Incorrect Left and Right Close and Midrange Tendencies

[eraserx13]

Anyone noticed that the Left/Right Tendencies for Close and Midrange Shot are off for 80% of the Opening Night roster?

For example, DeMar DeRozan has a left midrange tendency (T/SML) of 11 while his right (T/SMR) of 57. So his shot chart is skewed to the right side of the court. After running a sim, his shots from the right side is at around 70%.

However, when you look at his actual shot chart from the 18-19 season, its fairly even on both sides.

https://www.basketball-reference.com/play-index/shooting.fcgi?player_id=derozde01&year_id=2019&shot_distance=16-3P

Has anyone tried to address this by creating a roster? If so, which system?

[STLRams]

Re: Incorrect Left and Right Close and Midrange Tendencies

Not surprised, alot of the players tendencies dont seem true to life at all.

[infemous]

Re: Incorrect Left and Right Close and Midrange Tendencies

I haven’t researched it (statistically/numerically) but Lebron has always favoured shooting on the left side of the court but for some reason opening rosters had his hot spot on the right.

Also the fact Lebron has defensive leader to the point where the percentages come up under him is a joke.

I do genuinely believe that by putting all sorts of things into badges they’ve fundamentally ruined their ability to accurately stay on top of things like ratings, tendencies etc. And as the focus of gameplay has gone to my career it has chipped away at the basic tenets of the game that gave this game the success it currently enjoys.

I will not buy 2k on PS4 again and tbh they need to really find some technology that can automatically update all of these things because one man cannot and should not be tasked with this.

I remember discussing so frequently how Madden couldn’t have one dude responsible for all ratings and it’s the same for 2k. There are so many moving parts now the poor dude is set up to fail and the game suffers.

I don’t play PNO, MyTeam or MyLeague anymore because the latency makes it a slog where I’m constantly wondering why my player didn’t do what I input - and then whenever I look into why that was the case, more often than not some terrible rating/hot zone or badge is at fault.

Don’t get me started on rotations either - it’s gotten even worse in 2k20 and it has always been bad.

[Real2KInsider]

Re: Incorrect Left and Right Close and Midrange Tendencies

Anyone noticed that the Left/Right Tendencies for Close and Midrange Shot are off for 80% of the Opening Night roster?

For example, DeMar DeRozan has a left midrange tendency (T/SML) of 11 while his right (T/SMR) of 57. So his shot chart is skewed to the right side of the court. After running a sim, his shots from the right side is at around 70%.

However, when you look at his actual shot chart from the 18-19 season, its fairly even on both sides.

https://www.basketball-reference.com...distance=16-3P

From what I understand, they used shooting data to generate the shot location tendencies this year. If you'll notice, each zone should add to 100. So for example that should be looked at as 11% and 57%.

I have no idea what data-source got used if that's the numbers being generated, however. Could have been over a multi-year period. Also bears noting that Mid-range on BBRef is not precisely what the NBA uses (which I assume is what 2K uses).

[eraserx13]

Re: Incorrect Left and Right Close and Midrange Tendencies

From what I understand, they used shooting data to generate the shot location tendencies this year. If you'll notice, each zone should add to 100. So for example that should be looked at as 11% and 57%.

I have no idea what data-source got used if that's the numbers being generated, however. Could have been over a multi-year period. Also bears noting that Mid-range on BBRef is not precisely what the NBA uses (which I assume is what 2K uses).

Indeed. The values as a percentage should add up to 100. However the values under the rating tendencies do not necessarily need to. In 2k19 values never got close to 100 and the values in 2k20 are all over the place. From what I can tell, the percentages that you see in each zone in a players shot chart are governed by the rating of a single zone divided by the sum of the ratings of all zones.

That said, the rating tendencies and the resulting shot chart of each player after simming a full season are no where near what happened in real life. Harden shoots most of his 3s from the top of line as opposed to the corners. If you look at his shot chart after a season of sim, 70 to 80% of his shots come from the right corner from deep. So if the data is translated into a rating to form the tendencies, then whoever at 2k did a real poor job of porting.

In real life I am a data analyst and I see this type of shoddy work all the time. It appears that there might have been a data the issue where the source has a different data type than what is accepted by the actual roster. Because of this, the shot location ratings for most of the players are not consistent and is resulting in skewed results as evident by the shot chart leaning heavily to just one or two zones. Whatever the cause, this is just terrible unit testing and QA control by 2k. In other words, no one bothered to test the data.


Sent from my iPhone using Operation Sports

[eraserx13]

Re: Incorrect Left and Right Close and Midrange Tendencies

I haven’t researched it (statistically/numerically) but Lebron has always favoured shooting on the left side of the court but for some reason opening rosters had his hot spot on the right.



Also the fact Lebron has defensive leader to the point where the percentages come up under him is a joke.



I do genuinely believe that by putting all sorts of things into badges they’ve fundamentally ruined their ability to accurately stay on top of things like ratings, tendencies etc. And as the focus of gameplay has gone to my career it has chipped away at the basic tenets of the game that gave this game the success it currently enjoys.



I will not buy 2k on PS4 again and tbh they need to really find some technology that can automatically update all of these things because one man cannot and should not be tasked with this.



I remember discussing so frequently how Madden couldn’t have one dude responsible for all ratings and it’s the same for 2k. There are so many moving parts now the poor dude is set up to fail and the game suffers.



I don’t play PNO, MyTeam or MyLeague anymore because the latency makes it a slog where I’m constantly wondering why my player didn’t do what I input - and then whenever I look into why that was the case, more often than not some terrible rating/hot zone or badge is at fault.



Don’t get me started on rotations either - it’s gotten even worse in 2k20 and it has always been bad.

I totally agree with you. I made the jump from 2k19 so I didn’t have to deal with some of the bugs like the under the basket tendency or importing player DNA and the messed up season stats. Those got fixed in 2k20. Little did I know that this new game introduced a whole new set of bugs. I am constantly fiddling with the rosters so I can get realistic sim numbers. The badges really messed up the sim engine results. I have spent more time fiddling than actually playing a game. It’s BS when it comes to that and I am at that point where it is not worth it anymore. I am pretty sure this is my last year playing.


Sent from my iPhone using Operation Sports

[Real2KInsider]

Re: Incorrect Left and Right Close and Midrange Tendencies

Indeed. The values as a percentage should add up to 100. However the values under the rating tendencies do not necessarily need to. In 2k19 values never got close to 100 and the values in 2k20 are all over the place.

DeRozan
Close Tendencies (L-C-R): 15-14-71 (= 100)
Mid Tendencies (L-LC-C-RC-R): 11-23-8-1-57 (= 100)
3pt Tendencies (L-LC-C-RC-R): 20-4-42-14-20 (= 100)

In 2K19 and 2K18 these WERE all over the place. Clearly there was a philosophical shift in how these are approached. With that said I am unsure why DeRozan (or ANY scoring wing) would have such a high Mid tendency from the right corner.

We all can all agree that this doesn't appear to be based on 2019 data, and as such is worth investigating further.

From what I can tell, the percentages that you see in each zone in a players shot chart are governed by the rating of a single zone divided by the sum of the ratings of all zones.

If you mean on the NBA site:

https://stats.nba.com/events/?flag=3&CFID=33&CFPARAMS=2018-19&PlayerID=201942&TeamID=0&GameID=&ContextMeasure=FGA&Season=2018-19&SeasonType=Regular%20Season&LeagueID=00&PerMode=PerGame&Split=general&PlusMinus=N&PaceAdjust=N&Rank=N&Outcome=&Location=&Month=0&SeasonSegment=&OpponentTeamID=0&VsConference=&VsDivision=&GameSegment=&Period=0&LastNGames=0&DateFrom=&DateTo=&PORound=0&ShotClockRange=&MeasureType=Base&section=player&sct=zone

DST = % of the player's FGA. For example 3.2% for DeRozan in the right corner means his 42 shots were 3.2% of all his FGs, not of his mid-range FGs.

317 Mid-range FGA for DeRozan
His Mid Tendencies SHOULD have equaled roughly:
9-24-30-24-13
but instead:
11-23-8-1-57

That said, the rating tendencies and the resulting shot chart of each player after simming a full season are no where near what happened in real life. Harden shoots most of his 3s from the top of line as opposed to the corners. If you look at his shot chart after a season of sim, 70 to 80% of his shots come from the right corner from deep. So if the data is translated into a rating to form the tendencies, then whoever at 2k did a real poor job of porting.

Can you elaborate further, as I am unclear what you mean by translating into a rating. I don't believe that is the case as I'm understanding it.

In real life I am a data analyst and I see this type of shoddy work all the time. It appears that there might have been a data the issue where the source has a different data type than what is accepted by the actual roster. Because of this, the shot location ratings for most of the players are not consistent and is resulting in skewed results as evident by the shot chart leaning heavily to just one or two zones. Whatever the cause, this is just terrible unit testing and QA control by 2k. In other words, no one bothered to test the data.

I can buy that.