NBA 2k17 MyLeague After 5 Years: Data Showing Problems

[LorenzoDC]

UPDATE: We may have found some solutions for this. See here and here. I'll update this spot if we get a good work around team training slider set published and shared.

UPDATE 2: I've posted my slider set to fix this stuff here.

After all the contributions from people on this thread one about problems in MyLeague and 3 point shooting ratings over time, I decided to do a study to broaden the view of some of the problems I heard people talking about, and from my own observations of legacy issues. I thought it would be a good idea to put a study together for the devs to see what a lot of us are talking about.

In this study, I automated everything and used automated generated draft classes, and simmed ahead five years, from the season opener of 2016 to the season opener of 2021. First I'll show the data, and then I'll make some comments.



Here are the problems I see. They are caused by a combination of the composition of CPU generated draft classes, player progression and training/development AI, and CPU roster management AI:

• Over time, the league generates fewer real stars. Look at those changes in 90+ OVR players, and their average ages. While the draft classes do generate comparable numbers of players with potential at least A- or better, significantly fewer are A+ caliber league stars.
• However, over time more of the starting players in the league possess absolutely elite open shooting ratings from mid to 3pt range. Those changes are just incredible, and they make for a complete alternate universe league that in no way resembles the NBA.
• Some of the development in open shooting rates comes at the expense of contested shooting rates, but even taking that into account, shooting is way overpowered in later years, and the longer you sim forward, the worse it gets.
• Over time, the league includes many fewer defense first or at least 2-way players. This is something I have suspected for years but only now have I generated data to show it. This is due to some combination of draft class archetype composition and player progression/training.
• CPU roster management still doesn't pay enough attention to positional balance or to balancing between front court and back court personnel depth or even cap management (though I did not include cap numbers allocated by position in this study). I have some more information about the front court/back court balance of each team with 5+ players in one position, if anyone is interested. There's also a thread on this with more data here.

When you combine all of the above, you can see the MyLeague quickly become unrecognizable as an NBA simulation within a very short number of years. It becomes a league of mostly offensive players who are not elite overall but still better 3pt and mid range shooters than Steph Curry. And without 30 team control, CPU teams will not understand how to build rosters that can get teams through full seasons with injuries, because they too often neglect positional balance in making decisions.

The good news in MyLeague at least is that we have a lot of control to make edits. The quick edit feature for attributes this year is sensational on this front. And even though it's a little thing, so is the easy right stick cycling function to move you quickly across all the teams you manage.

The bad news is the CPU AI logic on all fronts (including draft classes, player progression and training, trades, free agent signings, scouting and everything else) is so overwhelmingly flawed that you end up spending more time trying to compensate for the lack of programming in the mode than you do actually playing the game.

And that's a big problem. I assume the same problems exist in MyGM where all those editing functions are absent.

I hope anyone at 2k who sees this understands this is a tough love critique. Last year I played about 300 12 minute quarter games in MyLeague, so this is my mode, and I'm a #simnation guy through and through.

I know this has been a short development cycle and I have no doubt whatsoever that the franchise mode developers completely want to get these things right. I just don't think 2k is giving them the mission and resources to work on these things or even address legacy flaws that I'm pretty sure they know exist. And for all their welcome focus on simmed stats over time, I have no idea how they can begin to get good results from those when these issues about league composition over time are so pervasive.

[Mauer4MVP]

Re: NBA 2k17 MyLeague After 5 Years: Data Showing Problems

Is there a way to set In-Season Training to manual? Just wondering if it's from CPU focusing just on 3 point shooting when training players.

[LorenzoDC]

Re: NBA 2k17 MyLeague After 5 Years: Data Showing Problems

Is there a way to set In-Season Training to manual? Just wondering if it's from CPU focusing just on 3 point shooting when training players.

Sort of.

With 30 team control, you can take over the training of every player in the league. But in order to make that tedious process a bit more manageable, you have the create and save a number of training templates based on different archetypes, and then manually set them up to each player in the league.

Once you go through the league once, those training plans stay with the player, so for subsequent years you only have to check and update the training plans of new players in the league.

There's a catch, though. There's a limit of 20 training profiles you can create and save locally to your console, assuming you don't share them out to the community. The publicly shared save file limit is probably tighter.

The problem then is there are more than 20 generally common player archetypes you'd want to make training plans for. That means that there will always be some archetypes you will have to do over and over again manually for each player with that archetype.

It's a very labor intensive process. I've done this for 20 player archetypes but I've only customized them so far by adjusting the balance of shooting training given to open versus contested shots. But it's not enough to overcome the scale of the problem.

I have not taken the next step to flat out cut the amount of training to players allocated to shooting, regardless of their archetype, and use those training points for defense. What will happen in that case is the progression system will see that you are training against archetype and cut back on progression a bit, so you may have to compensate for that with the progression slider.

The problem withe the lack of superstars over time really has to be overcome with draft class editing, making a few more A+ guys.

You could also, as I have always done, edit draft classes to include more defensive type players, but then you need to do not just attributes, but tendencies, hot zones and badges for each player manually. That ends up being a lot of work. And even then, you'd probably have to do what you've just suggested with the training plans for every player in the league.

It's just so much to have to try to fix and figure out. It makes it so that you have to spend 3 or more hours of editing for every maybe 1 hour you play the game. At what point does it stop being a "game" to "play" with all that extra work?

[trekfan]

Re: NBA 2k17 MyLeague After 5 Years: Data Showing Problems

Is there a way to set In-Season Training to manual? Just wondering if it's from CPU focusing just on 3 point shooting when training players.

You could manually save sliders to use for all players and then set those sliders for every player in the Player Training screen but that's

A) Insanely time-consuming and
B) Not addressing the progression issue in the offseason, which still inflates the open 3pt rating



I'm going to second the OP's post with gusto and it's completely in line with the results I'm getting on my end.

That said ... damn. I was genuinely hoping this wasn't looking as bad as it seemed. I've been hunting for a combination of sliders that stymies this, but there doesn't seem to be a setting we can affect that permanently stops this; all we can do is delay it with the sliders.

I continue to believe the main culprit is the AI progression engine. I can deal with the roster logic, that's been an issue in every franchise mode in history, but the offseason progression has never been this skewed. From what I'm seeing the progression engine is way overvaluing and rewarding the open 3pt shot and not focusing on other aspects of the NBA; the two-way stars, the elite rim protectors, the guys who live on defense (Tony Allen types), etc.

Lowering the effects of progression and in-season training only delays the inevitable and still leads to an unbalanced league, one with a lack of stars and rookies not hitting their true potentials. The lack of rookie progression at or near the default level of 50 is still an issue, but bumping that up will make the open 3pt rating issue worse

This needs to be tuned by the devs on the backend. If they can do it with player contracts, I hope they can do it here. I don't see any other solutions at this point for those of us who want to try to take their league as far as possible.

[joosegoose]

Re: NBA 2k17 MyLeague After 5 Years: Data Showing Problems

Amen brother, A+ post. We need more content like this; there are a number of issues that seem obvious to anyone who spends even a little time on these modes, but collecting the data for proof (as simple as it may seem) is immensely helpful to the devs.

I think some data collection on player contracts, contract options, and cap situations would be very helpful as well. I also wonder if we should make a singular post (maybe the OP of a dedicated thread) to keep a clean, 'consensus' list of important issues that need fixing/tuning in this mode. Then of course try to make sure a dev is aware of it.

It's just so much to have to try to fix and figure out. It makes it so that you have to spend 3 or more hours of editing for every maybe 1 hour you play the game. At what point does it stop being a "game" to "play" with all that extra work?

This is precisely what's holding me back a bit, both in terms of the game and in doing more data collecting and whatnot. I catch myself spending all my time not actually 'playing' and get overwhelmed. I'll never truly be a 30 team control guy for this reason, and it's exactly why I'm only lukewarm to the new Rotation Timeline. They're great to have and I love that I have the ability to do whatever I want, whenever I want, but I don't like feeling like I have to go the whole nine yards to get a realistic league. It starts to feel more like a sandbox than an actual game to me.

[Mauer4MVP]

Re: NBA 2k17 MyLeague After 5 Years: Data Showing Problems

I guess what I was getting at was not using any training at all and just adjusting the progression and regression sliders for the MyLeague as a whole. Possible that that could be a solution?

[trekfan]

Re: NBA 2k17 MyLeague After 5 Years: Data Showing Problems

I guess what I was getting at was not using any training at all and just adjusting the progression and regression sliders for the MyLeague as a whole. Possible that that could be a solution?

Not based on what I've experimented with. I killed in-season training and relied on just progression/regression, but the issue with open 3pt rating is still there, less pronounced early on but I had over 70+ open 3pt rating shooters five years in.

[LorenzoDC]

Re: NBA 2k17 MyLeague After 5 Years: Data Showing Problems

I guess what I was getting at was not using any training at all and just adjusting the progression and regression sliders for the MyLeague as a whole. Possible that that could be a solution?

No, because that doesn't change the shape or composition of the talent in the league, just the overall level of the talent in the league.

The problems I'm documenting are about the type of talent in the league over time. Plus, you can see that over time there are already too few superstars, so cutting down on training and progression makes that worse. And it doesn't solve the problem of players being overwhelmingly offense (shooting) heavy and defense weak as player types.

[Gramps91]

Re: NBA 2k17 MyLeague After 5 Years: Data Showing Problems

Great post! This is the stuff that's more important to me than anything when it comes to these games. Really hope they can patch some fixes. I simply don't have the time to control all 30 teams. Or 36 teams as I have in my current MyLeague.

[Mauer4MVP]

Re: NBA 2k17 MyLeague After 5 Years: Data Showing Problems

No, because that doesn't change the shape or composition of the talent in the league, just the overall level of the talent in the league.

The problems I'm documenting are about the type of talent in the league over time. Plus, you can see that over time there are already too few superstars, so cutting down on training and progression makes that worse. And it doesn't solve the problem of players being overwhelmingly offense (shooting) heavy and defense weak as player types.

Right, but that goes back to my original question. Is the problem that the CPU is focusing all in-season training on 3 point shooting? If so, you could increase progression rate without having that issue surface.

Maybe you already addressed that and I just misunderstood or missed it as a whole.

[joosegoose]

Re: NBA 2k17 MyLeague After 5 Years: Data Showing Problems

Right, but that goes back to my original question. Is the problem that the CPU is focusing all in-season training on 3 point shooting? If so, you could increase progression rate without having that issue surface.

Maybe you already addressed that and I just misunderstood or missed it as a whole.

It seems as if both in-season training defaults and offseason player progression (which can't be adjusted at all on our end, other than the rate at which it happens) seem to skew heavily towards open shooting, while comparatively neglecting contested shooting. Because we can't control it, we're completely depending on a fix from the devs for this issue.

[ffaacc03]

Re: NBA 2k17 MyLeague After 5 Years: Data Showing Problems

Great post Lorenzo !

I think My League is still tons of ways in order to achieve the label of editability/customization the mode needs to offer to its users, which in turn helps/aids the AI Logic.

While there are certain things that can and should be improved in terms of AI logic (players contract demands and scales, balanced draft classes, not huge deviations from actual statistical status quo, etc), I believe that while we get to a confort point and even after we get to that point, users definitively need to have a say on most if not all of the AI Logic variables.

Why ? Because these are very complicated matters, flooded by subjective, situational and intricated concatenated variables, plus, real life precedence of wtf moments that could justify some of the weird results we see on the AI decissions front. Therefor, it is very intrincate to tune such AI, so we need alternate tools to withstand while we get it ...

We need to:

1. Have a say on what type of stance does a team has in respect of minutes management, trades, signs/release, draft, coaching and budget.

- We should be able to set the team on a rebuilding stance, were it considers placing lineups more concentrated on giving minutes to rookies than stablished players. Leaning more into selling aged stars and role players for picks and young assets with potential. Focused on signing role players with raw potential and looking to draft possible superstars.
In draft/sign cases we should select between a few options: go for highest potential despite position, go for top potential on a given position of need, go for highest overall skills despite position, go for top overall skills of a given position of need, go for specific skill/archetype, go for most suitable to coaching philosophy, go for star, etc.
More of the same for other stances such as selling, buying, standing pat.
Also all stances should value rookies contracts and extensions a bit more, thus rebuilding and selling stances even more than the rest.
In coaching/staff cases, we should be able to edit everything as we do with players and even more, we need to have a "create a staff".

2. We need to have a say in things such as chemestry values, rivalry and variables that have impact on the rankings and on the sim stats results, such as offensive pace/efficiency and defensive pace/efficiency.

3. We need to have a say on the trade value of players and assets such as picks, we have to be able to edit the numers of "stars" given to a pickand or player, to make sure the AI doesnt under/over values it or do overvalue/undevalues it.

4. We need to have way more than 6 players within the whole league to be categorized as untouchables (30 at least), this would go long ways into aiding and providing a more realistic AI logic.

5. If space is need after a season and players need to be erased in order to make room for new ones, give us the option to select the ones to delete, preventing to have some weird choices.

6. Let us have some say on awards, all star selections and such.

7. We need to have all players vitals, ratings, tendencies, hotspots/hotzones and badges into a quick edit screen, as well as for staff edits.

8. We need to be able to edit/select body types for players and staff.


I know it sounds as a wish list but some of these things we need would certaintly aid the AI and improve the tools users have at hand to try and make My League their own universe.

Here is to hope for an even more expanded customization and editability.

P.S: Forgive the couple of throw in requests :c

[trekfan]

Re: NBA 2k17 MyLeague After 5 Years: Data Showing Problems

It seems as if both in-season training defaults and offseason player progression (which can't be adjusted at all on our end, other than the rate at which it happens) seem to skew heavily towards open shooting, while comparatively neglecting contested shooting. Because we can't control it, we're completely depending on a fix from the devs for this issue.

That's it exactly, the way progression works is heavily skewed towards offense, particularly on the open 3pt rating. Totally understandable the devs would prioritize that, but it's just too far into an extreme to maintain a balanced league.

We need Leftos and Davez to be made aware of this and hopefully they'll be able to tune it in a patch/update in the near future, but I imagine the two of them will be busy putting out the more immediate/gamebreaking fires first.

I don't anticipate a patch addressing this until at least Oct. 20th, one month post retail release, as this isn't necessarily "gamebreaking." It is for us guys who go 4-5 years deep into MyLeague and care about maintaining the overall balance/composition of the league, but we're few and far between -- spread out over the PC/Xbox/PS4 user base.

If we get to November and this still hasn't been addressed in some way, I will be very worried, but right now we're barely 2 weeks post-early tip off, so I'm not anywhere near the panic button.

The best thing we can do is to continue to experiment with sliders, document our results, and keep those save files (I imagine 2K may need to pull a few of them to see exactly what's going on, so we'll have to use our limited slots to save the ones with important results).

[LorenzoDC]

Re: NBA 2k17 MyLeague After 5 Years: Data Showing Problems

Right, but that goes back to my original question. Is the problem that the CPU is focusing all in-season training on 3 point shooting? If so, you could increase progression rate without having that issue surface.

Maybe you already addressed that and I just misunderstood or missed it as a whole.

The CPU does not focus all training on 3 pt shooting, at least, not according to the visible training plans for each player on each team. It only puts maybe 5% of training points to shooting for good shooters, and even less for other types of players.

Nevertheless, the results of a combination of training plus off season progression have supercharged 3pt AND mid range open shooting. So as other have said, there's much more going on here than what we as users can edit or control.

So if you increase (did you mean decrease?) the league wide progression slider the shooting progression problem will only get worse. If you decrease the league wide progression slider to 0, as trekfan has tested, it still does not overcome the juiced up shooting development across the league. It just dampens the tremendous increase a little.

Your question did make me think of something else I could maybe test, though. With 30 team control, you can specify training priorities for a whole team at once for each team. So you could go to the Team Training window, select fundamentals, and create a custom set of priorities among all the ones listed, and drop the shooting training sliders to zero if you want. I suspect players will still increase, just not by as much, but that would have to be tested.

I suspect that because based on what I've seen so far, the in season training customization available to the user only modestly shapes the training outcomes, by some marginal percentage but far from 100%. I'm going to guess it's under maybe between 5-15% based on my recollection of some things Leftos said years ago. Our training sliders do not give us full player training effects control, I don't believe.

All that doesn't even solve the problem of player progression, which we can see also contributes to juiced up shooting. Remember, training effects and player progression (which is off season), for the game's purposes, are different and separate.

Maybe I'll run another test to see how much can be managed using the team by team training menus, beyond the individual player level. I'd have to run a new league sim test with 30 team control. I'm not optimistic it would do anything more than help a little, maybe by 5-10%, os if I'm right it wouldn't solve the problem.

[Mauer4MVP]

Re: NBA 2k17 MyLeague After 5 Years: Data Showing Problems

The CPU does not focus all training on 3 pt shooting, at least, not according to the visible training plans for each player on each team. It only puts maybe 5% of training points to shooting for good shooters, and even less for other types of players.

Nevertheless, the results of a combination of training plus off season progression have supercharged 3pt AND mid range open shooting. So as other have said, there's much more going on here than what we as users can edit or control.

So if you increase (did you mean decrease?) the league wide progression slider the shooting progression problem will only get worse. If you decrease the league wide progression slider to 0, as trekfan has tested, it still does not overcome the juiced up shooting development across the league. It just dampens the tremendous increase a little.

Your question did make me think of something else I could maybe test, though. With 30 team control, you can specify training priorities for a whole team at once for each team. So you could go to the Team Training window, select fundamentals, and create a custom set of priorities among all the ones listed, and drop the shooting training sliders to zero if you want. I suspect players will still increase, just not by as much, but that would have to be tested.

I suspect that because based on what I've seen so far, the in season training customization available to the user only modestly shapes the training outcomes, by some marginal percentage but far from 100%. I'm going to guess it's under maybe between 5-15% based on my recollection of some things Leftos said years ago. Our training sliders do not give us full player training effects control, I don't believe.

All that doesn't even solve the problem of player progression, which we can see also contributes to juiced up shooting. Remember, training effects and player progression (which is off season), for the game's purposes, are different and separate.

Maybe I'll run another test to see how much can be managed using the team by team training menus, beyond the individual player level. I'd have to run a new league sim test with 30 team control. I'm not optimistic it would do anything more than help a little, maybe by 5-10%, os if I'm right it wouldn't solve the problem.

Ahh okay yeah then my idea is invalid. Thanks for the work, man! I'm going to be testing some things out now because of this as well.