Win11 vs XP - simulation results

[Dean-Atlanta]

Here's an example:

I create a game and generate 300 copies of it:

- Win11 - I have a 1.5 to 1 TD ratio vs my opponent
- XP - I have a 2.5 to 1 TD ratio vs my opponent

Same teams, same PPPs, same weather...

I just realized something...

I think you are using separate PC, so I assume you ae doing this:

Create game file on PC #1, duplicate it 300 times.

Create game file on PC #2, duplicate it 300 times.

Then you run the sims.

Creating those initital gam,e files on separate PCs might be accounting for that difference in scoring. Try this instead:

Create game file on PC #1. Duplicate it 300 times. Copy all 300 games file to a zip drive or put them on a shared folder if your PCs are networked. Copy those same exact 300 game files to PC #2. Sims same teams and PPPs on both PCs and see if you get that much variation in scoring. I suspect they will be closer but not exact. If you sim 1000 on each PC using same exact game files, your variation should decrease.

[Steve-San Diego]

What are you guys using to export game scores for 300 to 3000 sims?

All I see in wlogstats and logger is the ability to export play results, not the score / margin of victory for 100s of sims.

You can't be seriously writing down the scores of 300+ sims?!

[Rich-League Officer]

What are you guys using to export game scores for 300 to 3000 sims?

All I see in wlogstats and logger is the ability to export play results, not the score / margin of victory for 100s of sims.

You can't be seriously writing down the scores of 300+ sims?!

Steve, no score results are necessary.....from a large group of sims you can get TD data/FG data which tells the story.
In any batch of games, there are always going to be losses.
It also can give you TD to INT ratio which is useful.
If I was simming 5000 games, I'd be checking TD ratio against my opponent, my own TD/INT ratio, fumbles, FG %, KR/PR numbers....then obviously trying to improve on those by weeding out plays. Can't worry about the score.

[Dean-Atlanta]

I get the following from winlogstats:

Comp percent for both teams QBs
TDs scored by passing, both teams
INT by both teams
Fumbles by both teams
FGs by both teams
Rushing TDs by both teams

I wrote via AI a Python script that scans the log and tells me:

Points scored on avg per game by both teams (such as Denver 22.7, Atlanta 26.23)
wins per game (seeing more ties than I would epxect in sims) such as Atlanta 364, Denver 120, tied 16

This script takes seconds to run and spit out those numbers

I think it was on Discord one Friday night that Donovon discussed using AI to analyze a log file and suggest the most effective plays to use. I have nopt tried that yet.

I also use the tool that Brian created to convert the winlogstats DB to excel spreadsheet format.

I still think one of the most challenging parts of this process, no matter how good we get at doing the rest of it, is scouting the other team we play each week. That stillis quite time cosnuming fi it is to be done well. Using tools, scripts, AI, etc. can reduce the time needed to do the other parts of the game planning process.

Knowing how to all this well, Rich, are you eager to take one of those 5 remaining CPU teams in the PCFL and show us all how it's done?

[Mitch-Dolphins]

I used to sim 24,000 games in a week sometimes. However, I learned a while back that traditional simming was not as predictable as it needed to be for the time commitment.

[Justin-Chicago]

I used to sim 24,000 games in a week sometimes. However, I learned a while back that traditional simming was not as predictable as it needed to be for the time commitment.

So are you now doing non-traditional simming?

[Dean-Atlanta]

If we are talking about 24,000 simmed game in a week doing "traditional simming" that would be 80 game plans simmed 300 games each. I don't know there are enough plays to populate 80 game plans, so maybe that is 40 game plans simmed at 600 game each?

I'm not going to ask any of you what you are simming, how many game plans, how you are simming etc. all of that gets into divulging details of our own process, and I suspect a few of you are ahead of mos of us in how you do this, as exhibited by your win-loss records comapred to the rest of us.

Rich said on the show last season that with any simming process we do it is still "garbage in, garbage out" depending on what you input into your system. In the context of the discussion, In understood the inoput being referenced to be the scouting we do of the opposiong team. I think my scouting was way off at time last season (e.g. the Green Bay game) and pretty good at other times, such as my gams vs Denver or San Francisco. I re-examinied my game planning process in the off-season, and realized, no matter how much I improve the process, I still need toget better at scouting.

Traditional simming it still useful. It helps you become more competitive, but I went 5-11 several times relying on that. I think coahces like Rich and William and Thomas did this but they did so much more. The rest of might only be beginning to figure out what the so much more is.

[Matt-Jacksonville]

What are you guys using to export game scores for 300 to 3000 sims?

All I see in wlogstats and logger is the ability to export play results, not the score / margin of victory for 100s of sims.

You can't be seriously writing down the scores of 300+ sims?!

I won't give up the secret sauce, but there are ways without writing down the scores of 300+ sims.

[Dean-Atlanta]

I had AI write that Python script I mentioned, which isthe "secret sauce" for figuring out how many wins and losses and avcg points scored/given up in simmed games. Without revealing which team I simmed again, I get output that looks like this:

Team Name | Games | Avg Score
---------------------------------------------
Atlanta | 100 | 22.81 PPG
(team) | 100 | 24.84 PPG
=============================================

Team Name | Wins | Losses | Ties | Win %
-------------------------------------------------------
Atlanta | 46 | 48 | 6 | 49.0%
(team) | 48 | 46 | 6 | 51.0%
=======================================================

The ties are because the script reads the last reported score in each game and a game wonin overtime ends with the final score showing inthe log, so the scrip thiknis those are ties.

This works on 100 games, 300, evne 1000 games simmed.

[Mitch-Dolphins]

What are you guys using to export game scores for 300 to 3000 sims?

All I see in wlogstats and logger is the ability to export play results, not the score / margin of victory for 100s of sims.

You can't be seriously writing down the scores of 300+ sims?!

I won't give up the secret sauce, but there are ways without writing down the scores of 300+ sims.

Assistant Coach GPT does it for me.