http://www.basketball-reference.com/lea ... games.html
Nevermind I found it in the most obvious place possible lol
Basktball Game simulation
-
leerichardson
- Posts: 21
- Joined: Wed Jul 17, 2013 12:15 am
-
nbacouchside
- Posts: 151
- Joined: Sun Jul 14, 2013 4:58 am
- Contact:
Re: Basktball Game simulation
any of you guys seen this?
http://play.basketball-gm.com/
it's kind of a neat tool. you can upload custom rosters, and some redditors have made rosters representing current nba squads (http://www.reddit.com/r/BasketballGM/co ... roster_73/) as of today. those rosters have produced some wonky results simulating the full league for me. I've done it five times and each time some team that has no business winning a certain number of games happens. like washington winning 50 or the sixers winning 40 (!) games.
because of the way it is set up and what it's designed for, you can't do things like simulate 1000 seasons to get the most likely result. also, the rosters have some players rated... wrongly. it'd be neat if someone could tweak this open source set up and using actual statistics instead of the "skills ratings" that the system uses and could also make it so you could run 100s of simulations at once or single game simulations. here's the source code: https://github.com/jdscheff/basketball-gm
http://play.basketball-gm.com/
it's kind of a neat tool. you can upload custom rosters, and some redditors have made rosters representing current nba squads (http://www.reddit.com/r/BasketballGM/co ... roster_73/) as of today. those rosters have produced some wonky results simulating the full league for me. I've done it five times and each time some team that has no business winning a certain number of games happens. like washington winning 50 or the sixers winning 40 (!) games.
because of the way it is set up and what it's designed for, you can't do things like simulate 1000 seasons to get the most likely result. also, the rosters have some players rated... wrongly. it'd be neat if someone could tweak this open source set up and using actual statistics instead of the "skills ratings" that the system uses and could also make it so you could run 100s of simulations at once or single game simulations. here's the source code: https://github.com/jdscheff/basketball-gm
Re: Basktball Game simulation
Yes, I follow this project since ~2008 (it was written in Python back then (http://code.google.com/p/basketball-gm/). You can find this version as a branch in his github repo). The developer is a nice guy. You could raise an issue on github and ask if he could implement these changes. I think he would appreciate feedback from basketball and stats loving people. :)
Re: Basktball Game simulation
I want to see a game simulator in which shooters get "hot" and get more minutes than they normally would; and opponents react with more minutes given to defensive specialists.
Also, players would not tend to shoot as well when their best distributors and ball handlers are not on the floor. They'd also have more turnovers.
Also, players would not tend to shoot as well when their best distributors and ball handlers are not on the floor. They'd also have more turnovers.
Re: Basktball Game simulation
Also, players would not tend to shoot as well when their best distributors and ball handlers are not on the floor
does best distributors mean most assists?...
does best distributors mean most assists?...
-
leerichardson
- Posts: 21
- Joined: Wed Jul 17, 2013 12:15 am
Re: Basktball Game simulation
Wow that GM site is cool. Our game is using team/player statistics to simulate games. Our goal is to use this seasons stats to simulate the season, and see how similar our simulated seasons are to the actual season. But that github post will be crucial to see how I can actually store data and stuff. I'm a noob programmer
Re: Basktball Game simulation
Ideally your best ballhandlers are getting the ball to the best scorers in their favorite spots.bchaikin wrote:Also, players would not tend to shoot as well when their best distributors and ball handlers are not on the floor
does best distributors mean most assists?...
Assists are in fact a register of "successful" passes: those which result in a FG.
I'd also suggest -- not trying to be evasive here -- the "best distributors" are those players whose play improves their teammates' shooting the most.
It somewhat stretches the definition when you consider that 'setting picks' and 'spreading the floor' can have positive effects on 'distribution' to the optimal spots. But such acts at least facilitate the optimal distribution.
The point guard may dribble into the lane and pass out to the corner. He has facilitated his own pass, in this case. The guy who set the pick also facilitated, as did the decoy on the perimeter. They all conspired in the distribution.
Re: Basktball Game simulation
I want to see a game simulator in which.... players would not tend to shoot as well when their best distributors and ball handlers are not on the floor. They'd also have more turnovers... Ideally your best ballhandlers are getting the ball to the best scorers in their favorite spots. Assists are in fact a register of "successful" passes: those which result in a FG.
so in your game simulator, how would you model the 98-99 atlanta hawks and the 98-99 minnesota timberwolves? that year these two teams were 25th and 26th in overall shooting from the floor (a 43.9% vs 43.6% eFG%) yet while the t-wolves passed for assists on 66% of their FGM, the hawks did so on just 51%...
or how about the 78-79 san diego clippers and denver nuggets? both shot the same (48.3% vs 48.1% eFG%), but the nuggets assisted on 62% of their FGM and the clippers on just 41%...
so in your game simulator, how would you model the 98-99 atlanta hawks and the 98-99 minnesota timberwolves? that year these two teams were 25th and 26th in overall shooting from the floor (a 43.9% vs 43.6% eFG%) yet while the t-wolves passed for assists on 66% of their FGM, the hawks did so on just 51%...
or how about the 78-79 san diego clippers and denver nuggets? both shot the same (48.3% vs 48.1% eFG%), but the nuggets assisted on 62% of their FGM and the clippers on just 41%...
Re: Basktball Game simulation
I'd say teams that are more dependent on passes for scoring, are also more dependent on those players who have gotten the assists.
In 1999, the Hawks were a better team than the Wolves; their scorers weren't assisted as often, but they still shot as well.
In 1999, the Hawks were a better team than the Wolves; their scorers weren't assisted as often, but they still shot as well.
Re: Basktball Game simulation
you said:
I want to see a game simulator in which.... players would not tend to shoot as well when their best distributors and ball handlers are not on the floor. They'd also have more turnovers...
so then in a simulation how would you handle this:
In 1999, the Hawks were a better team than the Wolves; their scorers weren't assisted as often, but they still shot as well.
how would you model/simulate teams shooting and scoring better with better offensive efficiency but with less assists?...
you also said:
I'd say teams that are more dependent on passes for scoring, are also more dependent on those players who have gotten the assists.
i'd say you're wrong...
the 79-80 and 80-81 san diego clippers had identical rates of offensive efficiency - 104.4 and 104.5 pts/00poss scored (48.2% vs 48.6% eFG%s). yet in 79-80 they averaged just 48 ast per 100 fgm, and in 80-81 60 ast per 100 fgm. that's a very large 20% difference in assist rate, with no loss in offensive efficiency - yet in both seasons 4 of the 6 players to play the most minutes on the team were the same, swen nater, brian taylor (who was the starting PG both years), joe bryant, and freeman williams...
the 90-91 boston celtics saw a large 14% drop in ast/fgm ratio, from 68% to 59%, from 89-90, yet they shot better (52.0% vs 50.6% eFG%) and were slightly more efficient on offense (111.0 vs 110.6 pts/100poss), and also with 4 of their top 6 minutes players being the same both seasons - larry bird, reggie lewis, robert parish, and kevin mchale...
there are many other examples...
but there are few examples that show a correlation being a team's overall shooting or offensive efficiency being dependant on players getting lots of assists...
I'd say teams that are more dependent on passes for scoring,
all teams are dependant on passes for scoring...
are also more dependent on those players who have gotten the assists.
this however is untrue...
I want to see a game simulator in which.... players would not tend to shoot as well when their best distributors and ball handlers are not on the floor. They'd also have more turnovers...
so then in a simulation how would you handle this:
In 1999, the Hawks were a better team than the Wolves; their scorers weren't assisted as often, but they still shot as well.
how would you model/simulate teams shooting and scoring better with better offensive efficiency but with less assists?...
you also said:
I'd say teams that are more dependent on passes for scoring, are also more dependent on those players who have gotten the assists.
i'd say you're wrong...
the 79-80 and 80-81 san diego clippers had identical rates of offensive efficiency - 104.4 and 104.5 pts/00poss scored (48.2% vs 48.6% eFG%s). yet in 79-80 they averaged just 48 ast per 100 fgm, and in 80-81 60 ast per 100 fgm. that's a very large 20% difference in assist rate, with no loss in offensive efficiency - yet in both seasons 4 of the 6 players to play the most minutes on the team were the same, swen nater, brian taylor (who was the starting PG both years), joe bryant, and freeman williams...
the 90-91 boston celtics saw a large 14% drop in ast/fgm ratio, from 68% to 59%, from 89-90, yet they shot better (52.0% vs 50.6% eFG%) and were slightly more efficient on offense (111.0 vs 110.6 pts/100poss), and also with 4 of their top 6 minutes players being the same both seasons - larry bird, reggie lewis, robert parish, and kevin mchale...
there are many other examples...
but there are few examples that show a correlation being a team's overall shooting or offensive efficiency being dependant on players getting lots of assists...
I'd say teams that are more dependent on passes for scoring,
all teams are dependant on passes for scoring...
are also more dependent on those players who have gotten the assists.
this however is untrue...
-
leerichardson
- Posts: 21
- Joined: Wed Jul 17, 2013 12:15 am
Re: Basktball Game simulation
Yeah I think generally adding a boost to FG%'s if player X is on the court would be easy. If(player X is on the court) then(All other players on the court's shooting %'s in crease by some arbitrary amount). But I don't think that is a vital aspect of a game simulator. It's more important to have the main logic and structure of it down, and then you can add intricacies like this once the foundation is set.
Re: Basktball Game simulation
I didn't say all teams are equally dependent on assists for their scoring; in fact, I thought I clearly stated that some players and teams don't need to be assisted as much....there are few examples that show a correlation being a team's overall shooting or offensive efficiency being dependant on players getting lots of assists...
However, there are players whose FG are more than 90% assisted. Some of them shoot very high eFG%. Would their eFG% not go down if they didn't have their enablers, that is, the assist men?
Do the Clippers shoot or score just as well without Chris Paul? Is this really in question?
Isn't the real question about how much better his teammates shoot when he's on the floor?
When a team's star PG sits down, they tend to go with scorers who can create their own shot. Those who rely on crisp passing may not be of any use at all, at such times.
Re: Basktball Game simulation
Those 4 all shot either less frequently or less well, or both, in 1991.the 90-91 boston celtics saw a large 14% drop in ast/fgm ratio, from 68% to 59%, from 89-90, yet they shot better (52.0% vs 50.6% eFG%) and were slightly more efficient on offense (111.0 vs 110.6 pts/100poss), and also with 4 of their top 6 minutes players being the same both seasons - larry bird, reggie lewis, robert parish, and kevin mchale...
Dennis Johnson and John Bagley -- a couple of 49% shooters -- were replaced by Brian Shaw and Dee Brown, who both were around 52% (TS%)
And Kevin Gamble had his career year in '91 -- picked up his scoring from 5.1 to 15.6 per game, while his shooting jumped from .520 to .620
Team turnovers went from 13.4 to 14.0%, from '90 to '91
FT/FGA dropped from .251 to .228, further cutting into ORtg.
Re: Basktball Game simulation
I didn't say all teams are equally dependent on assists for their scoring; in fact, I thought I clearly stated that some players and teams don't need to be assisted as much.
what you have clearly state are generalizations. how about some specifics? which players? which teams?...
However, there are players whose FG are more than 90% assisted.
yes there are...
Some of them shoot very high eFG%.
yes, and some do not...
Would their eFG% not go down if they didn't have their enablers, that is, the assist men?
why would it? there are excellent high eFG% shooters on teams with both high and low assists PGs, or teams that get both high and low assists...
Do the Clippers shoot or score just as well without Chris Paul?
nope... but he's just one example...
toronto shot better without jose calderon, one of the league's best assist men. with calderon on the floor toronto - minus calderon's shooting - shot worse in 12-13 (10.5 ast/40min) with him on the floor...
Is this really in question?
you bet - there are many teams that shoot better with their starting PG than their backup PG where their starting PG is not a high assists PG. the ny knicks shot better with raymond felton on the floor last season than without, and he averaged just 5.5 ast/g (6.4 ast/40min). the knicks were dead last in the league in ast/fgm (.527) in 12-13 but shot the 8th highest eFG% as a team...
Isn't the real question about how much better his teammates shoot when he's on the floor?
many consider lebron james one of the best passers the league has ever seen, on par with or better than larry bird and others. from 03-04 to 07-08 he threw for the 5th most assists in the league, and played 81% of the time for cleveland. during this time cleveland threw for assists at a rate of 58 per 100 FGM (lge avg was 59)...
yet those 5 seasons, when you factor out lebron's own shooting, the cavs were the league's worst shooting team from the floor - a 46.8% eFG% (the rest of the league was between 47.2% and 53.0% eFG%). how is this possible if assists are so important? i can understand the cavs without lebron's stats not being perhaps best in the league in shooting from the floor, but dead last/worst? what does that tell you about all those assists he threw?...
do the math:
.468 = (.488)(0.81) + (x)(0.19)
x = (.468 - (.488)(0.81))/(0.19) = .383 or 38%
if the cavs other than lebron shot at least just a league average 48.8% eFG% when he was on the floor these 5 seasons, then they would have had to shoot just a 38% eFG% when he was on the bench to shoot overall a 46.8% eFG%. you really think they shot just a 38.3% eFG% in the 1/5 of the time he was on the bench over a full 5 year period?...
When a team's star PG sits down, they tend to go with scorers who can create their own shot.
when a team's star sits down, teams typically go with their backup PG...
Those who rely on crisp passing may not be of any use at all, at such times.
like who?...
Those 4 all shot either less frequently or less well, or both, in 1991.
bird, mchale, parish, and lewis combined shot better in 90-91 (52.9% eFG%) than in 89-90 (52.7% eFG%) despite a large drop in team ast/fgm of 14%. they obviously did not need alot of assists to shoot well...
what you have clearly state are generalizations. how about some specifics? which players? which teams?...
However, there are players whose FG are more than 90% assisted.
yes there are...
Some of them shoot very high eFG%.
yes, and some do not...
Would their eFG% not go down if they didn't have their enablers, that is, the assist men?
why would it? there are excellent high eFG% shooters on teams with both high and low assists PGs, or teams that get both high and low assists...
Do the Clippers shoot or score just as well without Chris Paul?
nope... but he's just one example...
toronto shot better without jose calderon, one of the league's best assist men. with calderon on the floor toronto - minus calderon's shooting - shot worse in 12-13 (10.5 ast/40min) with him on the floor...
Is this really in question?
you bet - there are many teams that shoot better with their starting PG than their backup PG where their starting PG is not a high assists PG. the ny knicks shot better with raymond felton on the floor last season than without, and he averaged just 5.5 ast/g (6.4 ast/40min). the knicks were dead last in the league in ast/fgm (.527) in 12-13 but shot the 8th highest eFG% as a team...
Isn't the real question about how much better his teammates shoot when he's on the floor?
many consider lebron james one of the best passers the league has ever seen, on par with or better than larry bird and others. from 03-04 to 07-08 he threw for the 5th most assists in the league, and played 81% of the time for cleveland. during this time cleveland threw for assists at a rate of 58 per 100 FGM (lge avg was 59)...
yet those 5 seasons, when you factor out lebron's own shooting, the cavs were the league's worst shooting team from the floor - a 46.8% eFG% (the rest of the league was between 47.2% and 53.0% eFG%). how is this possible if assists are so important? i can understand the cavs without lebron's stats not being perhaps best in the league in shooting from the floor, but dead last/worst? what does that tell you about all those assists he threw?...
do the math:
.468 = (.488)(0.81) + (x)(0.19)
x = (.468 - (.488)(0.81))/(0.19) = .383 or 38%
if the cavs other than lebron shot at least just a league average 48.8% eFG% when he was on the floor these 5 seasons, then they would have had to shoot just a 38% eFG% when he was on the bench to shoot overall a 46.8% eFG%. you really think they shot just a 38.3% eFG% in the 1/5 of the time he was on the bench over a full 5 year period?...
When a team's star PG sits down, they tend to go with scorers who can create their own shot.
when a team's star sits down, teams typically go with their backup PG...
Those who rely on crisp passing may not be of any use at all, at such times.
like who?...
Those 4 all shot either less frequently or less well, or both, in 1991.
bird, mchale, parish, and lewis combined shot better in 90-91 (52.9% eFG%) than in 89-90 (52.7% eFG%) despite a large drop in team ast/fgm of 14%. they obviously did not need alot of assists to shoot well...