Arguments against Analytics

[Borut]

There are a lot of ex-players ranting their frustrations towards analytics, due to not being able to get front office jobs in NBA teams.

I will try to be make a more rational point against analytics. There is a concept called ambiguity aversion, where one is more willing to take risks in the presence of set probabilities. Could the numbers be just soothing to the brain. We know probabilities imply their uncertainty, furthermore they need to be updated all the time. Therefore, they might serve as illusions of knowledge of the game.

[SNPA]

There are a lot of ex-players ranting their frustrations towards analytics, due to not being able to get front office jobs in NBA teams.

I will try to be make a more rational point against analytics. There is a concept called ambiguity aversion, where one is more willing to take risks in the presence of set probabilities. Could the numbers be just soothing to the brain. We know probabilities imply their uncertainty, furthermore they need to be updated all the time. Therefore, they might serve as illusions of knowledge of the game.

The argument against analytics is that it is simply a translation from the complexities of reality into a quantitative and thus reduced form and has only the illusion of objectivity when it is in fact subjective when it comes to interpretation.

It doesn't do a good job of telling you why something happened.

I better watch out, I said something like this recently and got called a quack. Lol.

[fianchetto]

There is not one rational argument against analytics in general. The presence of a field whose goal is to quantify and model the game can't hinder us in any way.

But arguments against the current state of analytics are easy to find:

- There's no way to verify what analytics are "good", whether or not they are mathematically correct and/or rigorous
- They can't adequately capture half of the game (defense) as well as many other important intangibles
- Player analytics do a poor job of sifting out the noise and focusing on the player's impact alone (what is impact anyway?)

[SNPA]

There is not one rational argument against analytics in general. The presence of a field whose goal is to quantify and model the game can't hinder us in any way.

But arguments against the current state of analytics are easy to find:

- There's no way to verify what analytics are "good", whether or not they are mathematically correct and/or rigorous
- They can't adequately capture half of the game (defense) as well as many other important intangibles
- Player analytics do a poor job of sifting out the noise and focusing on the player's impact alone (what is impact anyway?)

I agree. The mere presence of it is not a hindrance. The elevating of it to the "truth" or supremacy can be.

[penbeast0]

I better watch out, I said something like this recently and got called a quack. Lol.

You know what we say, "If its size and general visual conformation, measured walking speed and style, and recorded vocal utterings are within 1 standard deviation of those of a duck, it's probably a duck."

[SNPA]

I better watch out, I said something like this recently and got called a quack. Lol.

You know what we say, "If its size and general visual conformation, measured walking speed and style, and recorded vocal utterings are within 1 standard deviation of those of a duck, it's probably a duck."

Lol. This quacks me up.

[5paceman]

Gotta have both, but what analytics are keeping former players from FO's? I'd rather have a former player and basic train them on analytics/have a position dedicated to just that. They alone would be great if basketball were played on paper/computer by sexy robots, but unfortunately it occurs in the real world through complex humans. Strategies like take only threes and shots at the rim are kinda like well sure duh, and then breakdown against good competition. Turns out midrange 1v1 is where you end up against good defenses and teams will just defend the paint and line if you are so rigid; it's a game of counters. And these all encompassing stats are largely worthless except for a from the clouds view imo. You can say, "this seems to be the case," and then to actually figure out why or if your assumptions are correct, you have to deconstruct that advanced stat back down to it's pieces and watch actual hoops.

[reamily]

Its really good to have a former player having analytic background so that he can have a personal touch on players like for example, player x, I believe in you but ising our advance stats what your doing hurts the team and I can give you a footage on why you doing it wrong and what is also things that you go unnoticed but doing it right..unlike a guy who has business back ground and just use analytics to bombard his players out there..

[reamily]

Analytics is so useless in playoffs as jn regular season teams go relaxed in defense also experimenting on their own stuff..

[JonFromVA]

There are a lot of ex-players ranting their frustrations towards analytics, due to not being able to get front office jobs in NBA teams.

I will try to be make a more rational point against analytics. There is a concept called ambiguity aversion, where one is more willing to take risks in the presence of set probabilities. Could the numbers be just soothing to the brain. We know probabilities imply their uncertainty, furthermore they need to be updated all the time. Therefore, they might serve as illusions of knowledge of the game.

The argument against analytics is that it is simply a translation from the complexities of reality into a quantitative and thus reduced form and has only the illusion of objectivity when it is in fact subjective when it comes to interpretation.

It doesn't do a good job of telling you why something happened.

I better watch out, I said something like this recently and got called a quack. Lol.

Analytics is a broad term which some people use to refer to looking at stats and others use to refer to complex derivatives of stats often based on models and estimates.

It's pretty easy to examine a stat, so for instance if the Synergy data tells us that a player is a really good shooter from the corner even when guarded closely, it's just a click away to look at all of those plays and judge for yourself whether the stat is telling us something interesting or whether there was just a whole lot of ineffective defense going on.

The model based stats are interesting, but I feel raise more questions than they answer, and those questions can only be answered by digging in to the stats that went in to the model and of course what happened on the floor.

If I could train up NBA personal, I wouldn't just be explaining all these stats and models to them; but how they can dig deeper in to them to see and verify trends. I'd encourage them to be open to questioning their perceptions, and then how to go about evaluating their opinions and what they're seeing with their eye, and then how to go about testing it on the floor.

The eye test of a former player or coach who's been watching and breaking down the game for years is invaluable ... the trick is teaching them where they need to be looking and convincing them that they should be looking.

[KPT1867]

So here is my opinion on the subject. Analytics is a God-send at the macro-level. It allows analysts, coaches, players, GMs, etc. to ingest a plethora of information succinctly. It can be used to build teams and to come up with different coaching schemes. Things like that.

However, we have to be naive to think that it has the same effect on the micro level. That’s not how statistics work. There are just too many things to consider. Analytics isn’t going to tell us if a player broke up with his girlfriend and there fore he may play sporadically.

Great teams need to have a balance of analytics and keen coaching. They don’t need to keep chucking threes even after they miss 27 straight. I’m talking to you Houston Rockets

[falcolombardi]

So here is my opinion on the subject. Analytics is a God-send at the macro-level. It allows analysts, coaches, players, GMs, etc. to ingest a plethora of information succinctly. It can be used to build teams and to come up with different coaching schemes. Things like that.

However, we have to be naive to think that it has the same effect on the micro level. That’s not how statistics work. There are just too many things to consider. Analytics isn’t going to tell us if a player broke up with his girlfriend and there fore he may play sporadically.

Great teams need to have a balance of analytics and keen coaching. They don’t need to keep chucking threes even after they miss 27 straight. I’m talking to you Houston Rockets

is not that a bit of gambler fallacy in a way?

the assumption that because 3's are not falling the solution is not taking them at all? after all previous results dont change future results

shots not falling at first doesnt mean they will continúe falling or will "return to the mean" latter

even the bucks didnt stop spacing the floor with 3's and they were shooting like 30% on them

[Topofthekey]

It's easy to settle this debate

Teach an advanced AI everything about basketball analytics and task it with building the best team it could from the pool of existing NBA players

Form a group consisting of the best and most experienced basketball minds currently and have them similarly construct a team without using any analytics

Pit both teams against each other in a 7 game series and see which team comes out on top

[FlyingArrow]

It's easy to settle this debate

Teach an advanced AI everything about basketball analytics and task it with building the best team it could from the pool of existing NBA players

Form a group consisting of the best and most experienced basketball minds currently and have them similarly construct a team without using any analytics

Pit both teams against each other in a 7 game series and see which team comes out on top

Difficult to do that when you'll have 5 or 6 players on both teams.

[LiSTWithLani]

It's easy to settle this debate

Teach an advanced AI everything about basketball analytics and task it with building the best team it could from the pool of existing NBA players

Form a group consisting of the best and most experienced basketball minds currently and have them similarly construct a team without using any analytics

Pit both teams against each other in a 7 game series and see which team comes out on top

The Raptors are utilizing IBM Watson and I also think that they have excellent coaching and scouting staff.
Since implementing Watson, they have been a top team in the league and won a championship without a single player that was selected in the lottery.
If that isn't the closest thing in the league, I don't know what is. They have also been one of the winningest teens in the league consistently as well.
There is merit to this.

Sent from my Pixel 5 using RealGM mobile app

[DeBlazerRiddem]

Analytics if used well can really complement traditional coaching.

But they will only tell you what happened, not why it happened or more importantly what could happen.

[Wigginstime]

There is not one rational argument against analytics in general. The presence of a field whose goal is to quantify and model the game can't hinder us in any way.

But arguments against the current state of analytics are easy to find:

- There's no way to verify what analytics are "good", whether or not they are mathematically correct and/or rigorous
- They can't adequately capture half of the game (defense) as well as many other important intangibles
- Player analytics do a poor job of sifting out the noise and focusing on the player's impact alone (what is impact anyway?)

Here are a few good arguments against analytics

1. They drive players to make basketball decision that are counter productive to winning for the purpose of analytical performance. The best example is not talking a long shot at the end of a quarter. There is no reason to not take this shot it will always increase your probability of winning. In the 90s this shot was taken every time. Players today are more concerned with preserving their TS% / PER/ Etc than making a low percent long rage 3pt shot.

2. Analytics give a false sense of being able to evaluate basketball without actually watching the game. They purely analyze the end result without providing any information with regards of what led to that end point. They won't tell you if poor stats are due to a player performing badly, facing double teams, or simply deferring to a teammate who had a superior matchup

3. Analytics have noise which could result in really bad decision making. For example RPM in 2019 would have told you Eric Bledsoe was a top 15 player in the league and one of the biggest drivers for the success of the Bucks. An Analytics driven person would have said the Bucks should keep Bledsoe at all costs. Someone who actually watched the game was able to understand the drastic pitfalls of Bledsoe games and support making a trade of Bledsoe for Holiday despite the fact that it cost them 3 first round draft picks

[FNQ]

There is not one rational argument against analytics in general. The presence of a field whose goal is to quantify and model the game can't hinder us in any way.

But arguments against the current state of analytics are easy to find:

- There's no way to verify what analytics are "good", whether or not they are mathematically correct and/or rigorous
- They can't adequately capture half of the game (defense) as well as many other important intangibles
- Player analytics do a poor job of sifting out the noise and focusing on the player's impact alone (what is impact anyway?)

Here are a few good arguments against analytics

1. They drive players to make basketball decision that are counter productive to winning for the purpose of analytical performance. The best example is not talking a long shot at the end of a quarter. There is no reason to not take this shot it will always increase your probability of winning. In the 90s this shot was taken every time. Players today are more concerned with preserving their TS% / PER/ Etc than making a low percent long rage 3pt shot.

2. Analytics give a false sense of being able to evaluate basketball without actually watching the game. They purely analyze the end result without providing any information with regards of what led to that end point. They won't tell you if poor stats are due to a player performing badly, facing double teams, or simply deferring to a teammate who had a superior matchup

3. Analytics have noise which could result in really bad decision making. For example RPM in 2019 would have told you Eric Bledsoe was a top 15 player in the league and one of the biggest drivers for the success of the Bucks. An Analytics driven person would have said the Bucks should keep Bledsoe at all costs. Someone who actually watched the game was able to understand the drastic pitfalls of Bledsoe games and support making a trade of Bledsoe for Holiday despite the fact that it cost them 3 first round draft picks

1) Would love to see some evidence on this, I have not seen anything like this. And in the cases where a player has done something selfish for themselves (Ricky Davis triple double, Kobe's last game), its never been for the sake of analytics. Its been the sake for counting stats

2) Analytics are not responsible for user error. If someone who's unfamiliar with guns shoots themselves in the foot, you don't blame the gun. It did what it was supposed to do.

3) RPM never tells you who the best player is. RPM measures a player's effectiveness, in a role, on a team, to measure the success of that player in a specific role on a specific team. Steve Novak used to regularly top the charts of RAPM. No one suggested Steve Novak was anything more than a specialist - however he did show that if a player can play competent defense while being a marksman, they were extremely valuable. As for the rest, I refer you to point #2.

I dont any of those are even close to a good argument against analytics. Honestly, there aren't any. The people who constantly misuse these values are typically ones in a forum making a bad basketball argument, not analytics depts and coaches making bad decisions. The whole crusade against analytics is strictly cosmetic - you wish people in forums wouldn't misuse them to justify bad takes... but bad takes will exist even if analytics went away tomorrow. The other argument against analytics is 'old man yelling at cloud' - I dont understand them, I dont need them, thus they aren't valuable!

[Colbinii]

There is not one rational argument against analytics in general. The presence of a field whose goal is to quantify and model the game can't hinder us in any way.

But arguments against the current state of analytics are easy to find:

- There's no way to verify what analytics are "good", whether or not they are mathematically correct and/or rigorous
- They can't adequately capture half of the game (defense) as well as many other important intangibles
- Player analytics do a poor job of sifting out the noise and focusing on the player's impact alone (what is impact anyway?)

Here are a few good arguments against analytics

1. They drive players to make basketball decision that are counter productive to winning for the purpose of analytical performance. The best example is not talking a long shot at the end of a quarter. There is no reason to not take this shot it will always increase your probability of winning. In the 90s this shot was taken every time. Players today are more concerned with preserving their TS% / PER/ Etc than making a low percent long rage 3pt shot.

Is there any indication that taking these shots results in a higher Win %?

2. Analytics give a false sense of being able to evaluate basketball without actually watching the game. They purely analyze the end result without providing any information with regards of what led to that end point. They won't tell you if poor stats are due to a player performing badly, facing double teams, or simply deferring to a teammate who had a superior matchup

This isn't an argument against analytics, this is an argument for both watching the game and looking at analytics.

You could just as easily flip this, stating that watching the game makes people feel like they don't need analytics--yet our eyes miss far more of the game than analytics miss.

3. Analytics have noise which could result in really bad decision making. For example RPM in 2019 would have told you Eric Bledsoe was a top 15 player in the league and one of the biggest drivers for the success of the Bucks. An Analytics driven person would have said the Bucks should keep Bledsoe at all costs. Someone who actually watched the game was able to understand the drastic pitfalls of Bledsoe games and support making a trade of Bledsoe for Holiday despite the fact that it cost them 3 first round draft picks

That isn't what analytics state though. That's simply misusing analytics.

Its the same as saying that using a Hammer as a Saw means the Hammer is a bad tool--thats obviously not the case--its just a misuse of the tool.

To summarize what you said: Dumb people are dumb and will misuse their tools.