User Pass
Home Sign Up Contact Log In
Forum > Discuss GLB Issues With Catch22 > NOT A BUG - CLOSED > Wide Open Incompletion Case Study - Intended to answer Borts questions about the frequency of this error
Page:
 
Mike Rogers
Lead Tester
offline
Link
 
Making any change to passing has huge consequences, I (and most here) have been through a few already, at it is devestating with the amount of disruption it causes.

The amount of variables involved in changing the pass success %, then trying to adjust the defense to counter it, then balancing that with the multitude of player builds out there... well its a lot of work.

I think it would make a great improvemnt to the reality of the sims, but is it needed to continue the integrity of the game as a whole, is it the best use of Borts time.

I dunno to be honest.

What I will say is that having a portion of the offense running with a 'known bug' will lead to teams avoiding it or exploiting it and a shift in the dynamics of GLB.

 
lazloth
offline
Link
 
Originally posted by Mike Rogers
Making any change to passing has huge consequences, I (and most here) have been through a few already, at it is devestating with the amount of disruption it causes.

The amount of variables involved in changing the pass success %, then trying to adjust the defense to counter it, then balancing that with the multitude of player builds out there... well its a lot of work.

I think it would make a great improvemnt to the reality of the sims, but is it needed to continue the integrity of the game as a whole, is it the best use of Borts time.

I dunno to be honest.

What I will say is that having a portion of the offense running with a 'known bug' will lead to teams avoiding it or exploiting it and a shift in the dynamics of GLB.



Weren't you just on your soapbox demanding a change in the running game? This defect in the passing attack penalizes "better" WRs after they've beaten the coverage and will continue to grow worse as QBs and receivers gain in skill.
 
Bort
Admin
offline
Link
 
I guess my main concern is that current completion percentages are based on the current passing dynamic. Percentages for good teams are already in the 60-70 range often times and if I make a major change to that dynamic that's been in play for 4 1/2 seasons then we are going to see a large change in the performance of all QBs and WR's, and especially the higher leveled ones.

It's something I have to approach lightly and not just say "oh, here's a problem, swap that around, let it ride." I have to spend some time rebalancing to adjust to the new dynamic.
 
Bort
Admin
offline
Link
 
Originally posted by Modok
Originally posted by Bort

I've found the source of the "bug" if you want to call it that, and it's been the same since the first game ever played on GLB.

It has to do with the order of operations for calculating when a ball is catchable, and comes up when the receiver and ball get to the spot at around the same time, with certain angles used by the QB. Basically, right now the code progression is:

1. Move ball
2. Hit ground? Play over.
3. Near receiver? Try for catch.

If I switch it around to

1. Move ball
2. Check for catch
3. Check for ground hit

Then the receiver gets a chance to catch the ball before it gets checked for a ground hit.

Or, I can add an increased granularity proximity check for thrown balls that are very close to the receiver but not quite within the catching radius.

Now, the problem with changing this around the first way is that it totally changes the completion percentages, and changes a whole bunch of other stuff that is balanced around the original method, including deflections, interceptions, etc.

Changing it around the second way with an increased granularity check keeps other stuff balanced the same, but it will still end up boosting completion percentages enough that it would require adjustments to QB throwing to add more inaccuracy chances at longer distances. I'm not sure I'm comfortable doing that with the likely public response that goes along with it.


I am of the opinion, and I think most level headed people would agree, that is there is that big of mistake in the code, we need to fix it the right way.

1. Move ball
2. Hit ground? Play over.
3. Near receiver? Try for catch.

I think reordering it is the way to go because it identifies the problem and fixes it. The granularity proximity check is just a "band aid" to fix the problem that will still exist in theory.

Also, you mentioned that you may have to change the QB in your second option. Wouldn't it make more sense to leave the QB alone and make the changes with the cornerbacks and secondaries? Giving them a better chance to react and make a play on the ball.



Actually I think increased granularity is the better solution. If I just switch it around, then a lot of these balls will be considered "low passes" which means the receiver will have no chance to catch it in stride.
 
Staz
offline
Link
 
Originally posted by Bort
I guess my main concern is that current completion percentages are based on the current passing dynamic. Percentages for good teams are already in the 60-70 range often times and if I make a major change to that dynamic that's been in play for 4 1/2 seasons then we are going to see a large change in the performance of all QBs and WR's, and especially the higher leveled ones.

It's something I have to approach lightly and not just say "oh, here's a problem, swap that around, let it ride." I have to spend some time rebalancing to adjust to the new dynamic.


Keeping it around 60% is nice, just making catches in tight coverage a little tougher (40-50%), and wide open catches about 80-90%, that should balance out. If your coverage is poor enough to leave WRs wide open, you're going to allow a high completion percentage in real life.
 
winchester
offline
Link
 
Or when you are in double and triple coverage drop it to 0-10%, tight coverage 40-50% and when the WR is wide open 90-100%.
 
Staz
offline
Link
 
Wide Open: 80-90%
Single Coverage (Average): 60-70%
Tight Single Coverage: 40-50%
Double Teamed: 30-40%
Tight Double Team: 20%
Triple Teamed: 10%

Something like that?
 
Mike Rogers
Lead Tester
offline
Link
 
Originally posted by lazloth

Weren't you just on your soapbox demanding a change in the running game? This defect in the passing attack penalizes "better" WRs after they've beaten the coverage and will continue to grow worse as QBs and receivers gain in skill.


Err whats your point here?

My only point is it should be fixed if it doesnt overpower defenses, and that the fix needs to be tested because it effect so many situations in a game, and is very complex.

Which was exactly the same point I made in the Running game thread.
and the thread trying to fix Pass Rushing, Run Blitzing and DB coverage.

Originally posted by
Wide Open: 80-90%
Single Coverage (Average): 60-70%
Tight Single Coverage: 40-50%
Double Teamed: 30-40%
Tight Double Team: 20%
Triple Teamed: 10%

Something like that?


I Would hate to see pass completions exclude the builds of the WR or QB involved though.

And it should also include the distance from scrimmage etc.

Wide open inside 10 yards at 80-90% seems fine, but 30-40 yards downfield it seems a little high to be the average.

Last edited Sep 27, 2008 14:39:54
 
lazloth
offline
Link
 
Originally posted by Mike Rogers

Err whats your point here?

Apologies if I misread you. I took your comment to the effect of "do it slow, will break passing, etc" - very different from roll back tweaks to HB slam blocking.

Originally posted by Mike Rogers

Wide open inside 10 yards at 80-90% seems fine, but 30-40 yards downfield it seems a little high to be the average.

Wide open 30-40 yards downfield implies the defense has broken down: blown assignments or coverage that cannot compete with the offense. If a QB got a ball out to someone wide open (eg., no defender within 10yd?), I would not expect that play to result in an incompletion barring a very poor throw or a WR who simply cannot catch.

Tony Romo's early career was made with the long ball, wasn't it?
 
Mike Rogers
Lead Tester
offline
Link
 
Not saying its not possible, but to expect every QB irrespective of build to be able to hit a 40 yard pass 90% of the time if 'Wide open' is a little much.

This would result in one dimensional WR builds, (All speed and Agility) and one dimensional gameplans.

This is exactly what happened when it became apparent that a WR recieved a boost to catching whe 'wide open' last time round.

WR had at most 30 catching and never dropped the ball.

Bort has worked to avoid that happening again.

Thats all I want to avoid, any always works/fails plays.

Same reason I want to turn off the Weak I play so we can see more sweep plays used to balance the offense, and a draw play in the shotgun formation to stop defenses keying on that.

 
lazloth
offline
Link
 
There are other solutions long suggested that would be effective rather than gimping the passing game, and not allow the game to devolve into one dimensional play. (CBs jamming at the line, named assignments, etc.)

"Work always" plays could be neutered if heavier attention went into the game mechanics and plays. I'm not disagreeing with your spirit, but the execution leaves a stagnant game.
 
Enkidu98
offline
Link
 
Change the order of the checks and then undo some of the CB nerfs.

You make QB's and CB's happy. Obviously open receivers have a chance at the ball. CB's cover a little better again. So - Good coverage = lower catch percentage. Bad Coverage = good QB/Receiver numbers. Coverage as good as the Passing game = @55-65% completion rate. Which is about right
 
tonylieu
offline
Link
 
Originally posted by Bort
Originally posted by Mike Rogers

While I have to accept that Bort has found a sequance out of order, what I cant see is how that alone accounts for a sudden spike in this occurances this season.


It has to do with the ever increasing QB throwing power and WR speed, and how it relates to the granularity of update polling. We've probably just hit a point with certain combinations that hasn't come up before, and with the passing styles it's amplified it to be more noticeable.

General explanation of what I'm talking about with granularity. Obviously we can't sim every nanosecond of the play, so it's broken down into ticks every so often.

Example, with made up numbers:

If on tick 10, it says that the ball is 10 feet away, you're not going to be able to catch it on that tick. If on tick 11 it says that the ball is now only 1 foot away, you can now catch it. If it's also hitting the ground on tick 11, though, and we check that first, then you don't get a chance to catch it because it supercedes any "how close am I" checks.

It also might have to do with the additions of the passing styles I added last season, since they affect angles used to throw the ball, and decrease the randomness therein that used to be used instead, which probably "hid" things some.

Like I said before, the best "fix" is just to add some additional checks at increased granularity for catching. Once I do that, though, the QB percentages are going to go up since some of those passes that were incomplete are now going to be more catchable. Therefore, it will require some rebalancing of QB accuracy, especially on long throws, to ensure we don't have everyone at 90% completions all the time.


Hi Bort, sorry if I seem a bit slow today, but even if the "catch" check is done first at time tick 11 in your example, the ball should still be found on or near the ground, and should result in a very low possible successfull catch, especially when the receiver is running downfield away from the ball. In fact, it would make even less football sense IMHO, that the game should be altered to allow a speed receiver with low catching ability to be able to make such a catch.
 
tigerbait22
offline
Link
 
Tonylie.. FAILED in this thread... Now will you admit its an issue? And stop arguing with everyone here that has tried to be constructive?
 
tonylieu
offline
Link
 
Originally posted by tigerbait22
Tonylie.. FAILED in this thread... Now will you admit its an issue? And stop arguing with everyone here that has tried to be constructive?


Oh, now it's no longer a "bug". Sorry I missed the memo that downgraded it to "issue".
 
Page:
 


You are not logged in. Please log in if you want to post a reply.