Posted by fermor332002 on 3/9/2012 3:16:00 PM (view original):
5-6 new formations? Most guys practice for 2-3 OF and 2 for DF because of the amount of practice time aloted for formation and player practice. Will there be more time aloted for more formations? No use having more if you can not really use it.

Now do not get me wrong. I would welcome more formations if I am abule to practice them with out spreading my time too thin.

I wouldn't throw new formation in until we can address formation IQ, but I want to work on the engine with more options in mind.

Just because more formations are added, it doesn't mean you have to use them.  It's just more options.

Posted by gt_deuce on 3/9/2012 4:02:00 PM (view original):
Just because more formations are added, it doesn't mean you have to use them.  It's just more options.

That is just it. I would like to use them.

Any chance that we will see a FB, DE, ILB, OLB, S, and CB positions when recruiting?  Seems like it would help out if they were specific to that particular position in high school and possibly have high ratings than when you move them from like RB to FB.

Posted by mlemley on 3/10/2012 11:19:00 PM (view original):
Any chance that we will see a FB, DE, ILB, OLB, S, and CB positions when recruiting?  Seems like it would help out if they were specific to that particular position in high school and possibly have high ratings than when you move them from like RB to FB.

I guess I don't understand what you're wanting here.  The ratings don't change whether you play them at DE or DT, S or CB, ILB or OLB, RB or FB, right?

I understand they don't change for any of the players.  But like DE, you may find that position like in real life are more like LBs than they are DT.  More athletic, more speed and not as big as a DT.  I guess that is really the only position, but some of the others would be nice to see them set for those positions.  If you run a recruiting tool you will see that your top RBs may not be your best FBs and that the FBs ratings in the tool are rated not as high or career potential.  If your #1 FB rated at 62.33 is your #20th best RB rate at 68.45.  The FB will never be rated as high is he was kept at RB.  I understand that using a RB at FB, they do not lose any ratings like if you converted someone.  I am just saying that I think if there were players that were FBs to recruit, it would seem that maybe we good get there core value ratings better at that position.

Posted by mlemley on 3/11/2012 12:38:00 PM (view original):
I understand they don't change for any of the players.  But like DE, you may find that position like in real life are more like LBs than they are DT.  More athletic, more speed and not as big as a DT.  I guess that is really the only position, but some of the others would be nice to see them set for those positions.  If you run a recruiting tool you will see that your top RBs may not be your best FBs and that the FBs ratings in the tool are rated not as high or career potential.  If your #1 FB rated at 62.33 is your #20th best RB rate at 68.45.  The FB will never be rated as high is he was kept at RB.  I understand that using a RB at FB, they do not lose any ratings like if you converted someone.  I am just saying that I think if there were players that were FBs to recruit, it would seem that maybe we good get there core value ratings better at that position.

I think this applies more to haw Norbert affects the game engine with respect to player vs player match-ups. The ability to know that a FB can be set as "blocking" and the blocking ability makes a difference in that role would make the FB more valuable if they can block (maybe need to tweak RB settings to add more blocking ability). Same would be for blocking vs pass catching TE, possession vs speed receivers, run block vs pass block OL, run stop vs pass rush DE, etc. Another factor is if he designs more formations (Pro-set with HB/FB or 2 HB set-up etc). If the engine gets that good to allow more specific player qualities to set up a variety of player match-ups, the recruit values would have to be more able to develop those specific players. You can see this in Guess ratings where FB and DE don't live up to RB and DT values.


POPULATE HEISMAN!

Part of this update will be to define a little more which ratings affect different actions.  I'd like to see speedy backs excel more at wide runs and power backs excel at runs between the tackles.  I'd like to see faster receivers have a better chance of being successful down field.  First part is to work these considerations into the engine as well as trying to provide a little more control on how your team plays.  If it's a problem to have your slow FB being the target 30 yards down field, then there should be ways to make sure it doesn't happen.  Currently, the formations don't allow this, but if we work in the passing distributions again, we'll have to watch for this and allow some control over it.

First step is actually make sure these different ratings matter in the engine the way you think they would.  The next steps would be to review rating distributions to provide a little more variety in players at different positions and review how you then search for them, but that's something we will look at later.

Posted by fermor332002 on 3/9/2012 4:07:00 PM (view original):

Posted by gt_deuce on 3/9/2012 4:02:00 PM (view original):
Just because more formations are added, it doesn't mean you have to use them.  It's just more options.

That is just it. I would like to use them.

Then use them. But you are not compelled to use additional new formations and, if you do, you're going to have to make a series of decisions about where the practice minuets (or not) are going to come from. This is not a new phenomenon within GD - tradeoffs. And I don't think additional options that may lead to additional tradeoffs (or not) is a bad thing.

Posted by norbert on 3/12/2012 12:14:00 PM (view original):
Part of this update will be to define a little more which ratings affect different actions.  I'd like to see speedy backs excel more at wide runs and power backs excel at runs between the tackles.  I'd like to see faster receivers have a better chance of being successful down field.  First part is to work these considerations into the engine as well as trying to provide a little more control on how your team plays.  If it's a problem to have your slow FB being the target 30 yards down field, then there should be ways to make sure it doesn't happen.  Currently, the formations don't allow this, but if we work in the passing distributions again, we'll have to watch for this and allow some control over it.

First step is actually make sure these different ratings matter in the engine the way you think they would.  The next steps would be to review rating distributions to provide a little more variety in players at different positions and review how you then search for them, but that's something we will look at later.

Kind of tangential question for you, norbert, to satisfy my own curiosity...

After having extensive conversation(s) on passing distribution and what coaches are looking for in the passing game, do you see the benefits of what we're requesting, within the confines of our ability to control our team?  Or is this something that you see as superfluous, but are willing to "go along with" because the community wants it?

I definitely see the benefits to being able to determine who is getting targeted on your plays, and where they are getting targeted.  The new engine was supposed to address this through having specific slots that would be different depths and combined with your aggressiveness would determine the trends in throwing to one receiver over another, but between the lack of specific knowledge of which slot is which and the randomness of the pass aggressiveness settings, there really isn't much control at all.

I'm not sure just adding distribution will give us what we want either, which is why I'm still trying to keep this conversation going.  Distributions don't answer the problems with the randomness of the aggression settings, and I'll still have to deal with other parts of the passing like what happens when the QB is under pressure or the primary target is covered.  There are some things that can cause what ever distribution controls we add to not reflect the outcome of the game, which means we need to do a better job of explaining why they didn't happen that way.  Might be more acceptable if your top receiver doesn't get targeted as much if you see the other team is always covering him.

Which brings up some other issues.  If we can add control over which receivers are targeted more, should there be ways to counter that on defense to double-team or focus on a receiver.  How covered does a receiver have to be for the QB to throw to him.  Should there be a way to force more throws to covered receivers.  We could probably open a can of worms where this new setting begets this new setting and so on, so I think we need to take one step at a time and get to where we have more control over our team but not to where it's so complicated that it might not be clear what all the combination of settings would produce.

So, yes, I get it and I agree.  I think from the technical side and potential ability of the engine, the new engine was a step in the right direction, but it fell short when dealing with this type of control.  There are other places where I think the new engine fell short, but I think it's salvageable.  We tried to add more emphasis on individual play and used some methods to tell which individuals are more likely to affect certain plays which lead to the distribution settings going away, but it's pretty clear we need to rethink how that is working and get back to where we had that control.

In a few weeks, I should be spending more time on the GD update and I might pull out certain conversations from this thread into separate conversations.  There is still a lot I'd like to get ironed out before I start changing code.

It might be informative to pull the old code from GD 1.0 that handles the passing/distribution aspect of the game, then imagine how you might improve upon that to get the precise result you're looking for.  I say that, not to bring that exact system back, but simply because people basically seemed pretty happy with the inputs->results->tweaks->new results loop that we had.  It may not have been perfect... and it may not even have worked exactly how we were told and/or how we thought it did, but there was a satisfying input-to-result connection that seems to be so severely lacking in the new engine.

From there, you could compare-and-contrast this reimagined GD 1.0 code and see how to upgrade the 2.0 code.  I don't even think you'd have to actually re-code 1.0... just imagine how it could've been improved and that should be instructive of how to shape the 2.0+ (or 3.0 or whatever we're calling it) code.

GD 1.0 to 2.0 is apples to oranges in many cases.  There is a built in distribution number in 2.0 that is calculated based on the player's slot in the formation and his ratings.  It gives player's with higher receiving ratings a higher distribution number.  I'm pretty sure I can tap into that to allow the distribution to be input from setting rather than calculated.  What it doesn't address is how to change those distributions based on distance or other factors of the game, like if you are in the red zone.  I have a couple ideas and when I get some more time, I'll try to put together a few mocks to see if they make sense.  It will be easier for me to see if the game plan setting will work if I can see them.

I'm thinking we can have some way to set base distributions on a formation either one time or each time you set it in the play book, and then have one setting on a player that modifies his distribution number like we modify a player's 3-pt shots in HD.  But until I look at all the settings, I'm not sure if it will be too complex or too confusing.  I don't think we can go back to the GD 1.0 method of setting depth charts and distributions on the formations themselves, but I will keep that in mind when trying to come up with something.

I've heard that 1.0 was even written in a different "language" than 2.0, so would that have an effect??

I don't think we can go back to the GD 1.0 method of setting depth charts and distributions on the formations themselves, but I will keep that in mind when trying to come up with something.

This is what I was driving at.  Just got a little wordy.