Friday, October 15, 2010

Understanding NBA 2K's Tendencies and Synergy Data

As most NBA 2K gamers know by now, the tendencies in NBA 2K11 are preventing gamers from enjoying a realistic experience. This is a major issue, because these tendencies are so ambiguous except to a select few world-wide (including 2K Sports' staff) that very few have the sufficient knowledge required to accurately evaluate them for a single team (much less the entire league). 2K Sports has kept people in the dark regarding how these tendencies were/are formulated. Now that the tendencies are greatly effecting gameplay, gamers are essentially being held hostage by the imaginary man behind the curtain.

I am hopeful that I can use my summer's Synergy Sports research and find the solution.

First, let's list all the tendencies that are in NBA 2K11.

Shot Tendency
Inside Shots
Close Shots
Mid-Range Shots
3pt Shots
Putback Dunks

Drive the Lane

Pull Up

SETUP PRE-DRIBBLE (total of 100)
Pump Fake
Triple Threat
No Triple Threat

SETUP DRIBBLE (total of 100)
Straight Dribble

Drive Right

Step Back
Half Spin
Double Cross
Behind the Back
Hes Cross
In And Out
Simple Drive

Pass Out

FINISHING MOVE (total of 100)
Hop Step
Spin Layup
Simple Drive

Stepback Jumper
Pull Up Jumper
Reverse Pivot Jumper

Inside Upfake

Triple Threat Shot
Self Post
Alley Oops
Use Glass
Draw Foul

Use Pick
Set Pick
Use Offball Screen
Set Offball Screen
Spot Up
Post Up
Give And Go

POST MOVES (total of 100)
Quick Spin
Drop Step
Step Through
Up And Under
Face Up
Back Down
No Move

POST SHOT (total of 100)
Hook Shot
Roll Shot
Step Away

Flashy Passes
Throw Alley-Oop

Hard Foul
Take Charge
Play Pass Lane
On Ball Steal
Contest Shot
Commit Foul

Geez, that's a lot.

Now, many of these tendencies are just cosmetic, and for all intents and purposes are not impacting the gameplay. "Dribble Penetration" and "Shot Type" tendencies that dictate what moves a player performs are all well and good, but they don't make one bit of difference if the player never has the ball in his hands. For instance, it would annoy gamers if Dirk Nowitzki took a hook shot every single time he posted up (NBA 2K5 ring a bell?), but it's more important that he's actually posting up the proper amount before worrying about how it's going to look.

So let's take all those styles out of the gameplay equation. That leaves us with Shooting Tendencies, Drive the Lane, Inside Shots, Various Shots, Freelance, and Defense as the main tendencies that are affecting the gameplay.

Truth be told, 2K has done a good job (though not perfect) with the shooting tendencies. NBA Hotspots is not a difficult reference to use, after all (at least when it's operational - it won't be again until the regular season starts). The most important thing is that we can actually say these tendencies are good, because they are quantifiable by statistics. Drive the Lane and Inside Shots (Dunks) can also be calculated using statistics. Even Defense can be calculated via advanced statistics! So that leaves us with Freelance and Various Shots as the two forms of tendency that we have very little knowledge on.

Or do we? This summer Synergy Sports Technology made available it's expansive database with data from the 2010 season. Among the data that we can see?

Pick and Roll - Ball Handler
Pick and Roll - Roll Man
Spot Up
Post Up

Do these look familiar? They should. They're the freelance tendencies. This was a major breakthrough for me because I realized that many of 2K's Freelance tendencies were just thrown together. There are many players (especially centers) who have 0 in all categories. Implementing this data into my rosters is what helped me improve upon 2K's gameplay last season. Now, it might not only be the key to saving this year's game, but the key to the perfect basketball simulation.

Here are some freelance tendencies for the New York Knicks for us to pick apart.

95 Ewing Jr.
94 Stoudemire
87 Felton
83 Randolph
66 Turiaf
50 Mason
49 Azubuike
20 Gallinari
17 Douglas
16 Chandler
11 Mozgov
10 Walker
10 Curry
10 Rautins
0 Williams

Patrick Ewing Jr. (95) vs Shawne Williams (0)
The first thing you surely noticed was Patrick Ewing's son with 95 Iso tendency - greater than the Knicks' star player, Amare Stoudemire. This appears to be a fluke, or at the very least, some sort of joke. Ewing Jr's tendencies are significantly different from last season (they are high enough to be Ewing Sr's, so I presumed they gave him his father's tendencies by accident, but this was not the case). Bear in mind that he didn't play at all last season (and has never even played in the NBA, in fact), and that 2K had no reason to adjust them. I'll give them the benefit of the doubt and presume this was a computer error. Now surely most of you don't care about Jr's tendencies since it's so rare that he sees the court in the game. However, I saw him get into a game due to Amare's foul trouble, and he proceeded to brick 3 shots in 5 minutes in the 4th quarter of a close game. Stuff like that cheapens the gameplay experience, which is why we are harping on it now. But like I said, I'll give 2K the benefit of the doubt and won't harp on it again.

Now, at the very bottom we have Shawne Williams. Williams has 0 in all Freelance tendencies EXCEPT for the new tendencies that 2K added this season (Set Offball Screen, Give And Go, and Touches). Now you might say, Williams is unlikely to make the Knicks, who cares? Fine. I'm not going to go crazy over editing tendencies for a player that will never see the court. Fine. However, you might also presume that he has those tendencies because they ignored players in the free agent pool. This is NOT the case. Williams was on an NBA roster last season, and played in the NBA for 3 years prior to that. His freelance tendencies should have been formed with 2009 data, but this never happened, and there were MANY OTHERS in 2K10 like Williams who shared the same "0 Freelance" fate. I have not checked every player to confirm this, but that's not my point anyway. As all 2K vets know, 2K spends a great deal of time editing ratings for star players, but has barely touched most backup players in the last 3 years. It only makes sense that their tendencies are just as ignored. Most backups have Freelance tendencies that are simply not accurate.

Williams and Ewing are unlikely to make the team, but they are the most extreme examples on this list. Let's compare some actual rotation players on the Knicks now, since that will be having a more direct impact on gameplay experiences.

Raymond Felton (87) vs Toney Douglas (17)
I find this one a bit odd because Felton is considered a "pass first" PG while Douglas is considered a "scoring" PG. This is simply 2K Sports not knowing the players.

According to Synergy, 49 of Douglas' 471 plays (10.4%) were isolations. Conversely, Synergy lists 126 of Felton's 1127 plays (11.2%) as isolations. Using this data, it doesn't take a math wizard to figure out that a 70 point discrepancy is incredibly off-base.

What makes this completely inexcusable, is that 2K Sports is a million dollar company, and access to this data cost me a mere $30. They could buy TWO subscriptions for the cost of ONE sold game. If 2K is looking to save money they should just hire me and I'll gladly continue to pay for my own Synergy account :)

Ronny Turiaf (66) vs Wilson Chandler (16)
Anyone who knows anything about the NBA knows this is wrong. Turiaf is a non-scorer and any offensive havoc he causes is a direct result of his off-ball contributions (Hustle plays and pick and rolls). Chandler on the other hand is a SF that is athletic/skilled enough to play SG and PF as well. He may not have the greatest handle, but he certainly attacks off the dribble much more than Turiaf. Off to Synergy for some confirmation.

Turiaf: 14 Isos in 221 plays (6.3%)
Chandler: 114 Isos in 1032 plays (11.0%)

It's also worth noting that Turiaf posted a meager 0.43 Pts per Possession on Iso's (3-11 FG, 7.1% FoulDraw, 14.3 TO%). This was clearly not something he looked to do, and was only used as a last resort (i.e. with the clock running down). Let's look at the rest of Turiaf's Synergy to get an idea of what he does on offense.

0.0% P/R Ball Handler
30.3% P/R Roll Man
6.3% Isolation
0.5% Off Screen
6.3% Spot Up
9.0% Post Up

5 Use Pick
98 Set Pick
66 Isolation
10 Use Screen
5 Spot Up
56 Post Up

Virtual Turiaf CLEARLY won't behave the same as Real Turiaf. If Real Turiaf actually has the ball in his hands, he will look to post up, not take his man off the dribble. Meanwhile he catches and shoots just as frequently (or should I say, infrequently) as he goes 1-on-1, but that won't be the case with Virtual Turiaf in NBA 2K11.

Chandler on the other hand is Isoing just as much as Felton and Douglas. One of these players is an 87. The other two are 17 and 16. Clearly we need to find a a balance in where these guys should be rated. Should they all be in the teens? Should they all be in the 80s? Or should they meet somewhere in the middle? To find that out, we need to compare more than just two players at a time.

17.2% Randolph (83)
14.8% Stoudemire (94)
13.5% Gallinari (20)
11.2% Felton (87)
11.0% Chandler (16)
10.4% Douglas (17)
9.1% Azubuike (49)
8.8% Walker (10)
7.7% Mason (50)
6.3% Turiaf (66)
2.3% Curry (10)

As you can see, these values are all over the place. 3pt specialists Mason and Azubuike will go 1-on-1 much more than Chandler & Gallinari (which, along with "Touches" helps explain why they never looked to score in the simulations that I ran), while Walker is sandwiched between the two and has a lower Iso tendency than anyone on the team sans Curry.

What we do know is that Randolph & Stoudemire are both tops on the team and have 80+ Iso. Felton isn't close to them which means his 87 needs to be dropped, while Chandler/Douglas need to be raised. If we bring them to Azubuike/Mason's level, it means they need to be dropped, and the question becomes, how far?

Now you can all see why editing tendencies is a chore, even with the data at your finger tips. If you don't know how to apply data, it's useless. Fortunately, I was able to develop a system & scale that many of you played for yourself this summer.

Speaking of scales, does 2K have a system in place? Let's compare their 2K11 Iso tendencies to 2K10 Iso tendencies.

59 Randolph
49 Felton
46 Gallinari
30 Mason
20 Azubuike
19 Stoudemire
7 Chandler
5 Turiaf
0 Douglas - Like I said, there were plenty of these
0 Walker

Judging by this set... I mean, it looks like 2K just used guess work this year. If anything, 2K10 actually has the better tendencies! While compiling my 2K10 tendencies I was under the impression that 2K used 2009 synergy data... I cannot say the same for this season. Looking at 2K10's tendencies without even knowing the synergy data I'd say the only guys who needed changes were Amare (way too low), Chandler (should at least be higher than Roger Mason, not 2 points ahead of Turiaf), and Douglas/Walker (who obviously deserve better than zero). Yet 2K11 looks like they threw their tendencies into a blender... and then advertised them as a masterpiece.

All this talk and effort, and this is just ONE tendency for ONE team. Synergy data is all but required to accurately adjust Freelance tendencies for the entire league. The good news is that it can be done (and I already HAVE done it). As I maintained throughout the summer, my research WILL translate, and is practically required now since it appears 2K's tendencies got EVEN WORSE.

(Aside: Since I'm the only one with that data, it's in every video game basketball junkies best interest that I not drop dead. Or get married. Whichever is worse).


Ah, yes, you're right. The dreaded "touches" tendency that has lead to star players averaging 40 shots per game without fail.

Let's just say Synergy will help me out with that one too.

As anyone who played Ogre Battle: March of the Black Queen will tell you... the magician never reveals his secrets :)


Shelb said...

Hire this man!

Mike said...

Interesting read. Even though I think you're very arrogant at times I believe you definitely deserve the job over the guy who thinks Jon Brockman is an below average rebounder.

And please don't get married. I couldn't survive with 2K's rosters.

Rashidi said...

LOL, so you'd rather see me dead than married?

Actually, that might sum up the 2K communities' general feeling towards me perfectly.

Anonymous said...

Hey Rashidi you the man bro when you finally update for ur first will be cant wait. Haveb even start my association cant do it with 2f horrible offical rosters.

Mike said...

lol, that came out the wrong way. I'm sure deep down inside the people who insult you appreciate your work. They just don't realize it yet.

Anonymous said...

Very impressive.

Anonymous said...

lol, and people think nogster has superior tendencies.

he just does a 'once over' on a old roster file and he's done,for the year. and yet ppl still use what will be a dated roster by opening day to play multiple seasons with.

this my friend will be the ultimate roster once tendencies are adjusted tested and adjusted once more. i would only play online if not for this roster.
last year 2k11 for 2k10 balled like playing a super-sim human and if you had pull up j's (read:midrange game) i might still be playing it.

Oliver Baden said...


Only really new to the '2k community' so you definitely have a clean slate with me, don't see any of this supposed arrogance. At least you are walking the walk and doing something about it, it would be arrogant to sit behind a computer and complain about 2k but not do anything. So kudos.

Being a bit of a self-confessed noob because im so new, I was just wondering if your roster updates only impact on tendencies of players? They don't affect sliders right? i suppose the two are connected to some extent in that they impact gameplay and CPU decisions, but not directly related?

also is your update for PS3 (please say yes lol)

cheers man

Rashidi said...

Being a bit of a self-confessed noob because im so new, I was just wondering if your roster updates only impact on tendencies of players?

I am more known for my ratings, which are much more fun to discuss (as well as edit). This year I'm spending way too much time on tendencies. Most of my ratings are the same as last year (since end of year stats haven't changed), so I've spent most of the last 2 weeks trying to figure out this terrible gameplay. Only I'm starting to realize that some of it is unfixable.

They don't affect sliders right?

No, although I'm starting to believe this game simply cannot be played on default sliders.

also is your update for PS3 (please say yes lol)

They will be for both systems. I am working on 360, while forwarding my updates to another user for PS3.

They should also be for PC once the proper editing tool is released.

Anonymous said...

Hi i have noticed that your rosters are very promising, so i would like to know if there are disadvantages to using your rosters instead of the official ones? Can we play online with your rosters? And will the nba today games show with your rosters? sorry for being a noob