Warrior General Games

I am not sure how many time we should complain about it before it get fixed.

Here is a game where the a run package that had the following (all plays start at 100%, offense is SSM and package only called on 1st and 2nd down) :
1) WI Off tackle called 12 times (only for gains, except two fumble to end the game)
2) WI HB slams called 3 times (only for gains)
3) SI counter weak called 3 times (two gains and a fumble, the fumble came first)
4) SI HB dive called 0 times
5) trips pitch strong called 5 times (3 losses and 2 gains shorter then 2 yards)
AA is set to adjust quickly.

The pitch play was adjusted up once due to the fumble , then adjusted down twice. This does not included adjust down when other plays got adjusted up.

Something is not working, or the information we were given about AA is false. In my calculations for the pitch play
1) start at 20%
2) adjusted up after the fumble : 20% /0.85 = 23%
3) adjusted down after 0.5 yard loss 23% - 15% = 8%
4) adjusted down after another 0.5 yard loss 8% - 15% = 0%

The ply get called 3 more times, with no positive adjust. Even if I assume that a small loss is half the adjustment -7.5%(Bort quotes has no hint of that) I still cannot explain how this play is called after the 3rd loss (23 - 3* 7.5 = 0.5%)

Since one play in the package is never called. Something is very fishy here.

http://goallineblitz.com/game/game.pl?game_id=1981184

A couple of note to bug mods:
1) I do not need to show any more example, since this is a mathematical impossibly happening.
2) If you do not believe my calculation read the WIKI. Maybe I have reading issues and got it wrong.
3) I seen this weird behavior in the past. But now it is a very clear case.
4) this is not a whining thread, since the play in question didn't cost us the game.

I figure Auto Adjust is like drinking and darts, you just keep drinking more and hopefully you hit a bullseye at some point. I

Originally posted by The Big Cack Envy
I figure Auto Adjust is like drinking and darts, you just keep drinking more and hopefully you hit a bullseye at some point. I

It is actually good 90% of the time. But sometime .....

I've scouted the linked game, but don't have the Wiki link handy. Help?

My guess is that you're adjusting the numbers in a vacuum instead of recalibrating all of the percentages to equal 100% after every play, but I'm willing to look over the math because I also have my doubts about whether auto adjust is really WAI.

Originally posted by CDZYO
I've scouted the linked game, but don't have the Wiki link handy. Help?

My guess is that you're adjusting the numbers in a vacuum instead of recalibrating all of the percentages to equal 100% after every play, but I'm willing to look over the math because I also have my doubts about whether auto adjust is really WAI.

You guess wrong.

I didn't took into consideration the effect of positive adjust of the WI off tackle (since I don't know how it is divided) . But that would only take the pitch play lower not higher.

By the way, my guess is one of the two
1) Sim has major problems when loading AI to a game, which cause funny staff like that (due to wrong initial %). Siomething like the X-factor bug.
2) The information Bort dish out is half true or wrong. So a 0.5 yard loss will generate a very small adjustment and not the full 15%.

I trust you, but will verify.

Originally posted by CDZYO
I've scouted the linked game, but don't have the Wiki link handy. Help?

Edit: My guess was two-fold. One, that you may have not taken the effect of negative adjustment into account. Meaning, when one play fails, it increases the probability of the other plays. Two, that you may have not accounted for a percentage reaching zero. For example, if you have plays A, B, C and D, and the % for D is already at 0%, if A succeeds again, it can't lower D anymore. But would it decrease B and C by a larger amount to compensate (since any adjustment is a zero-sum game)?

I understood. This is why it is hard to find clear examples.

However in this game there was only one relevant positive adjustment for the pitch play and I took it into considerations. It may adjusted 20% to more then 23%, maybe 25% or even 30%, still the play was not suppose to be called the 5th time after 3 losses. .

Originally posted by tragula
I am not sure how many time we should complain about it before it get fixed.

Here is a game where the a run package that had the following (all plays start at 100%, offense is SSM and package only called on 1st and 2nd down) :
1) WI Off tackle called 12 times (only for gains, except two fumble to end the game)
2) WI HB slams called 3 times (only for gains)
3) SI counter weak called 3 times (two gains and a fumble, the fumble came first)
4) SI HB dive called 0 times
5) trips pitch strong called 5 times (3 losses and 2 gains shorter then 2 yards)
AA is set to adjust quickly.

The pitch play was adjusted up once due to the fumble , then adjusted down twice. This does not included adjust down when other plays got adjusted up.

Something is not working, or the information we were given about AA is false. In my calculations for the pitch play
1) start at 20%
2) adjusted up after the fumble : 20% /0.85 = 23%
3) adjusted down after 0.5 yard loss 23% - 15% = 8%
4) adjusted down after another 0.5 yard loss 8% - 15% = 0%

The ply get called 3 more times, with no positive adjust. Even if I assume that a small loss is half the adjustment -7.5%(Bort quotes has no hint of that) I still cannot explain how this play is called after the 3rd loss (23 - 3* 7.5 = 0.5%)

Since one play in the package is never called. Something is very fishy here.

http://goallineblitz.com/game/game.pl?game_id=1981184

A couple of note to bug mods:
1) I do not need to show any more example, since this is a mathematical impossibly happening.
2) If you do not believe my calculation read the WIKI. Maybe I have reading issues and got it wrong.
3) I seen this weird behavior in the past. But now it is a very clear case.
4) this is not a whining thread, since the play in question didn't cost us the game.




Were any of the runs a "significant" play. For example, a TD. I'd bet those cause a larger adjustment than a run on the 50.

Where did Bort say it was -15 for a loss? I tried finding it but could not.

Originally posted by Catch22
Where did Bort say it was -15 for a loss? I tried finding it but could not.

1) There is a quote saying it is -3%, -6% .. -15% depending on the rate of AA (was in a Q&A). Search the wiki for 'very quickly'

2) When he explain the issue, he never ever said there is half an adjust. Since he answered tons of those question I am assuming there are no half negative adjustments.



Originally posted by wiki

Bort
February 6, 2011 Question I assume that the 5 settings (very quickly, quickly, medium, slowly, very slowly) give a different change to the % given in a certain play based on if its successful or unsuccessful. Could you share how much % each of these settings will change a plays %? Also, could you share what is considered a successful/unsuccessful run/pass play?
Answer "very quickly" is 5 times faster than "Very Slowly." Very slowly will adjust an output by between -3 to 3 depending on what happened (meaning "very quickly" will do from -15 to 15). For runs, a gain of 6 or more is a great gain, more than 3-6 is a good gain, 0-3 is neutral, and a loss is negative. For a pass, a 15 yd gain is considered really good, 6 yds is good, a loss is bad, and the rest is neutral. If the play was a turnover or a score, that's an additional + or -.

Originally posted by tragula
Something is not working, or the information we were given about AA is false. In my calculations for the pitch play
1) start at 20%
2) adjusted up after the fumble : 20% /0.85 = 23%
3) adjusted down after 0.5 yard loss 23% - 15% = 8%
4) adjusted down after another 0.5 yard loss 8% - 15% = 0%

Considering that the trips pitch play isn't called until the ninth call of the package -- which makes for eight prior adjusting events -- your analysis is already flawed.

I do wonder if the 0.5 yard losses were treated as neutral. I also wonder if the bug is in how auto-adjust treats those 100% inputs.

Originally posted by tragula
A couple of note to bug mods:
1) I do not need to show any more example, since this is a mathematical impossibly happening.
2) If you do not believe my calculation read the WIKI. Maybe I have reading issues and got it wrong.
3) I seen this weird behavior in the past. But now it is a very clear case.
4) this is not a whining thread, since the play in question didn't cost us the game.
That's a load of arrogant crap. If you run the same package in another game and get a different result (say, SI HB Dive called 15 times and WI OffTackle called 0 times), then what's your bug?

Originally posted by CDZYO

Considering that the trips pitch play isn't called until the ninth call of the package -- which makes for eight prior adjusting events -- your analysis is already flawed.


But only one of them was an adjustment down. The pitch play started from 20% can you imagine it got over 30% from a single adjustment of another play ? If I would take into account the positive adjustment it would only make my case clearer.

Originally posted by CDZYO

That's a load of arrogant crap. If you run the same package in another game and get a different result (say, SI HB Dive called 15 times and WI OffTackle called 0 times), then what's your bug?


If 1+1=2 is true every day of the week, it is not a statistical result. Events where I can show a mathematical problem cannot happen every game (and it is not that I check every package every time).

The bug is not about the dive played called 0 time (approx 1/100 event, I can live with that) or Off tackle called 12 times. It is about a play just should not been called 5 times.

Because the feedback a play that is adjusted up getting called many times, or a play that never get called are common results. However a play that is only adjusted down which keep getting called is an issue.

Originally posted by wiki


Bort
February 6, 2011 Question I assume that the 5 settings (very quickly, quickly, medium, slowly, very slowly) give a different change to the % given in a certain play based on if its successful or unsuccessful. Could you share how much % each of these settings will change a plays %? Also, could you share what is considered a successful/unsuccessful run/pass play?
Answer "very quickly" is 5 times faster than "Very Slowly." Very slowly will adjust an output by between -3 to 3 depending on what happened (meaning "very quickly" will do from -15 to 15). For runs, a gain of 6 or more is a great gain, more than 3-6 is a good gain, 0-3 is neutral, and a loss is negative. For a pass, a 15 yd gain is considered really good, 6 yds is good, a loss is bad, and the rest is neutral. If the play was a turnover or a score, that's an additional + or -.


I interpreted Bort's statement on a negative adjustment to work similar to how a positive adjustment works. Meaning, a small loss on a play causes a smaller adjustment than a big loss. I also believe, that auto adjust will never take a play to zero%. And I don't think it' is as straight forward as the OP suggests. Pretend I have 6 rushing plays in a package. Play 1 is set to 50% and the others are all set to 10. Auto adjust set to very quickly

First play of the game play 1 goes for 10 yrds. So, we think we increase the 50 to 65 and the others are all reduced .. but by how much? In my example because each of the remaining 5 plays has the same % to fire, they are all reduced equally. But I think that if one of those plays had a 20% chance to fire that it gets adjusted by a different amount then the others who have different percentages.

Also - when looking at mathematics, random numbers and percentages ... we cannot use yesterday's history to influence today's dice rolls. By that I mean -- if you were standing at a craps table and watched the shooter throw a 7 eight times in a row -- what are the odds that the next roll would be a 7 ? Yup -- the same odds as on that shooters first roll.

So -- we need to remember that the RNG does not pay attention to its most recent result when conjuring up the next random number, it simply generates one. So, if your package has a play in it that has s 1% chance of being called -- it can be called twice in a row.

I do not profess to know how the Auto Adjust works -- I have only shared my thoughts.


Peace,

Fish

The issue is, instead of explaining how the auto adjust can produce unexpected results, perhaps the "adjusting" needs to be looked at so that if a play fails repeatedly, it does NOT keep getting called. I think that's what people want. What good does it do anyone to have adjusting AI if it can't be manipulated to produce desired results?

^

Originally posted by IdRatherFlyFish

...

I never understood why bug forum, should be sunday school forum. It is not about can we interpret Bort's quotes in a way that make everything ok. He is a man not god, if he wrote something that is unclear/incorrect then he should be notified about it and fix it (either sim or his answer).