New Sticky Note

The new user interface is in preview!

Want to check it out? Click here! (If you don't like it, you can still switch back)

League Forums

Main - General MFN Discussion

Re: The Playcalling Matrix Randomizer

By setherick
9/22/2016 2:56 pm
raymattison21 wrote:
setherick wrote:
I also wouldn't include me in this discussion. My offensive game plans are usually much more complex. They just aren't being called or aren't working. The latter I can fix. The former I can't unless I just use rules.



Much like infinities more complex (using rules) might actually increase the power of the dice roll by lower the possble options.


But I don't limit plays, and I watch the weight of the plays that have the chance to be called, and basically nothing makes any sense mathematically or otherwise.

Re: The Playcalling Matrix Randomizer

By Infinity on Trial
9/22/2016 3:00 pm
setherick wrote:
raymattison21 wrote:
setherick wrote:
I also wouldn't include me in this discussion. My offensive game plans are usually much more complex. They just aren't being called or aren't working. The latter I can fix. The former I can't unless I just use rules.



Much like infinities more complex (using rules) might actually increase the power of the dice roll by lower the possble options.


But I don't limit plays, and I watch the weight of the plays that have the chance to be called, and basically nothing makes any sense mathematically or otherwise.


Also, I limited plays in certain situations because the dice rolls were unreliable, not the other way around.

Re: The Playcalling Matrix Randomizer

By jdavidbakr - Site Admin
9/22/2016 4:02 pm
setherick wrote:
raymattison21 wrote:
setherick wrote:
I also wouldn't include me in this discussion. My offensive game plans are usually much more complex. They just aren't being called or aren't working. The latter I can fix. The former I can't unless I just use rules.



Much like infinities more complex (using rules) might actually increase the power of the dice roll by lower the possble options.


But I don't limit plays, and I watch the weight of the plays that have the chance to be called, and basically nothing makes any sense mathematically or otherwise.


If you could save a gameplan that you feel did not work as you expected and let me know what you named it, I can investigate to make sure something odd is not happening.

Re: The Playcalling Matrix Randomizer

By setherick
9/22/2016 4:18 pm
jdavidbakr wrote:
setherick wrote:
raymattison21 wrote:
setherick wrote:
I also wouldn't include me in this discussion. My offensive game plans are usually much more complex. They just aren't being called or aren't working. The latter I can fix. The former I can't unless I just use rules.



Much like infinities more complex (using rules) might actually increase the power of the dice roll by lower the possble options.


But I don't limit plays, and I watch the weight of the plays that have the chance to be called, and basically nothing makes any sense mathematically or otherwise.


If you could save a gameplan that you feel did not work as you expected and let me know what you named it, I can investigate to make sure something odd is not happening.


Sent you a PM.

Re: The Playcalling Matrix Randomizer

By trslick
9/22/2016 6:58 pm
You guys sound over educated in this stuff, I feel like dumb and dumber! I rolled the dice one
night, in one of the few times I coached baseball,highschool age! The other teams pitcher was
a stud,striking out all the kids.We were down 1 to 0 in the third and I was looking over the book
there it was,our 2 left handed hitters had walked! I told the kids to bat left handed all of them, they said they couldn't, I told them just to crowd the plate and wave the bat a little! It worked,we
won 2 to 1,the kid could not pitch to a lefty and the coach didn't take him off the mound! I was
booed and cussed at, I told them if they could find a rule in the book I broke,we would give the
win to them. I slept fine that night! You find a way.

Re: The Playcalling Matrix Randomizer

By jdavidbakr - Site Admin
9/22/2016 8:36 pm
setherick wrote:
Sent you a PM.


I briefly looked at this, stepping through the code, here's what the first logic flow looked like:

First, the play type was selected - it happened to select outside run on this spin (15% probability). Now, you only had two outside runs in your game plan, both were in 1RB/1TE/3WR personnel, which you had at 10% - however, since these two plays were the only plays available to choose from, obviously one of these plays were called. So, there was a 15% probability - or more than one in every ten plays - that one of those two outside runs would be called.

Remember - the play type is chosen before the play itself is. The weights in the popup kind of get skewed if you don't have a well-rounded set of plays because, as you can see from this example, not having any outside run plays in the personnel that you wanted to run most of your plays from negated any plays from those formations once the play type was selected. To get the results you were wanting, you probably should have reduced or eliminated outside runs from first down in addition to removing the personnel.

Before I dig deeper, does that sound like a plausible explanation from your observations?

Re: The Playcalling Matrix Randomizer

By setherick
9/22/2016 8:41 pm
That could explain some of the second down calls because it looks like outside run was left at 30% for medium.

But this is all kind of proving my point. If I can't trust the weights because they get skewed, why am I always told to use them as a guide for determining what play gets called first? Play weights should probably only show the weight versus the play being called and not factor in personnel at all.

Basically, this is grown irritating beyond belief.
Last edited at 9/22/2016 8:42 pm

Re: The Playcalling Matrix Randomizer

By jdavidbakr - Site Admin
9/22/2016 8:47 pm
setherick wrote:
That could explain some of the second down calls because it looks like outside run was left at 30% for medium.

But this is all kind of proving my point. If I can't trust the weights because they get skewed, why am I always told to use them as a guide for determining what play gets called first? Play weights should probably only show the weight versus the play being called and not factor in personnel at all.

Basically, this is grown irritating beyond belief.


I probably need to dust off my statistics book and review how to best display the probability in this case. The weights are calculated on the popup by multiplying the two sets together, but it breaks down when you have an uneven distribution of plays through the formations and play types.

Re: The Playcalling Matrix Randomizer

By setherick
9/22/2016 8:48 pm
jdavidbakr wrote:
setherick wrote:
That could explain some of the second down calls because it looks like outside run was left at 30% for medium.

But this is all kind of proving my point. If I can't trust the weights because they get skewed, why am I always told to use them as a guide for determining what play gets called first? Play weights should probably only show the weight versus the play being called and not factor in personnel at all.

Basically, this is grown irritating beyond belief.


I probably need to dust off my statistics book and review how to best display the probability in this case. The weights are calculated on the popup by multiplying the two sets together, but it breaks down when you have an uneven distribution of plays through the formations and play types.


This, coincidentally, was how I thought that plays were being determined. But, yeah, displaying a union of sets here isn't the easiest thing. At least I know now which number is actually controlling play selection. (Now, it'll be a question of whether it is easier to evenly distribute the personnel sets and let play selection determine the play run or do more aggressive mathematics when determining my game plan or just use a three play game plan. The last sounds the most preferable.)
Last edited at 9/22/2016 8:51 pm

Re: The Playcalling Matrix Randomizer

By jdavidbakr - Site Admin
9/22/2016 8:57 pm
setherick wrote:
jdavidbakr wrote:
setherick wrote:
That could explain some of the second down calls because it looks like outside run was left at 30% for medium.

But this is all kind of proving my point. If I can't trust the weights because they get skewed, why am I always told to use them as a guide for determining what play gets called first? Play weights should probably only show the weight versus the play being called and not factor in personnel at all.

Basically, this is grown irritating beyond belief.


I probably need to dust off my statistics book and review how to best display the probability in this case. The weights are calculated on the popup by multiplying the two sets together, but it breaks down when you have an uneven distribution of plays through the formations and play types.


This, coincidentally, was how I thought that plays were being determined.


That actually used to be the case, I changed it very early on because it was just as confusing for people back then. What happens is, again, if you don't have a balanced active playbook - let's say you have 30 pass plays and 10 run plays - and you have a situation where you are 80% run and 20% pass, you are -not- going to run 8 times out of ten, because each pass play has a weight of 20 and each run play has a weight of 80, but the sim of 30 pass plays at weight 20 = 600 and the sum of 10 run plays at weight 80 = 800, so you will only run 57% of the time (simple example, assuming all formations carry the same weight, but it can be extrapolated further given the formations). I think the number of users who focus on run/pass ratio is much greater than those who are dialing into the formations.

That's ultimately the problem we're running into - as always, I'm open to suggestions if you think there's a better way to present it - maybe breaking the 'view plays' down by play type, i.e. '25% short pass' headings followed by the list of all available short pass plays and their relative weights. Actually, as I type this I think this may be a good resolution to this issue.