Warrior General Games

I assume play selection is a dice roll, right? Comparing my playbook to what actually happened per stobies scouter:

Offensive outside play book vs assassins including priority and actual times run:

Outside:
Weak I Sweep Weak (5) - 8 times
Diamond Counter Weak (4) - 4 times
Trips Pitch Weak (5) - 4 times
Weak I Counter Sweep Weak (2) - 1 time
Proset Sweep (5) - 0 times
Strong I Sweep (5) - 4 times
GL Sweep (3) - 9 times
Big I Sweep Left (2) - 2 times


All together I had 32 plays run from the outside playbook.

IF the way it works is: a 5 priority play gets "5 marbles in a bag" and a 2 priority play gets "2 marbles in a bag" (no clue if this is right):

There were "37 marbles" in the outside playbook.

3 of those marbles were for GL Sweep (3 / 37 = 8% chance)
Where standard probability would say on 39 plays - GL Sweep would be run 3 or 4 times -- here it was run 2 or 3 times that. Someone smarter than me could figure up the standard deviation there and how likely it is. I know this is a very small sample size - and I don't really feel like going back and looking at my other games to see if somehow GL Sweep is 'favored' no matter what the playbook says.


But as we've so often discussed: How do we punish teams that spam the same plays over and over -- perhaps we should first be looking at if the sim is actually paying attention to play priorities -- or if the sim is actually favoring certain plays over other plays (like it was doing with Romo's bug where his priorities were not saving correctly for certain plays unless he added them to the playbook in the right order).

In this game as an example - to an observer - it probably looked like I spammed GL Sweeps -- and GL Sweeps were my most run play -- but that wasn't my intention - the sim did it on its own.

Is it possible the sim is preferring certain plays -- ADDING to the spam / repeat play complaints because of a bug or error?

Random number generation for a computer isn't a dice roll; it's a table lookup. As a result, there are always "streaky" spots.

The only way to test the system overall is to use a larger sample size. If a larger sample results in a more appropriate distribution, then you hit a "streaky" spot.

If a larger sample shows a problem, then in a smaller sample it can get really distorted.

My guess is you hit a "streaky" spot.

I think you're trying to get away with an intended spam playbook which you dont want to get blamed for

But anyhow, I somehow have the feeling sometimes that plays that worked have a higher chance to get called right away. Can be wrong tho

What about the position of the play in the playbook. It's always felt to me like it starts on one end of the playbook every time, so the first or last 1 or 2 plays have a better chance of being called than the plays in the middle regardless of their priority. Just my lol observation.

bias definitely DOES NOT work like your marbles in a bag example.

If we weren't competitors I'd educate you here, lol. It's much more like auto-adjust in GLB1. The higher you set a play bias the more volatile it is. A play bias of 5 on a single play with everything else at 1 likely means it's getting called 95% of the downs, especially so if you're running it with success.

Note: defense has it's quirks when things aren't exact matches, especially with WR counts (typical issue for newb DCs).

Originally posted by InRomoWeTrust
bias definitely DOES NOT work like your marbles in a bag example.

If we weren't competitors I'd educate you here, lol. It's much more like auto-adjust in GLB1. The higher you set a play bias the more volatile it is. A play bias of 5 on a single play with everything else at 1 likely means it's getting called 95% of the downs, especially so if you're running it with success.

Note: defense has it's quirks when things aren't exact matches, especially with WR counts (typical issue for newb DCs).

well good news....we aren't competitors

Originally posted by InRomoWeTrust
A play bias of 5 on a single play with everything else at 1 likely means it's getting called 95% of the downs, especially so if you're running it with success.



I'm not trying to learn the tactics better -- I'm pretty comfortable with how things have gone the last 6 seasons.

But are you saying auto adjust is a factor here- and the more successful a play is - the more the sim tends to run it?
If so - then couldn't removing auto adjust as a factor on offense (and sticking with the tactics) remove some of the spam in game plans and generally make people just a little bit happier? (not saying it is a cure - but saying it could be an improvement even if slight)

OP has lost his marbles

Originally posted by DeeVee8
OP has lost his marbles

That was perfect.

Originally posted by InRomoWeTrust
bias definitely DOES NOT work like your marbles in a bag example.

If we weren't competitors I'd educate you here, lol. It's much more like auto-adjust in GLB1. The higher you set a play bias the more volatile it is. A play bias of 5 on a single play with everything else at 1 likely means it's getting called 95% of the downs, especially so if you're running it with success.

Note: defense has it's quirks when things aren't exact matches, especially with WR counts (typical issue for newb DCs).

WR Counts has no factor in play calling what so ever.

Originally posted by InRomoWeTrust
bias definitely DOES NOT work like your marbles in a bag example.

If we weren't competitors I'd educate you here, lol. It's much more like auto-adjust in GLB1. The higher you set a play bias the more volatile it is. A play bias of 5 on a single play with everything else at 1 likely means it's getting called 95% of the downs, especially so if you're running it with success.

Note: defense has it's quirks when things aren't exact matches, especially with WR counts (typical issue for newb DCs).

Originally posted by Parab00n
WR Counts has no factor in play calling what so ever.

Parab00n is correct, WR counts on D plays are meaningless. It's either Pass, Pass/Run, Inside Run and Outside Run. I'm not even sure "Medium Outside Run" is a subset of Medium Pass like it implies. I doubt it. It doesn't seem to fire that way for me.

Pass is the generic control on Distance and now (with the recent changes) Inside/Outside is direction.

Basically everything is based on Pass Short or Pass Medium. If you follow some of the Defensive Playbook & Tactics blog posts I created, then this is explained.

There's dummy-proofing in regards to WR counts. Run some tests, you'll see.

Originally posted by InRomoWeTrust
bias definitely DOES NOT work like your marbles in a bag example.

.

Ok then what is it like? Because no one will say if the placement of the play in the playbook as an effect. We all know the fucking red bars mean jack when it comes to play calling.

I know I added two screen passes to the bottom of my medium playbook against Harrisonburg last game. 1 pip on their "priority" bar. They seemed to get called an inordinately large % of the time.

just in case position plays a role - the above is the order in which the plays were added to my book