User Pass
Home Sign Up Contact Log In
Forum > Discuss GLB Issues With Catch22 > RESOLVED BUGS > Something is broken with GLB's autoadjust - ISSUE RESOLVED
Page:
 
GMathiasf
offline
Link
 
I don't know how I prove it, or exactly what's wrong, but something has been screwy this season. I assume that it's auto-adjust related.

This was from today's game http://goallineblitz.com/game/game.pl?game_id=1813341 :

Auto Adjust Slowly
10% WR Clear Out
10% Shotgun Smash
10% I Pitch weak Payton
10% HB Out
10% WI WR screen
10% sb counter Payton
10% TE Cross Slant
10% PS HB Blast
10% PS HB Counter
10% SB Dive weak

Here's what we got:

44 Shotgun Smash
4 WR screen strong
4 HB Blast
3 WR Clear Out
3 SB counter Strong
3 TE Cross Slant
2 HB Counter Weak
2 HB Out
2 Dive Weak
1 I PItch Weak

And it's not like the play was breaking huge plays or something. It was a good play call, but by the end of the game it was completed 15 out of 44 times for 4.19 yards/play. Hardly justification for it to continue being called.

Had a similar thing happen to a slightly lesser extent in the pre-season. I didn't document it as well, but here's what I got. http://goallineblitz.com/game/game.pl?game_id=1780073

I had like 10 plays in at 10% each. WI WR In got called 24 times, and the 2nd time it was called was a 102 yard INT return for a TD. Some plays (FB Clear, Weak Overload, Double Hitches, SB HB counter strong) got called 0 times. For those of you with statistics experience. What are the odds that a play gets called 24 times while 4 others get called 0, when all are supposedly an equal 10% chance of happening (granted slow auto adjust is on, but seriously is this an obvious bug that the INT return TD got counted as a positive?)?
Edited by Catch22 on Sep 22, 2011 01:05:48
Edited by Catch22 on Sep 16, 2011 08:07:06
 
Novus
offline
Link
 
I don't think it's Auto-Adjust, because I have Auto-Adjust turned OFF entirely, and I'm still getting odd results.

Example:
In this game ( http://goallineblitz.com/game/game.pl?game_id=1786993 ) I had an outside-play package with the following 4 plays in it:

25% - Strong I FB Screen Weak
25% - Weak I HB Sweep Weak
25% - Strong I HB Sweep Strong
25% - Single Back HB Counter Strong

And here's how the play distribution turned out... bear in mind, this is with Auto-Adjust turned COMPLETELY off:

04 - 17.4% - Strong I FB Screen Weak
00 - 00.0% - Weak I HB Sweep Weak
03 - 13.0% - Strong I HB Sweep Strong
16 - 69.6% - Single Back HB Counter Strong

None of those plays were in any of my other packages at all, only in that one. So something is screwy with the random-number-generator itself.

I've had a few other similar occurrences as well, though they weren't quite that extreme, so I was willing to write them off to, "Well, that's the RNG for ya." But this one example made me take notice. And now your thread makes me think I'm not alone.
 
GMathiasf
offline
Link
 
yah, I've noticed some strange occurrences in my smaller packages as well, but with a small sample size, it's difficult to prove anything. My short yardage package today was 40-20-10-10, so basically 50% chance play 1 gets called, and it was dialed up 8 out of 9 times. Though that's not crazy, it's still statistically very unlikely (less than 2%). By itself, that's not enough to prove anything.

According to a probability calculator, the odds of an event occurring 44 times out of 65 when it's a 10% chance of happening each time (so if there was no auto-adjust) is <0.000000000001 or less than 1 in 1 trillion. That's borderline statistically impossible without a bug of some kind.
Edited by GMathiasf on Sep 15, 2011 18:16:29
Edited by GMathiasf on Sep 15, 2011 18:11:29
 
tautology
offline
Link
 
Originally posted by GMathiasf


less than 1 in 1 trillion

That's borderline statistically impossible without a bug of some kind.



Agreed 100%

I have seen mind-boggling improbabilities a few times with the AI.


THERE IS DEFINITELY SOMETHING GOING WRONG SOMEWHERE.


This needs attention.
 
tonnyrat
offline
Link
 
Originally posted by tautology
Agreed 100%

I have seen mind-boggling improbabilities a few times with the AI.


THERE IS DEFINITELY SOMETHING GOING WRONG SOMEWHERE.


This needs attention.


me too. i thought AA was fucked in the pssn compared to last ssn.
 
whodey08
offline
Link
 
Originally posted by coreyls18
weird, used exact same O and D. And look at the carries for brady and whodey...nothign close like last game. Also I have the screens on auto-adjust but we ran that damn FB screen that was 0/7...auto adjust is awful


 
whodey08
offline
Link
 
I agree something is definitely off somewhere. I have been an OC for several seasons for several teams and as I watch games there are a ton of times this season that have left me scratching my head. The percentages of plays called is just whacky.
 
johnbarber
offline
Link
 
agreed to all points above
 
dbreeze
offline
Link
 
friggin' bizarro ball......
 
tautology
offline
Link
 
Originally posted by GMathiasf

According to a probability calculator, the odds of an event occurring 44 times out of 65 when it's a 10% chance of happening each time (so if there was no auto-adjust) is <0.000000000001 or less than 1 in 1 trillion. That's borderline statistically impossible without a bug of some kind.




To put this in perspective, you are likely to win the powerball lottery 5000 times before you would see the probability distribution exhibited in this game.


Seriously.

Let's take a look. This is a bug that can destroy the gaming experience.
 
tautology
offline
Link
 


You will draw a royal flush off the flop 1.5 million times before you see a game like this.


Please mark as a bug and let's get this into the system.
 
cowtesticles2001
udder pressure
offline
Link
 
tautology has spoked
 
Catch22
offline
Link
 
Couple things guys.

1) Need links to the games in question
2) Need links to the AI for the game in question

Not taking this any further until I can actually review the data and information available.
 
Novus
offline
Link
 
Originally posted by Catch22
Couple things guys.

1) Need links to the games in question
2) Need links to the AI for the game in question

Not taking this any further until I can actually review the data and information available.

Cool. Good to see this being taken seriously.

I've already made changes to my AI and Packages since the last game, so that won't help you unfortunately. But I have a game coming up later this afternoon. If I see any weird play distributions within a package today, I'll immediately save copies of the AI and Packages in-question and set them aside for you to look at later.
 
TrevJo
offline
Link
 
Originally posted by Novus
I don't think it's Auto-Adjust, because I have Auto-Adjust turned OFF entirely, and I'm still getting odd results.


Yeah I think anyone doing these experiments with auto-adjust should also do them with auto adjust off.
 
Page:
 


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