NBA 2K15 News Post

Revamped Overall Formula for NBA 2K15 Player Ratings Explained

10:38 PM - September 3, 2014 by bedwardsroy19

Promoted from Forums to homepage, written by Mike Stauffer, NBA 2K Production Assistant for rosters and ratings.

As NBA 2K15 Overall ratings are released, I think it is important to point out that NBA 2K15 will feature a revamped formula to determine a player's overall rating. This year a player's overall rating will be determined by what "type" of player they are. By having more dynamic overall rating formulas it allows for players who are highly valued by teams to have a more representative "Overall" rating. Specialists such as great defenders, shooters, and rebounders will have an overall value that properly displays their worth to an NBA team.

The goal of the newly calculated Overall formulas is to as accurately as possible show a player's value in the NBA. The new formula will really highlight those players that play a very important role on their team, but may not be the prototypical player at their position. All players in the NBA provide some sort of value to their team, and the revamped Overall formulas should really illustrate that in NBA 2K15.

(More details below, as Mike answers questions from the community.)

So if I have two SG's with identical attributes but one is a "Defensive" and the other is "3PT" they will produce two different overalls?

Mike: Good question! No, the individual ratings will determine what player type (per position) to use.

If a player gets traded to another team, does the overall change?

Mike: No. It is 100% based on a player's attributes.

So basically what you are saying is that the 2K ratings being released paint an incomplete picture without releasing the player types as well?

Mike: Not exactly, and this post should address other confusion in the thread:

For every player, their individual ratings will run through all of the overall formulas possible for a given position. Whichever formula generates the highest overall is what is displayed in the roster. The individual attributes matter more than ever in determining a players overall.

Can you could elaborate as to what are some of the overall formulas possible for a given position might be based on?

Mike: We will be elaborating on this in the future, this was just to add some perspective in the ratings that are being released. But in the past, player's overall rating by position was calculated by a singular formula. In 2K15 there are many different formulas that are calculated that will look at certain individual attributes with more weight than others to account for different player types. Of those formulas the highest Overall is selected. The overall formula from games past is still there, but many other formulas are being calculated to accurately display the overall worth of a player who specializes in a certain part of the game.

In years past many claimed Overall ratings "mean nothing". This year they should. Player's Overall rating should finally be the point of debates.

It sounds like they are just removing the filter of position in determining overall rating. Essentially- ratings will be calculated pretty much the same, but instead of a given player's position dictating which of the 5 potential overalls they receive (either pg,sg,sf,pf, or c overall) they tweaked it so a players overall is the highest of the 5 formulas. They simply eliminated a position penalty and put the onus on the ratings. I don't think it's as in depth and evolved as some of you are guessing.

Mike: I'm not sure where this assumption is coming from. As I've said a few times in the thread, there are a variety of Overall formulas that are now being calculated per position to better represent unique players that may not be all around players for their position. These players are still very valuable to a team and in 2K15 their rating will reflect that value.

Just added these comments from Leftos.

Each Position has a set of Archetypes (or Player Types, if you will). All-Around, Athletic, Defensive, etc. Some positions share some of them, some have some unique ones too. A player has an overall per player type per position. So a player has (NumberOfPositions * NumberOfPlayerTypesPerPosition) overall ratings. Each Positional Player Type has its own overall formula (so PG All Around has a different formula than PF All Around has a different formula than C Defensive).

Whenever his overall rating gets updated, we calculate all the overall ratings for his position by player type. The player type that gives him his the highest overall is the one we determine "most compatible", so we assign it to him along with that Overall.

So yes, a player that might be C All-Around but as years go by sees his offensive attributes regress but you've made sure to keep him up to par defensively using (cheap plug but I'm a dev) our new in-season Training system, might see his player type change to C Defensive, and his value to the team will still be there.

As for team-building AI, we've taken some steps to make sure teams take into account specific needs as far as more specific skills go. So if that player losing his offensive ability means that the team lost their main source of points, they'll value players that can bring the team's scoring up more than players that might be contributing to needs sufficiently covered (such as defense, in this example). (Let me clarify that the examples of "offense" and "defense" are shallow and the "skills" teams look at go beyond that; there's 13 different categories actually.)

That said, Team Style is still a factor, so teams won't all "average out" by trying to cover skill needs, if a coach prefers to play a certain way. So if a team prefers outside scorers over inside scorers, it won't value inside scorers as much, even if it has more of a need for them than another team.

Also, we pay much more attention at position stacking which has been a problem for years in the franchise. Teams are much more aware of trying to build each position with a player of starter quality, a decent bench player and a 3rd string backup (less important but good to have). Not every team is going to be perfect, and if they were, they wouldn't be able to replicate my frustration with Detroit's roster all these past years. :P More than 3 players in a position starts making teams reconsider unless their skills and secondary position mean that they're of significant value to the team. No more "Oh, another 80+ point guard available? Never mind that we have 5 of them already, let's get one more since we can fit him under the salary cap!" I cringed so bad when someone showed me a screenshot of that and I made a point of starting to fix that logic that very same day.

You can see how a system like this gets really complicated. I'm very happy with our new overall formulas and logic this year, and along with the improvements to team-building (which isn't going to be perfect but we've taken strides in the right direction, trust me) gives us nice results and a nice base to continue working off of based on this year's feedback.

A lot of things have been changed this year, so I'm really looking forward to fresh feedback once you get your hands on the game.

Game: NBA 2K15Reader Score: 8/10 - Vote Now
Platform: iOS / PC / PS3 / PS4 / Xbox 360 / Xbox OneVotes for game: 64 - View All

NBA 2K15 Videos

Member Comments

# 61 eko718 @ 09/04/14 11:55 AM

I really don't understand the desire to remove overalls. There won't ever be a perfect formula, because who knows what the perfect formula is? But the harm of having overalls is unclear to me. If they were to be removed, there would still need to be an alternative to organize players... hard to imagine a fantasy draft and even prospect scouting in Association with no such system of organization. There will always need to be a method by which to arrange players, whether it be overall, letter grade, etc. A letter grade would be ambiguous and potentially spark more debate. I prefer the specificity of the overall rating.

# 62 NDAlum @ 09/04/14 12:06 PM

Quote:

Originally Posted by eko718


	I really don't understand the desire to remove overalls. There won't ever be a perfect formula, because who knows what the perfect formula is? But the harm of having overalls is unclear to me. If they were to be removed, there would still need to be an alternative to organize players... hard to imagine a fantasy draft and even prospect scouting in Association with no such system of organization. There will always need to be a method by which to arrange players, whether it be overall, letter grade, etc. A letter grade would be ambiguous and potentially spark more debate. I prefer the specificity of the overall rating.

Who said there was any harm to having overalls?

Players in today's NBA don't have any overalls attached to them. Rookies in the scouting combine don't have overalls attached to them.

Can't you look at individual attributes and make your own determination? The overall rating is just a number and has no substantial meaning to me. I'm confident I could go in there and make a 90 OVR player that nobody would want on their team.

# 63 knick9 @ 09/04/14 12:09 PM

I'm lost but I know a couple of things:

In the past users look mainly at speed, dunk, and 3pt shooting ratings because they are the most valuable.

And signature skills SHOULD factor into a player's overall rating because some are really powerful. Two equal players shouldn't be rated equal if one has posterizer and finisher and the other has nothing.

# 64 seanbarkley @ 09/04/14 12:13 PM

Quote:

Originally Posted by NDAlum


	Who said there was any harm to having overalls? Players in today's NBA don't have any overalls attached to them. Rookies in the scouting combine don't have overalls attached to them. Can't you look at individual attributes and make your own determination? The overall rating is just a number and has no substantial meaning to me. I'm confident I could go in there and make a 90 OVR player that nobody would want on their team.

...to you (and me and a lot of OS users), but it has for most of the 'not so hardcore' fans/players of this game. Many people don't have the time or simply don't want to go so deep into the game, just that, and take into account that those are the vast majority. IMO 2K is combining both worlds perfectly with the system they are developing.

# 65 Real2KInsider @ 09/04/14 12:18 PM

Quote:

Originally Posted by Da_Czar


	Interesting but my question is what data would drive a players value per system when all players have not performed under each system. wouldn't that be mostly up for interpretation or is there existing data that shows a players effectiveness in a particular system?

I think what he means to say is more or less an evolution of the coach slider system.

For example we know D'Antoni favors spotup 3pt shooting, transition, and pick & rolls.
Prime Steve Nash would rate higher in his system, while Pau Gasol would rate lower.
The catch is not all 30 teams are coached by someone like D'Antoni whose offensive approach is easily defined due to it's extremities.

It's also not necessary and just confuses things. If a player has a higher OVR in a specific system than that messes with trade values considerably. Because then teams where the player doesn't fit will be placing a higher priority on that player.

Quote:


	We don't necessarily know how a player will fare in a given system until we see him in it do we ?

There are some markers when you look at play-by-play data, but it's ultimately true.
Jamal Crawford serves as a good example. He was a combo guard before joining the Clippers and now is exclusively a SG. His efficiency has gone up because he takes considerably more spotup threes now. However, given his player history, it would have been just as likely that he'd take possessions away from Chris Paul. There isn't really any way to conclusively know until the player takes the court. Kevin Garnett didn't look like such a great pickup when the season started, did he?

Quote:


	To me it just initially I think that would cause as many issues as it solves especially since it could be mostly user opinion that would drive it.

In short, yes.

# 66 eko718 @ 09/04/14 12:30 PM

Quote:

Originally Posted by NDAlum


	Who said there was any harm to having overalls? Players in today's NBA don't have any overalls attached to them. Rookies in the scouting combine don't have overalls attached to them. Can't you look at individual attributes and make your own determination? The overall rating is just a number and has no substantial meaning to me. I'm confident I could go in there and make a 90 OVR player that nobody would want on their team.

And how do you translate a persons eye test determination of a player's skill in real life into a video game?

# 67 NDAlum @ 09/04/14 12:34 PM

To me it's not that difficult to quickly scroll through ratings and get an idea of what a guy is good and/or bad at.

# 68 eko718 @ 09/04/14 12:48 PM

Quote:

Originally Posted by NDAlum


	To me it's not that difficult to quickly scroll through ratings and get an idea of what a guy is good and/or bad at.

Sure, as can I, but that's not the point I was making. There will still be need to organize players in the game. Would you prefer when scouting prospects in Association for the players to be randomly arranged to where you would have to go in and blindly scout guys to discover who the top players are? There will always need to be a system of organization.

When I said I don't see the "harm" in having an overall rating, I mean I don't understand the argument for removing it... maybe you can elaborate on why removing it would be beneficial?

# 69 NDAlum @ 09/04/14 01:01 PM

Prospects:
Give projections and player comparisons.

How to organize:
List them alphabetically or by projected starters first. They can have OVR under the hood, I just don't like the visual attribute as I believe it is doesn't hold much substance. You can have two 90 rated guys and they could be polar opposites

Issue I see with overall rating:
I see guys use it to justify ripping off the CPU in all sports games that I play (NBA/MLB/NFL). I think people put way too much stock into OVR.

# 70 JazzMan @ 09/04/14 01:20 PM

So we can potentially see CPU teams go after AND start the guys who are one-dimensional yet incredibly valuable to the team, like Kyle Korver or Tony Allen?

This is great news.

# 71 eko718 @ 09/04/14 01:23 PM

Quote:

Originally Posted by NDAlum


	Prospects: Give projections and player comparisons. How to organize: List them alphabetically or by projected starters first. They can have OVR under the hood, I just don't like the visual attribute as I believe it is doesn't hold much substance. You can have two 90 rated guys and they could be polar opposites Issue I see with overall rating: I see guys use it to justify ripping off the CPU in all sports games that I play (NBA/MLB/NFL). I think people put way too much stock into OVR.

And that's my point I guess, as this seems to be the main issue folks have with the overall rating. I understand that, but ultimately, there is no concrete advantage to removing it, especially when the reason is one of visual/subjective annoyance. Whether under the hood or visible, it or some iteration of it is necessary. It drives so many systems in the game that it would probably break the game to remove it completely.

Maybe they can implement an option to visually turn it off, but either way they seem to be moving in the right direction with trying to make it more specific and meaningful.

# 72 BobbyColtrane @ 09/04/14 01:31 PM

Quote:

Originally Posted by Da_Czar


	Interesting but my question is what data would drive a players value per system when all players have not performed under each system. wouldn't that be mostly up for interpretation or is there existing data that shows a players effectiveness in a particular system? We don't necessarily know how a player will fare in a given stystem until we see him in it do we ? What are the definitive markers you would use to clearly define each system especially when some guys either have no discernable "system" or run a mixture of both. To me it just initially I think that would cause as many issues as it solves especially since it could be mostly user opinion that would drive it. So not saying it couldn't work just asking how would it work and if that would really be better ? Another thread maybe ? IDK LOL. ps apologies for typo's I am using a phone on a treadmill.

Czar off the treadmill and get back to work I want those playbooks TIZZIGHT!!!!!

# 73 franzis @ 09/04/14 02:15 PM

Quote:

Originally Posted by Da_Czar


	Interesting but my question is what data would drive a players value per system when all players have not performed under each system. wouldn't that be mostly up for interpretation or is there existing data that shows a players effectiveness in a particular system? We don't necessarily know how a player will fare in a given stystem until we see him in it do we ? What are the definitive markers you would use to clearly define each system especially when some guys either have no discernable "system" or run a mixture of both. To me it just initially I think that would cause as many issues as it solves especially since it could be mostly user opinion that would drive it. So not saying it couldn't work just asking how would it work and if that would really be better ? Another thread maybe ? IDK LOL. ps apologies for typo's I am using a phone on a treadmill.

I will keep it simple and realistic:
- first of all, you have to simplify the notion of "team system" taking as example the Team DNA in NBA LIVE 09/10: a kind of breakdown of possessions by type of plays (P&R, ISO, 3pt, etc etc) based on actual NBA stats
- you have to describe the single player style as breakdown of FG% by type of plays (as above) based on actual NBA stats
- when Player X who has his best FG% in Play Y, is traded from Team W to Team Z, IF Team Z uses more (less) Play Y than Team W, THEN Player X receive a boost (penalty) to his offensive ratings (and thus to Overall)...same logic of the Floor General Sig Skill

Obviously it has some limits (too much emphasis on scoring, no attention to defence) that requires extra reasoning, but it seems pretty feasibile to me and I believe that something like that is already in the 2k code (when plays are automatic assigned to players according to their attributes)

# 74 stillfeelme @ 09/04/14 03:10 PM

Quote:

Originally Posted by franzis


	I will keep it simple and realistic: - first of all, you have to simplify the notion of "team system" taking as example the Team DNA in NBA LIVE 09/10: a kind of breakdown of possessions by type of plays (P&R, ISO, 3pt, etc etc) based on actual NBA stats - you have to describe the single player style as breakdown of FG% by type of plays (as above) based on actual NBA stats - when Player X who has his best FG% in Play Y, is traded from Team W to Team Z, IF Team Z uses more (less) Play Y than Team W, THEN Player X receive a boost (penalty) to his offensive ratings (and thus to Overall)...same logic of the Floor General Sig Skill Obviously this system has some limits (too much emphasis on scoring, no attention to defence) that require extra reasoning, but it seems pretty feasibile to me and I believe that something like that is already in the 2k code (when plays are automatic assigned to players according to their attributes)

The system thing can be confusing but when I think system it is usually:

Triangle, Motion, Princeton offense, Pick and roll. How do you tell what player can do well in a triangle vs. another offense especially if a player has never played under that system? I think the problem lies with the team/coach/philosophy.

The offense/defense suffers if you place your players in situations that don't line up with their strengths. The player doesn't lose actual value at the position based off the scheme though. Think of this some players play a large portion of their college career in zone defense, does this mean they can't play man to man how?

# 75 franzis @ 09/04/14 03:25 PM

Quote:

Originally Posted by stillfeelme


	The system thing can be confusing but when I think system it is usually: Triangle, Motion, Princeton offense, Pick and roll. How do you tell what player can do well in a triangle vs. another offense especially if a player has never played under that system? I think the problem lies with the team/coach/philosophy.

I agree with you but I don't think it's possible to implement that in the game in such terms (as Czar wondered).

That's why my first assumption was to simplify what we mean with "team system" in 2k perspective

# 76 threattonature @ 09/04/14 05:07 PM

Quote:

Originally Posted by Da_Czar


	Interesting but my question is what data would drive a players value per system when all players have not performed under each system.

It is what teams do now when searching for free agents, just making blind guesses over who will fit into the style they play. My thought is more for myleague/myteam. Each team should have their own variables they look for to calculate the overall rating based on a system. Most coaches have a basic system in place rather it's more uptempo, pick and roll, 3 point shooting, inside/out. So the data that would drive their value in different systems are their existing ratings.

Quote:

Originally Posted by Da_Czar


	wouldn't that be mostly up for interpretation or is there existing data that shows a players effectiveness in a particular system? We don't necessarily know how a player will fare in a given stystem until we see him in it do we ?

What I'm suggesting is just using the existing attributes and the system would determine the overall. So a player with high speed, quickness, jumping, steals would rate higher in an uptempo system or be more valuable in that type of system compared to being in a Spurs or Memphis system where that athleticism would not matter near as much. Basically what I'm suggesting is a way to give different teams, a more distinct identity and style.

Quote:

Originally Posted by Da_Czar


	What are the definitive markers you would use to clearly define each system especially when some guys either have no discernable "system" or run a mixture of both.

Which teams do you think don't have an identifiable system? I think especially now with the SportVu tracking so much data that it's definitely possible to gather what type of style each team plays and what types of skillsets would fit in each one.

Quote:

Originally Posted by Da_Czar


	To me it just initially I think that would cause as many issues as it solves especially since it could be mostly user opinion that would drive it. So not saying it couldn't work just asking how would it work and if that would really be better ? Another thread maybe ? IDK LOL. ps apologies for typo's I am using a phone on a treadmill.

I do agree with how everything works it would be extremely complicated but eventually could tie any many different factors. The way I see it could possibly work long term is in myGM it could affect the type of coaches that each owner would hire. The coach based on the system he's running can determine which players they attempt to go after in free agency and on the flip side and play a factor in free agents determining what team they choose to sign with.

I think the one thing missing in association mode is teams having individual personalities or styles after a year or two because they're all going after the same players based off the same overalls.

# 77 Da_Czar @ 09/04/14 09:26 PM

Quote:

Originally Posted by threattonature


	It is what teams do now when searching for free agents, just making blind guesses over who will fit into the style they play. My thought is more for myleague/myteam. Each team should have their own variables they look for to calculate the overall rating based on a system. Most coaches have a basic system in place rather it's more uptempo, pick and roll, 3 point shooting, inside/out. So the data that would drive their value in different systems are their existing ratings. What I'm suggesting is just using the existing attributes and the system would determine the overall. So a player with high speed, quickness, jumping, steals would rate higher in an uptempo system or be more valuable in that type of system compared to being in a Spurs or Memphis system where that athleticism would not matter near as much. Basically what I'm suggesting is a way to give different teams, a more distinct identity and style. Which teams do you think don't have an identifiable system? I think especially now with the SportVu tracking so much data that it's definitely possible to gather what type of style each team plays and what types of skillsets would fit in each one. I do agree with how everything works it would be extremely complicated but eventually could tie any many different factors. The way I see it could possibly work long term is in myGM it could affect the type of coaches that each owner would hire. The coach based on the system he's running can determine which players they attempt to go after in free agency and on the flip side and play a factor in free agents determining what team they choose to sign with. I think the one thing missing in association mode is teams having individual personalities or styles after a year or two because they're all going after the same players based off the same overalls.

Good stuff! Def something for us to discuss.

# 78 Colts18 @ 09/04/14 09:38 PM

Or the AI's gm could look for players with certain grades in specific attributes. Like a gm of the Celtics is looking for a PG with at least a B in athleticism. And this should be in the coach profile

# 79 rgp913 @ 09/04/14 10:04 PM

It sounds like they are just removing the filter of position in determining overall rating. Essentially- ratings will be calculated pretty much the same, but instead of a given player's position dictating which of the 5 potential overalls they receive (either pg,sg,sf,pf, or c overall) they tweaked it so a players overall is the highest of the 5 formulas. They simply eliminated a position penalty and put the onus on the ratings. I don't think it's as in depth and evolved as some of you are guessing.

# 80 bedwardsroy19 @ 09/04/14 10:49 PM

Quote:

Originally Posted by rgp913


	It sounds like they are just removing the filter of position in determining overall rating. Essentially- ratings will be calculated pretty much the same, but instead of a given player's position dictating which of the 5 potential overalls they receive (either pg,sg,sf,pf, or c overall) they tweaked it so a players overall is the highest of the 5 formulas. They simply eliminated a position penalty and put the onus on the ratings. I don't think it's as in depth and evolved as some of you are guessing.

I'm not sure where this assumption is coming from. As I've said a few times in the thread, there are a variety of Overall formulas that are now being calculated per position to better represent unique players that may not be all around players for their position. These players are still very valuable to a team and in 2K15 their rating will reflect that value

« Previous 1 2 345 6 7 Next »

More Links

Revamped Overall Formula for NBA 2K15 Player Ratings Explained