Warrior General Games

... Auto Adjust. 6 settings. Off, Very Quickly, Quickly, Medium, Slowly, Very Slowly. Been using Very Quickly for many seasons and today the 2x4 of logic hit me upside the head and asked me WHY? And I didn't have an answer other than logic. Which brings me here.
Has anyone done it differently and stuck with it?
Has anyone had any extensive work with the variations and learned anything useful?
Is it all just crap and I should ignore it completely?
Is off really 'OFF'?
Any help, insights, and thoughts appreciated. Sometimes it's just good to have other points of view.

I use Quickly but I have no good reasoning why. Hope this helps.

I use off from when started OCing. I scout every game that matters and if I make a mistake or DC is smarter so be it. Don’t want cpu mess with my tactics, why let Borf guided AI take over my offense?

Very quickly for this reason. If everything goes as expected, it will have no reason to adjust, however if something isn't working as expected, then it can compensate by calling one of the other plays I've prepared for the game. Thats literally exactly what I'd do if it was a real game of football.

The more I use a play throughout the offense, the lower I set the auto adjust. My preference is 3 plays each in two or three separate packages with all set very quick.


Originally posted by Theo Wizzago
... Auto Adjust. 6 settings. Off, Very Quickly, Quickly, Medium, Slowly, Very Slowly. Been using Very Quickly for many seasons and today the 2x4 of logic hit me upside the head and asked me WHY? And I didn't have an answer other than logic. Which brings me here.
Has anyone done it differently and stuck with it?
Has anyone had any extensive work with the variations and learned anything useful?
Is it all just crap and I should ignore it completely?
Is off really 'OFF'?
Any help, insights, and thoughts appreciated. Sometimes it's just good to have other points of view.

Hmmm... was thinkin mostly the same. When I first started my stuff was way over-complicated and Very Quickly worked for that... but now that I've pared down everything to much less that the AI has to have to choose from, I began to wonder if Very Quickly wasn't messing with things.

Yeah completely depends on your own Ai. Packages have their own Auto Adjust also. There are times when very quickly isn't fast enough to get back to a big play, and times when a lucky roll by the other team makes VQ kick in too much.

There's often a bit of value to be had in it though. Like a 3rd and 2 for example. In the Big i formation, you can Blast Left or Right for the A-gap. You can Slam Left or Right, for the B-gap. You can Off Tackle left or right for the C-gap.

Depending on how the d-line is set against that formation, one of those gaps could be much more reliable than the others. You don't really care which run you use, you just need 2+ yards. It would be swell if the Ai could pick out the one or ones that were working and avoid the ones getting stuffed.

or just Slam it and pray, twice if you need to

Dr. E alluded to something interesting back there...number of options. I find Auto-Adjust works much better in a package of 6 plays than with a package of 10 or 12.

Type of play matters too, and the volatility of that play. To use my Palookas as an example, if I have a package with both inside and outside plays in it, we might call an outside play that our opponent anticipated and set up his defense to stop it. But if our speed-QB gets lucky, breaks a couple of tackles, and goes off for 75 yards, Auto-Adjust will crank up that play for a while -- setting up a parade of 2-yard losses while the inside runs in the same package get neglected. So we keep different types of plays separate from each other in their own packages.

My GLBrain isn't what it used to be.

But I think that I might have thought it was 15%,12%,9%,6%,3% change for successful/unsuccessful play outcomes, based on whether you were very quickly (15%) to very slowly (3%).

For rare play scenarios (like 3rd and short, or 4th and short) I think I set to very-quickly, since I'd only load 2-4 options and it only happens so many times a game, so I want the sim to adjust ASAP. In my mind if 4 plays were 25% each, then one was successful it'd becomes 40%-20%-20%-20%. If the same play was succesful again it'd become 65%, 12%, 12%, 12%. Not sure if that based on a bort quote, or based on my own conjecture though.

For other more used scenarios I still used quickly since I don't think the other settings matter enough to make a dent. Still would try to limit plays somewhat, since otherwise the auto-adjust will never have time to actually do anything worthwhile.

I use medium and always have. I also don't use packages. My experience, packages add another later of auto adjust, and often times the AI wants to get stuck calling the same handful of plays no matter how many are in a package. My O-AI's started to blossom when I rid them of packages, set 10-15 plays as outputs to each input and then used only the auto adjust that is built into the input function of the AI. This gave me the most consistency.

Well, TBPH, my OAI's work pretty good overall. Each season I learn just a little bit more and make improvements but the process I started with has mostly remained the same. I use %'s inside the packages to dictate which plays fire most often (or should, anyways). I have learned to limit the number of plays in a package. I almost never go more than 5 and prefer 4. Hell, I even have a few packages that only have 2 plays in them. But %'s tend to play out as designed... mostly. Also, the %'s inside the AI's themselves further limit which packages get run in which situation which further limits the chances of some plays happening. So with all that, mostly I get something like this;
Out of 10 times I run an OAI about 7 of those I get a decent, but average result. Always good enough for a win as long as everything else works too (Defense and ST's). About twice out of those 10 times I get pure perfection. I mean it's like the OAI knows exactly what defense is on the field and calls the perfect play for that situation. Maybe not for every single play but when it counts it's always spot on. I love those games. But... 1 out of every 10 times... it gets seriously stupid. I >KNOW< my packages and the percentages and all that so when I see a low % possibility play get called 3 or 4 times in a game... and it's not only not working but not working in a bad way... and it's not the only low % play getting called... I just can't figure it. To me it would be like having weighted dice that always went AGAINST the weights and rolled opposite of what they're designed to roll. Head-shaking. So I began to wonder about the one thing I haven't changed since I started this game and that's the Auto Adjust. I don't know if the 1-in-10 wacky results are from what I'm doing/not doing... or is it all just part of the Bort master plan that every now and then, sh*t gets crazy. I can live with it if it's the latter... but I can fix it if it's not so I just gotta figure out which one it is. I don't want to mess with success... I just want to improve on it.

Originally posted by Kenshinzen
I use off from when started OCing. I scout every game that matters and if I make a mistake or DC is smarter so be it. Don’t want cpu mess with my tactics, why let Borf guided AI take over my offense?

It simply changes the % frequency of your own plays/packages, it doesn't resort to "Bort AI" unless you're talking about the casual settings, or your packages didn't total up right, etc.

3/6/9/12/15% adjustment. Which one fits, depends how many packages you're using (if at the OAI level), or how many plays you've stuffed in (if at the package level).

I tend to have a fair number of packages, but keep the plays within them tight - so I go slowly at the OAI level, and quickly inside the packages.

Originally posted by Theo Wizzago
But... 1 out of every 10 times... it gets seriously stupid.

Just a thought, but if the low % calls you keep seeing that are "not working" are screens ... There's a problem when completed screens for a loss = play was a success. So that could be part of the problem if *most* everything else is failing, and it's adding 15% divided between the packages that weren't just called - you can find yourself auto-adjusting to calling almost nothing but screens if it's set too high, and your other packages/plays are having a rough go.

Wheee, ignore the double post.

Originally posted by Gambler75
Just a thought, but if the low % calls you keep seeing that are "not working" are screens ... There's a problem when completed screens for a loss = play was a success. So that could be part of the problem if *most* everything else is failing, and it's adding 15% divided between the packages that weren't just called - you can find yourself auto-adjusting to calling almost nothing but screens if it's set too high, and your other packages/plays are having a rough go.



No... not specific plays like screens. So I'll try to give example. Maybe on 4th and 5 yards, out of FG range but close and kinda too close to punt (say 37 yard line for a low/medium leveled team?) and I'd rather go for it than not. So in my '4th and going for it' setup... most of the plays in some of the packages are pass plays that work well and are correct distance. Also, since we're talking 4th and long, all pass play packages get the highest % or which one's the AI most likely chooses.
But one package is run plays (lowest % chance of being selected). Inside the run package there are 4 plays. The play most likely to work gets the highest % assigned to it. So maybe one play is a HB sweep using the scat HB to catch the defense off guard maybe if tagging as a 'receiver'. Just enough to sneak out 5 and an inch, eh? But because it's a risky play it's something I ONLY want to see once in a great while so it gets the LOWEST % chance assigned to it.
Further... lets just say that, somehow, that scenario happens 3 times in that game... and ALL 3 times it calls the same, lowest %, scatback run play. And all 3 times it fails miserably. And it only happens in that game and I won't see anything like it for another 9 or more games. I'm trying to figure out why the AI called that play over and over when, clearly it wasn't working... and clearly it was the LOWEST likely play to occur since passing was given the most % and even in the run plays it was the 'least %' of possibly happening.
So I'm trying to figure out why the %'s and everything get ignored once in a blue moon. I can eliminate several things except for 'That's just Bort baby' or 'Your auto adjust is too high'. One I can fix... the other I just live with.