Hyperbole in ratings...

[jeebs9]

Re: Hyperbole in ratings...

I think Ray Allen is a notch below Steph, even as great as Ray was, not sure I put him there, just my opinion, Ray Allen is definitely one of the all time great shooters of all time, but what Curry does night in and night out all while being the top guy for the D to contain puts him on another level for me.

I actually was watching Round table and some of the guys were debating greatest shooters of all time, and I forget who all it was that still out Reggie over Ray. Mainly because Ray never carried a team on his back, like Reggie did,, nor did he hit all the clutch shots that Reggie did. Now there is no bigger Reggie fan than myself, but not sure if I can agree with those guys,lol, but they do make good argument for that being the case.

I wasn't saying Ray Allen is better or less. I was more referring to being 90 plus rating. But it's debatable if Ray Allen carried a team. And sonic and bucks were close. But in my opinion. MIller had big games just as much as Allen did for his teams. There isn't much of a difference in my opinion.

Okafor should at least get a boost in driving layup.

[HowDareI]

Re: Hyperbole in ratings...

As far as his inability to shoot, I'm not seeing it being as a result of double teams, but simply that he's a bad shooter from anywhere outside 6 feet. Since I live on the West Coast and the Sixers being as bad as they are, the chances I have of actually watching them play outside of highlights the next day are slim to none. So I've got to dig into the stats.

Here we go, first thing I did was look into his shot breakdowns, which can be done here:

http://stats.nba.com/player/#!/1626143/stats/shooting/

He looks decent on the box, but 52 for 206 (25.2%) for jump shots? That's atrocious. So I decided to look deeper into that and look into the shot charts.

http://on.nba.com/1O8MCDA

He looks good from the right side of the floor, but once again atrocious from everywhere else. But maybe it's because he's being closely guarded, like you claim, so I decided to look further into that:

NBA.com Shot Tracker

Nope. Looking at "Closest Defender shot > 10ft.", we can see that he consistently shoots awful from ten feet and out whether he's closely guarded or not, and most of the time he's taking open shots.

So the last thing I did was look into the video for some of his shots to find out a little more. That can be done by going back to the link below, clicking on the Mid-Range FGA under the Shot Area and clicking Video. Most of what I saw were really poor open misses. The offense would get confused and Okafor would be standing there open around the free throw line, so they'd get the ball to him, and he'd take the shot and miss. And not just miss, but miss poorly. Shots hitting the backboard instead of the rim. Bricks off the very edge or back of the rim. If I was any of the coaching staff looking at this, I'd ask him not to shoot from that area anymore this year. Not even if wide open.

Looking at both stats and video are vital, IF you want to evaluate a player as an NBA fan or coach. I think when building an NBA AI system in terms of its structure, you need to study the game thoroughly from poring over video after video to find patterns in the way that real NBA players and teams behave, to correctly capture the right animation triggers or flow or reactions or state transitions. For creation of ratings for the NBA 2K video game, however, stats are about 90% of the game. The way game AI works, under the hood, is all based on logic, state machines, and probabilities. It's all stat driven. So it would make sense that the numbers that drive the AI behaviors are also based solely off stats, where possible. I've been re-rating players purely and without bias from stats for close to two decades now, and this method hasn't steered me wrong yet.

Lol I never said Okafor was a good shooter, just (like you said) down low where he's actually in control and not rushed he's fine.

He has a nice touch from the line, and anybody will tell you he has a soft touch when he finishes. But he might never be a good jumpshooter, like Rondo, because his hands are gigantic. And some claim it's a lot harder to shoot like that...maybe it's true...maybe not (Kawhi).

You even confirmed what I said when most of the time he's rushed or pushed out of his comfort zone because of poorly run sets or making him bail us out when he's a center, not a shooter.

I do agree someone needs to sit down and tell him to get more aggressive and fight harder for position, stop settling. It's gonna help him and the team as a whole. It's weird to think we haven't done that considering our GM and the whole analytics thing he's known for.

But there's always two different ways to look at things...I'm not a solely numbers guy because in certain cases they can't show you why someone's doing what they're doing. At least you looked for video as well for visual proof as to what the numbers show...that's (to me) showing you know what you're talking about.

[SonicMage]

Re: Hyperbole in ratings...

Was Iverson a good mid range shooter?

I know, that surprised me, too. But throughout the prime moments of his career, he was decent, especially for the volume of shots he took.

Shot Chart

And here's his MVP season:

http://www.basketball-reference.com/players/i/iversal01/shooting/2001/

41.5% from midrange is actually not bad, a little higher than average, especially for a guard.

The rest of his career he dropped down to 37-39% except for one good season in Denver.

[SonicMage]

Re: Hyperbole in ratings...

I do agree someone needs to sit down and tell him to get more aggressive and fight harder for position, stop settling. It's gonna help him and the team as a whole. It's weird to think we haven't done that considering our GM and the whole analytics thing he's known for.

I don't know, does Hinkie give advice about coaching as well as using statistics to evaluate player acquisitions?

Sam Hinkie always puzzles me as an analytics guy. Chemistry and solid veteran leadership are just as important to a winning team as talent, some analytics gurus ignore the qualitative measures that exist alongside the quantitative ones. Any statistician worth his salt will tell you that 'tanking' is a faulty strategy to take given the way the current lottery is structured. Sure, the team with the worst record gets the best shot at the first pick, but its still only 25%. 3 out of 4 times the team with the worst record won't get the top pick. In fact, historically, the worst team has received the top pick only 1 in 10 times. Is that worth the loss of revenue while struggling to get good young players? And if you don't build a solid trust to create a winning atmosphere around them, those top picks will bail town as soon as their next contract expires. As a long-time Clippers fan, I know the pain of free agents abandoning ship when we were under Sterling's reich. Ugh, I'm getting off-topic here.

Anyways, good luck to your team, I hope they get on the right track soon.

[jeebs9]

Re: Hyperbole in ratings...

I know, that surprised me, too. But throughout the prime moments of his career, he was decent, especially for the volume of shots he took.

Shot Chart

And here's his MVP season:

http://www.basketball-reference.com/players/i/iversal01/shooting/2001/

41.5% from midrange is actually not bad, a little higher than average, especially for a guard.

The rest of his career he dropped down to 37-39% except for one good season in Denver.

I just put 01-02 Jordan vs Iverson... Big different if you ask me.

[SonicMage]

Re: Hyperbole in ratings...

I just put 01-02 Jordan vs Iverson... Big different if you ask me.

Where did I say he was better than Jordan? I just said he was better than the average player, which is right around 38.5%.

[jeebs9]

Re: Hyperbole in ratings...

Where did I say he was better than Jordan? I just said he was better than the average player, which is right around 38.5%.

Sorry I did explain myself. Very sorry. I was using Jordan as a good mid range shooter. While Iverson on the other hand is just an above average like you said right? But their ratings in the small department are very similar. Iverson is probably a 92-95 rated mid range while Jordan is probably 95-98. I'm pretty sure you can find other players that shot just like Iverson. And they aren't rated that high. But Iverson needs to be rated that high. Could you imagine if they rated him like his stats? There would be an outrage.

edit: This is 06 Allen Iverson http://2kmtcentral.com/16/players/9616/allen-iverson

[SonicMage]

Re: Hyperbole in ratings...

Sorry I did explain myself. Very sorry. I was using Jordan as a good mid range shooter. While Iverson on the other hand is just an above average like you said right? But their ratings in the small department are very similar. Iverson is probably a 92-95 rated mid range while Jordan is probably 95-98. I'm pretty sure you can find other players that shot just like Iverson. And they aren't rated that high. But Iverson needs to be rated that high. Could you imagine if they rated him like his stats? There would be an outrage.

edit: This is 06 Allen Iverson http://2kmtcentral.com/16/players/9616/allen-iverson

That card definitely has some mistakes. Standing Shot Close and Moving Shot Close should be in the high 70's/low 80's. Standing Layup should be in the 60's (short guards suck at standing layups right under the rim because of the shot protection). Shot IQ and Offensive Consistency should be lowered a lot. The rest seem about right. He'd still be about high 80's overall which is actually really good.

I have Iverson at about 84-87 for midrange for his prime years, but I haven't done the rating scales for that midrange attribute yet so it's hard to say. I don't know if I agree that there'd be an outrage, not if Iverson plays like himself. And I don't think his ratings would be as low as you expect. But he wouldn't be the highest rated player from that MVP season. Certainly not over Shaq or Duncan.

[HowDareI]

Re: Hyperbole in ratings...

I don't know, does Hinkie give advice about coaching as well as using statistics to evaluate player acquisitions?

Sam Hinkie always puzzles me as an analytics guy. Chemistry and solid veteran leadership are just as important to a winning team as talent, some analytics gurus ignore the qualitative measures that exist alongside the quantitative ones. Any statistician worth his salt will tell you that 'tanking' is a faulty strategy to take given the way the current lottery is structured. Sure, the team with the worst record gets the best shot at the first pick, but its still only 25%. 3 out of 4 times the team with the worst record won't get the top pick. In fact, historically, the worst team has received the top pick only 1 in 10 times. Is that worth the loss of revenue while struggling to get good young players? And if you don't build a solid trust to create a winning atmosphere around them, those top picks will bail town as soon as their next contract expires. As a long-time Clippers fan, I know the pain of free agents abandoning ship when we were under Sterling's reich. Ugh, I'm getting off-topic here.

Anyways, good luck to your team, I hope they get on the right track soon.

Me and my friends have this same discussion anytime we're out watching the game together.

Young players (unless they're Lebron or MJ) aren't gonna just be All-Stars without some sort of mentor, whether it's on the bench coaching or playing. We had Jason Richardson before and he was that guy for the guards. Now we have Carl Landry and no offense but he was never a 20/10 guy for our bigmen to really learn from. Go out and get a guy like Jermaine O'Neal or something....

At this point tho why would any respectable former all star wanna come here lol.

I really think if we don't make any moves next season something will be done about our management because at a certain point they're running one of the most storied franchises into the ground. We have so much cap someone will sign just for the extra money we could give them.

And I can't talk about AI I'm way too biased in that regard so I'll just keep my mouth shut there lol

[Comduklakis]

Re: Hyperbole in ratings...

Lost in the many side arguments (although interesting) is the some of the original points about issues with ratings. I strongly agree that tendencies are a major issue. They need a ton of tweaks, particularly in regards to raising points in the paint.

I don't think NBA player's ego is the issue with overall. I think the issue is how overall impacts rotations, free agent signings, contracts, even fantasy drafts. Because all these types of decisions are tied to overall, it is difficult for 2k to figure out how to make specialists valuable but not give them ridiculously high overalls. Some of the 3 and D guys should really having overalls in the 50s. They should have abysmal driving ability. They should have low scores in rebounding, even speed. But instead those numbers are inflated so that they will get signed by a team, so that they will get minutes, and so that teams can trade them for equal value. Otherwise, due to the importance of overall to the AI "GM", we would see D league guys with high physical ratings getting signed over guys like Anthony Tolliver or Jared Dudley or Channing Frye. It's a flaw in the system and I don't know that it can be solved. Unless 2k can figure out a way to change how teams value players, whether in signing them, trading them, drafting them, and giving them playing time, then the issue with inflate overall ratings will continue and we will continue to see less seperation between the stars and the non-stars than we should.

[Real2KInsider]

Re: Hyperbole in ratings...

But instead those numbers are inflated so that they will get signed by a team, so that they will get minutes, and so that teams can trade them for equal value. Otherwise, due to the importance of overall to the AI "GM", we would see D league guys with high physical ratings getting signed over guys like Anthony Tolliver or Jared Dudley or Channing Frye. It's a flaw in the system and I don't know that it can be solved.

The flaw in the system is that those D-League players DO get signed over most role players by the time a season advances.

I have never understood the perception that role players should have absurdly low ratings when the reality is there isn't all that much separating role players and stars. When injuries happen these players are more than capable of stepping into increased roles without the team falling apart at the seams (as replacing a 90 OVR with a 50 OVR would suggest).

Unless 2k can figure out a way to change how teams value players, whether in signing them, trading them, drafting them, and giving them playing time, then the issue with inflate overall ratings will continue and we will continue to see less seperation between the stars and the non-stars than we should.

Given that the vast majority of players are rated in the 70s and that there were only 50 players w/ 80+ rating entering the season I don't really see how inflation was an issue this year.

[SpeedyClaxton]

Re: Hyperbole in ratings...

In NBA there shouldn't be drastic rating differences between players, doesn't matter if they are superstars or benchwarmers, ratings should be mid 70's to 90's, hate seeing players with 65 rating like they are high school ballers lol, every player can put up 20 ppg if they are given enough playing time and i speak that from experience of playing basketball. If you give Curry 10 minutes per game he would probably average around 10-15 ppg but since he plays 30+ mins each game ofcourse he will put up big numbers.

Giving an NBA player 65 rating is outta this world, it's like saying this dude is ******** but heck he plays in nba so he should be approx 20 times worse player than LBJ, that's ridiculous, in NBA every player is able to put up 20+ points if he is given a proper chance or space regarding his playing minutes.

[Comduklakis]

Re: Hyperbole in ratings...

The flaw in the system is that those D-League players DO get signed over most role players by the time a season advances.

I have never understood the perception that role players should have absurdly low ratings when the reality is there isn't all that much separating role players and stars. When injuries happen these players are more than capable of stepping into increased roles without the team falling apart at the seams (as replacing a 90 OVR with a 50 OVR would suggest).




Given that the vast majority of players are rated in the 70s and that there were only 50 players w/ 80+ rating entering the season I don't really see how inflation was an issue this year.

I agree that there is some signing of guys out of free agency that in reality would never be touched. But if you lowered overalls for role players it would be even worse. As of now there are guys in the D league that are superior in physical attribuates in real life but are blinking idiots when it comes to defensive or offensive awareness. That's what keeps them in the D league. The issue is that 2k can't translate that idiocy very well.

Maybe I didn't make myself completely clear. Within the confines of the game, overall is problematic. The reason being is that for whatever reason (I'm no game programmer) 2k struggles to make stars look like stars compared to role players. The assumption that I and some others make is that it's because the ratings lack seperation. Now I disagree that the level of role players and stars isn't that different. You put Kyle Korver with the same attention that LeBron or Curry gets and the guy is worthless. He's also not helping a huge amount in a lot in other ways.

If the game could really delineate speed differences on the court between a 92 and 94 that would be wonderful. But IMO it doesn't. In real life, two pgs can both be fast, but just a tick more quickness and you've got a mismatch all game.

to combat this the only way I see it is 2k figures out how to make small differences translate onto the court or they make the differences larger but figure out how to have AI "gms" and "coaches" value certain skills and skill packages, particularly with how they fit with other players on the team. I'm not sure I see either fix being particularly likely.

[SpeedyClaxton]

Re: Hyperbole in ratings...

Remove overall rating completely from the game, it's what ruins the game for majority of people. Best case scenario would be to give players some type of grades for each game segment, for example you have LeBron James driving layup at A while you got on other side Tristan Thompson driving layup at C- ; without any numerical attributes just grades, without overall, make players height and weight matter, how the hell can 6'7'' wing who weighs 250 lbs be faster than 6'7'' wing who weighs 220 lbs ?

Implement full realtime physics into game, collision system, rim, ball, audience, referees, everything in arena should be made to comply with realtime physics even dudes that sell food in stands. Animations should be removed completely also because currently game is too heavy on animations, from collision animations to ball striping animations everything seems scripted and plastic. Instead of that they should make game rely on physics so if 2 players collide with each other result of that should be player A height + weight against player B height + weight + the force and speed of collision and then you got final result of that clash, that would be most realistic scenario for any sports game, i'm always up for some random moments in the game.

2K doesn't offer many random/nonpredictable outcomes, i know there aren't games that i could possibly see DeAndre Jordan/Brandon Knight posterizations or from 2 days ago Alec Burks posterization of Jon Leuer because current engine offers xy amount of animations and counteranimations and when you put those numbers that's the number of animations you can see in this game and that is very predictable or dull after 100+ played games, instead of that they should make game engine from zero, implement physics engine for everything from ball, jerseys, refs, cheerleaders, rim, stanchion (why not make shot clock go down after some really powerful slams) all the way to the crowd.

GTA V is one of these next gen games that offer quick peek at physics engine, you never fall same way, there is difference if you fall from 10 ft or 5 ft, there is difference if you hit concrete from wood..and the list goes on, that's the definition of unpredictable gameplay, in current state game rely too much on ratings, animations and that always creates predictable outcomes for us players.

[Real2KInsider]

Re: Hyperbole in ratings...

Maybe I didn't make myself completely clear. Within the confines of the game, overall is problematic. The reason being is that for whatever reason (I'm no game programmer) 2k struggles to make stars look like stars compared to role players.

This is largely because they use worthless attributes like Offensive Awareness (Shot IQ) for an artificial boost. Attributes are not weighed equal to their true value (in game or otherwise). Standing Dunk is a pretty important rating for bigs and it doesn't affect Overall one iota.

to combat this the only way I see it is 2k figures out how to make small differences translate onto the court or they make the differences larger but figure out how to have AI "gms" and "coaches" value certain skills and skill packages, particularly with how they fit with other players on the team. I'm not sure I see either fix being particularly likely.

What I and others have done is adopt a 5-point scale for various attributes to create separation. 2K already does this for various ratings like Consistency. Nobody at 2K can explain the veritable difference between 73 On-Ball Defense and 71, aside from it being a number derived from a formula, which makes for a curious design.