User Pass
Home Sign Up Contact Log In
Forum > Suggestions > Have a 'success' setting in the offensive packages and/or AI
Page:
 
ZoSo1
offline
Link
 
Originally posted by Novus
Following up on Romo's question... what percentage did you have each play at? Were all 8 plays at 100% each like a lot of people do nowadays, for a total of 800%? Or did you have each play at 12.5% each, for a total of 100%? (12.5% x 8 = 100%)

Makes a HUGE difference. Auto-Adjust simply adjusts the raw number you have in your package.

If you have all 8 plays at 100% each, Auto-Adjust set to Very Quickly, and you run a play that gets stuffed for a 1-yard loss, here's what would happen:
- The bad play would drop to 95%.
- The other 7 unused plays would each increase to 100.7% each.
- There's not much difference between 95% and 100.7%, so the 95% play would still have just under a 1-in-8 chance of getting called -- pretty much the same chance as before.

If you have all 8 plays at 12.5% each, Auto-Adjust set to Very Quickly, and you run a play for a 1-yard loss, here's what would happen:
- The bad play would drop to 7.5%.
- The other 7 unused plays would each increase to 13.2%.
- The 7.5% play would now drop from a 1-in-8 chance of getting called to a 1-in-13 chance of getting called -- significantly smaller.

...

See how that could make a big difference, Hitman?



800% would not be possible....... see below

from the Q&A Wiki search

answered by: Bort October 1, 2011

Question: could we clarify how percentage Bias works in the offensive package section? when creating a new offensive package, you can set a percentage Bias. i thought i remember many seasons ago that we were told: 1. the numbers need to add up to at least 100 (to avoid bugs) 2. the numbers can add up to over 100 3. the numbers are not hard percentages can anyone help clarify (confirm/deny) this? i have been trying to work some packages with many, many plays and instead of trying to figure out %s, i believe i can leave it at the default 100 and let the sim randomly choose a play (equally). or have you found that it consistently calls certain plays (like the top one)?

Answer: The numbers will be adjusted in-memory to add up to 100 in whatever ratios you entered them in at when the sim starts up. For example, if you have 4 plays set to 100, they will be adjusted to all be 25 when the sim runs so the code can handle things as if they were 100%. If you had 4 plays set to 50, they would also be adjusted to 25, since they are all equal. If you had two plays with one of them set to 100 and the other set to 50, they would be adjusted to 66.6% and 33.3% since that is the ratio based on the whole (150).
 
bhall43
offline
Link
 
Ya I knew Bort had clarified that at some point but I was too lazy to look up the QnA on it. The numbers are all adjusted to hit a straight 100% at the beginning of the sim.
 
Novus
offline
Link
 
Originally posted by ZoSo1
800% would not be possible....... see below

from the Q&A Wiki search

answered by: Bort October 1, 2011

Question: could we clarify how percentage Bias works in the offensive package section? when creating a new offensive package, you can set a percentage Bias. i thought i remember many seasons ago that we were told: 1. the numbers need to add up to at least 100 (to avoid bugs) 2. the numbers can add up to over 100 3. the numbers are not hard percentages can anyone help clarify (confirm/deny) this? i have been trying to work some packages with many, many plays and instead of trying to figure out %s, i believe i can leave it at the default 100 and let the sim randomly choose a play (equally). or have you found that it consistently calls certain plays (like the top one)?

Answer: The numbers will be adjusted in-memory to add up to 100 in whatever ratios you entered them in at when the sim starts up. For example, if you have 4 plays set to 100, they will be adjusted to all be 25 when the sim runs so the code can handle things as if they were 100%. If you had 4 plays set to 50, they would also be adjusted to 25, since they are all equal. If you had two plays with one of them set to 100 and the other set to 50, they would be adjusted to 66.6% and 33.3% since that is the ratio based on the whole (150).


That's weird. I could swear I remember reading the exact opposite somewhere.

I do know this, though... when I used to just set every play in my package at 100%, Auto Adjust didn't work well for me at all. When I switched to making all of the plays in a package add up to a grand total of 100%, suddenly Auto Adjust started working for me the way I expected it to.

I also know this: it wouldn't be the first time Bort INTENDED his sim to work a certain way, only to discover later that it's actually been working a different way the whole time due to something he overlooked. That's not to say that Bort is incompetent or lazy or anything, just that his sim is INCREDIBLY complicated and has lots of moving parts that can interact with each other in highly unexpected ways.

So, I'm sticking with my post from Page 1. I can't speak for anyone else's experience with Auto-Adjust, but MY experience is that it works better when I make sure the play percentages add up to a grand total of 100%.
 
bhall43
offline
Link
 
As a guy that uses the 100% on every play package rule of thumb. I can say I definitely feel it works fine as long as you keep screens and the regular plays in separate packages.
 
Novus
offline
Link
 
Originally posted by bhall43
As a guy that uses the 100% on every play package rule of thumb. I can say I definitely feel it works fine as long as you keep screens and the regular plays in separate packages.


Don't get me started on screens wreaking havoc with Auto Adjust...

And actually, that feeds back to my point about Bort's sim being so complicated, because I'm pretty sure he didn't intend for screens to totally bollocks-up Auto Adjust, but for some odd reason, they do.
 
SteveMax58
offline
Link
 
Originally posted by ZoSo1

Answer: The numbers will be adjusted in-memory to add up to 100 in whatever ratios you entered them in at when the sim starts up. For example, if you have 4 plays set to 100, they will be adjusted to all be 25 when the sim runs so the code can handle things as if they were 100%. If you had 4 plays set to 50, they would also be adjusted to 25, since they are all equal. If you had two plays with one of them set to 100 and the other set to 50, they would be adjusted to 66.6% and 33.3% since that is the ratio based on the whole (150).

This is exactly what I recalled and why I have always used 100% (assuming I want an equal bias) until AA kicks in.

Originally posted by Novus

I can't speak for anyone else's experience with Auto-Adjust, but MY experience is that it works better when I make sure the play percentages add up to a grand total of 100%.


My experience is that setting AI outputs to 100% and using AA-Very Quickly does not seem to adjust as quickly as I would expect. And I run mainly 2 AI outputs per AI input.

Of course, the sample sizes for any given AI input are pretty small to start with so its hard for me to say it doesn't work as I might have an AI input triggered 4-6 times at most in a game. So the adjust % would need to be like 25% (possibly more) on Very Quickly to truly adjust "Very Quickly". At least from a defense perspective...there are too many variables to expect any AI input to get triggered too any more times so therefore Very Quickly needs to be almost an immediate gigantic swing in % bias.
 
bhall43
offline
Link
 
Originally posted by Novus
Don't get me started on screens wreaking havoc with Auto Adjust...

And actually, that feeds back to my point about Bort's sim being so complicated, because I'm pretty sure he didn't intend for screens to totally bollocks-up Auto Adjust, but for some odd reason, they do.


They work fine in a package by themselves.
 
dusk883
Zone Monster
offline
Link
 
+1
 
InRomoWeTrust
Lead Mod
offline
Link
 
I'm with Novus on this one. I understand that the percentage biases operate as a ratio of the total amount, but from my experience auto adjust probably isn't operated with that in mind.

Have to wait until the next Q&A
Edited by InRomoWeTrust on Jan 17, 2013 20:16:45
 
The Hitman
offline
Link
 
Originally posted by InRomoWeTrust
I'm with Novus on this one. I understand that the percentage biases operate as a ratio of the total amount, but from my experience auto adjust probably isn't operated with that in mind.

Have to wait until the next Q&A


OK I ran another experiment and I will have to admit it works better than 10+ seasons ago, before I went to the 100%s.

BUT - I would still like to see a 'success slider' because I'd like to be able to control what I consider to be a success.
 
InRomoWeTrust
Lead Mod
offline
Link
 
Originally posted by The Hitman
I would still like to see a 'success slider' because I'd like to be able to control what I consider to be a success.


truf
 
bhall43
offline
Link
 
I get the feeling that would kill a lot of gameplans if we controlled what we consider to be success in the current environment.
 
dbreeze
offline
Link
 
Originally posted by The Hitman
........................BUT - I would still like to see a 'success slider' because I'd like to be able to control what I consider to be a success.



here here! +1

noone has explained AA quite as I see it so I'll pray I'm right and just keep quiet......
 
T2
Killuh
offline
Link
 
but OMGZ OCs, what happens when the defense has a success slider and our successes collide????!!!!! WHICH SLIDER WILL BREAK???!!

+1

 
Primate
offline
Link
 
I think it would be nice to know what defines success. Is 2 yards a success for an inside running play? Is 4 yards a success for a screen?
 
Page:
 


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