[ropeParticipant]

I have to admit that a lot of the advanced statistical analysis leaves me cold as a fan. I really can’t read articles that have tables of numbers, charts and graphs without making my mind go numb. For instance, I personally don’t even like the per 36 / per 40 stuff, which is really simple to understand. There is no way that you can say that an individual player’s production will go up or down if their minutes were increased. I do know however that some teams have dived reall deeply into the realm. Makes sense if you do it for a living to break things down into minute detail. I know Memphis went hard in this direction, going as far as to hire John Hollinger to provide the same kind of thing he was doing for ESPN, but for real. Two questions, I guess. First – what other teams are known for applying advanced statistics in their player evaluations and team construction? Are there some teams doing the "Moneyball" thing more than others. Second, how much interest in this do you have as a fan? Do you like that kind of analysis?

[ChoppyParticipant]

I think it’s best used as a comparative analysis tool. As as means of comparing two similar players, breaking down their games and boiling their production down into a single number can provide an easy reference point to judge between players. For example, you might use it to say that Player A is better on the break than Player B, or that Player B has higher win-shares when partnered with a pass first PG. I think the league has gone too far in that direction now. The league had to embrace analytics more than it did but has now gone way too far. I think a balance needs to be made between a more analytic approach and ‘the eye test’. Basketball knowledge goes a long way. As we all know, there are things players do on a basketball court that contribute to wins but can’t necessarily be quantified.

I like the analysis – I find it interesting looking at the small details that make a player good/bad. But I do take it with a hint of caution not to trust it implicitly.

[ChoppyParticipant]

I think it’s best used as a comparative analysis tool. As as means of comparing two similar players, breaking down their games and boiling their production down into a single number can provide an easy reference point to judge between players. For example, you might use it to say that Player A is better on the break than Player B, or that Player B has higher win-shares when partnered with a pass first PG. I think the league has gone too far in that direction now. The league had to embrace analytics more than it did but has now gone way too far. I think a balance needs to be made between a more analytic approach and ‘the eye test’. Basketball knowledge goes a long way. As we all know, there are things players do on a basketball court that contribute to wins but can’t necessarily be quantified.

I like the analysis – I find it interesting looking at the small details that make a player good/bad. But I do take it with a hint of caution not to trust it implicitly.

[nick5354Participant]

The first name that comes to mind is Morey. I am getting the feeling he is constructing his team like it is a 2k game. He needs more emphasis on chemistry.

[nick5354Participant]

The first name that comes to mind is Morey. I am getting the feeling he is constructing his team like it is a 2k game. He needs more emphasis on chemistry.

[tli232]

I build predictive models for a living and to be honest I find it pretty interesting what the NBA teams are doing in terms of analytics.

I’ve interviewed with a few teams for analytics positions and they had some great basketball analytics questions for me to solve mathematically. Really fascinating stuff from a purely academic perspective.

(e.g. what is the mathematical increase in likelihood of a vicotry of a block by a backcourt player vs a block by an interior player on own man vs block by interior player on help defense)

However, I agree that it takes someone with a solid math/ stats background to competently make sense of all the stats.

The stats that the media quotes are totally misleading if they “dumb it down” for the average Joe. Because the media just wants a sound bite, they’ll just quote some random Hollinger stats and call it advanced statistics.

In reality, how the NBA teams use this “analytics stuff” is through logistic regressions/ stochastic modeling/ co-variance of multiple variables/ Markov Chains/ et cetera/ ad nauseum that are very closely guarded in term of methodology.

The “advanced stats” you hear about on TV quoted by analysts is just for show. NBA teams go MUCH deeper than using PER/ EFF.

[tli232]

I build predictive models for a living and to be honest I find it pretty interesting what the NBA teams are doing in terms of analytics.

I’ve interviewed with a few teams for analytics positions and they had some great basketball analytics questions for me to solve mathematically. Really fascinating stuff from a purely academic perspective.

(e.g. what is the mathematical increase in likelihood of a vicotry of a block by a backcourt player vs a block by an interior player on own man vs block by interior player on help defense)

However, I agree that it takes someone with a solid math/ stats background to competently make sense of all the stats.

The stats that the media quotes are totally misleading if they “dumb it down” for the average Joe. Because the media just wants a sound bite, they’ll just quote some random Hollinger stats and call it advanced statistics.

In reality, how the NBA teams use this “analytics stuff” is through logistic regressions/ stochastic modeling/ co-variance of multiple variables/ Markov Chains/ et cetera/ ad nauseum that are very closely guarded in term of methodology.

The “advanced stats” you hear about on TV quoted by analysts is just for show. NBA teams go MUCH deeper than using PER/ EFF.

[tuck243Participant]

need to realize that it doesn’t always equate to what type a player you have…  Negative or Positive ways…   I do think the 36 Per thing is a bit ridiculous to spout when there isn’t a reasonable way to know what someone will average with extended minutes…   But it does provide you will some insight on players that are getting around 20 minutes per game…   You can use the 36 per to get an estimate what you think they "might" average…  That’s not something you should bring into a debate because it’s too many varibles at play…

"Advance stats" should be used as a tool to understand the game further…  I feel like people will look at Hollinger PER and automatically use that in their discussion for or against a player…  Which majority of people don’t even know what’s in that formula to recieve that stat…  Furthermore, PER doesn’t equate the "eye test"…  I always felt to properly evaluate a player you need a combination of stats and seeing a player to make a correct assumption… with both it’s STILL it’s a crap shoot on how good a player currently is, or how good he will become…   

  

[tuck243Participant]

need to realize that it doesn’t always equate to what type a player you have…  Negative or Positive ways…   I do think the 36 Per thing is a bit ridiculous to spout when there isn’t a reasonable way to know what someone will average with extended minutes…   But it does provide you will some insight on players that are getting around 20 minutes per game…   You can use the 36 per to get an estimate what you think they "might" average…  That’s not something you should bring into a debate because it’s too many varibles at play…

"Advance stats" should be used as a tool to understand the game further…  I feel like people will look at Hollinger PER and automatically use that in their discussion for or against a player…  Which majority of people don’t even know what’s in that formula to recieve that stat…  Furthermore, PER doesn’t equate the "eye test"…  I always felt to properly evaluate a player you need a combination of stats and seeing a player to make a correct assumption… with both it’s STILL it’s a crap shoot on how good a player currently is, or how good he will become…   

  

[Author]

Posts