may not be representative. A guy who does great as a backup in say five OOC games against weak opponents may not be very good at all in conference play against tough opponents.....especially if you move him into the starting slot.

yet, I cannot resist looking at results and thinking about adjusting distro. My compromise is to look at performance and consider adjusting distro whenever I get grades - so that is four times a season. I resist taking small samples too seriously. But, if a guy is shooting 40% playing the five after 12 games that at least raises the question whether his distro is too high - or something else needs an adjustment

Personal preference maybe, but I adjust according to performance after every game. Usually the adjustments aren’t staggering. I find it pivotal, though, especially as new players develop.

Does anyone else use Kenpom’s points per possession formula to analyze offense production?

And Gil, I take minutes into consideration for distribution to achieve best-possible offense production.

Usually, the top scorers on any team are also going to be starters. The way I set up distribution, my starters’ mean distribution might be a 10. Backups might be a 5.

Minutes played, though, are important to consider when you look at a team’s offense production as a pie chart.

For example, if every player played an equal amount of 200 minutes in a game and that is represented by a ratio of 1:1. Adam’s ratio is 1.5:1 of the minutes and Jones has a .7:1 ratio. If Jones also is more efficient on offense, I take his low ratio of minutes played into consideration when setting distribution.

I look at it less as which player is on the court when and more as a pie chart.

what is the formula for that?

this one?

Points/(FGA-OR+TO+(0.475xFTA))


or this one?

Points/(FGA+TO+(0.460xFTA))


or something else?

i generally do it like craig, lots of tweaks based on how efficient the player is.

except distro can never equal 13, or 6 or 16, for obvious reasons. (same for practice minutes)

also, not sure why, but i never use 0. i put the worst offensive players at 1.
if you use 1 as your minimum, you kind of need to keep overall up closer to 100 so the 1s become a smaller piece of the pie, as intended.

have any of you ever looked at what percent of minutes the starters are with the starters?
seems hard to predict who will be on the floor together. but maybe it usually works out that way?

I know a lot of coaches who just do not change distribution plans during the season. It is just too much effort, especially in 2 a day Worlds. Which is why Man to Man is my favorite defense. Press or zone you can keep it the same and it will not make much of a difference, but M2M you have to change it up for your opponent if you want to be competitive.

Posted by oldave on 2/11/2021 1:50:00 PM (view original):
what is the formula for that?

this one?

Points/(FGA-OR+TO+(0.475xFTA))


or this one?

Points/(FGA+TO+(0.460xFTA))


or something else?

I use points as numerator.

FGA-OR+TO+(.46(FTA)) as the denominator.