Analyzing a league for parity, balance and realism

Sliders, settings and other adjustments to make the game more realistic.
User avatar
Danchat
Hall of Fame
Posts: 2709
Joined: Wed Jul 27, 2011 9:07 pm
Location: Minnesota

Re: Franchise guide for the Grudlies universe

Postby Danchat » Wed Jan 04, 2012 11:12 pm

I could find some time to help with this. I've always wanted balanced stats............
One thing I need is a stats site which tells us "yards after first hit/contact" for a RB, and "yards after catch" for a receiver.

I'll take a look around the internet for some stat sites.
And before you ask, I don't program in VB and I don't have visual studio.
Franchise Journal - Atticus Jenkins: Head Coach of the San Francisco 49ers
http://www.footballidiot.com/forum/viewtopic.php?f=2&t=2387&start=400

Re: Franchise guide for the Grudlies universe

Sponsor

Sponsor
 

User avatar
torontogrudlies
MVP
Posts: 837
Joined: Fri Jul 15, 2011 4:56 pm

Re: Franchise guide for the Grudlies universe

Postby torontogrudlies » Wed Jan 04, 2012 11:37 pm

Danchat wrote:I could find some time to help with this. I've always wanted balanced stats............
One thing I need is a stats site which tells us "yards after first hit/contact" for a RB, and "yards after catch" for a receiver.

I'll take a look around the internet for some stat sites.
And before you ask, I don't program in VB and I don't have visual studio.


LOL...glad we got that out of the way!

I programmed the "box score generator" utility. It was the first thing I really did in VB, although I've done other similar programming in the past. The interface is generic and there might be version compatibility issues, but it was a working model. This utility I'm considering, actually will be similar in many ways to the boxscore, and will even use the same core code, changing table and field names. It reads info from various tables in the franchise file, processes them and delivers output (maybe a text file, I was thinking folks might have a use for tracking changes.)

Raston
MVP
Posts: 701
Joined: Thu Aug 11, 2011 6:57 pm

Re: Franchise guide for the Grudlies universe

Postby Raston » Sat Jan 07, 2012 11:22 am

This will be interesting to see if you can get believable results in Franchise by doing some programming. I do alot of simming too and each Franchise sim is different as though there are various scenarios created by programming for the playout of the season. Some stats such as HB yardage, LB interceptions and sacks, receptions, and a few other stats always seem high. Lowering certain physical stats helps but much depends on what profile is loaded for the Franchise season. For example, one sim I will have the QB's fall into believable results for their ratings then another sim they might do fantastic and rate higher than expected. Or I'll have a HB who did good in real life 2011 yet in a Franchise he might do lousy or get hurt.

Administrator
Site Admin
Posts: 83
Joined: Fri Jul 15, 2011 4:30 pm

Re: Franchise guide for the Grudlies universe

Postby Administrator » Thu Jan 12, 2012 11:12 pm

Ok. Here's my initial work on an application which will analyze a franchise file, and compare certain stats and numbers against typical NFL numbers.

This first piece looks at:

1. Average wins by division champions
2. # of "cellar dwellers": teams with < 4 wins
3. # of blowout games, margin of victory > 27 points
4. # of shutout games, one team didn't score

These will hopefully be helpful in figuring out the parity in a league, i.e. how much of a gap separates the upper and lower echelon of teams.

As an example, here's output of the previous season in my franchise:

Code: Select all

Results
-------
Average wins by division champ:  11
(NFL average                  :  11.25)
---------------------------------------
# of teams with < 4 victories :  3
(NFL norm                     :  3)
---------------------------------------
Number of blowouts (28+ points:  11
(NFL norm                     :  22)
---------------------------------------
Number of shutouts            :  3
(NFL norm                     :  9)
---------------------------------------


The NFL numbers, at this stage, are just hard-coded, but in the long run we'd want to have them fed in from a .CSV file, so that they can be customized as desired. The blowouts/shutouts came from a stat in ESPN magazine from last year; the division wins and cellar dwellers are based upon looking at standings for the past four or five years.

This app would ideally be run on a franchise file at the conclusion of the regular season, anytime prior to starting new season, but it's looking at regular-season results only. There's also nothing coded to prevent it from being run during any week in the season, but of course that wouldn't give much in the way of useful info.

Analyzing my own league's results: looked at the past three seasons, and the division wins average/cellar dwellers are pretty much consistent with NFL, but the # of blowouts are about half the NFL norm, and there are significantly fewer shutouts (3, 1 and 4 over the past seasons.)

I'd be curious to know what other franchise owners might get with these numbers? Or if maybe my criteria aren't all that conclusive, and/or need to take different things into account?

User avatar
superben21
MVP
Posts: 622
Joined: Fri Aug 05, 2011 1:44 pm
Location: Ohio

Re: Analyzing a league for parity, balance and realism

Postby superben21 » Fri Jan 13, 2012 10:21 am

I feel like I'm getting incorrect results. I'm running this at the end of the regular season:

Code: Select all

Results
-------
Average wins by division champ:  3.375
(NFL average                  :  11.25)
---------------------------------------
# of teams with < 4 victories :  1
(NFL norm                     :  3)
---------------------------------------
Number of blowouts (28+ points:  3.375
(NFL norm                     :  22)
---------------------------------------
Number of shutouts            :  1
(NFL norm                     :  9)
---------------------------------------

Any Ideas?
~ Superben21 | An active moderator at one point or another.

User avatar
torontogrudlies
MVP
Posts: 837
Joined: Fri Jul 15, 2011 4:56 pm

Re: Analyzing a league for parity, balance and realism

Postby torontogrudlies » Fri Jan 13, 2012 1:04 pm

I will take a look, I must have done something wrong in the code.

I'll bet I did something stupid when I copied and pasted certain pieces of the code, and forgot to have it show the # of blowouts (it seems to be showing the division wins.) This could also explain the confusing results I got...the average division wins, and the # of cellar dwellars pretty much matched the NFL, but the shutouts and blowouts were much lower.

I will correct this in the next update, which is also going to include analysis of the pass to run ratio around the league.

I'm not certain on the division winners average, it worked OK on mine. Would you be willing to allow me to use a copy of your .fra file for more detailed testing? I only have mine to test against, although the win-loss records seem like they should be recorded the same.

User avatar
torontogrudlies
MVP
Posts: 837
Joined: Fri Jul 15, 2011 4:56 pm

Re: Analyzing a league for parity, balance and realism

Postby torontogrudlies » Fri Jan 13, 2012 4:56 pm

Yikes. I just proofread my code and found a couple of glaring errors in it. Like, jdhalfrack and packfan would be rolling over in their graves if they had them. :lol:

This should now work better. I've also added the piece which gives info about the pass to run ratio, although I didn't yet gather up the NFL norms to compare them to.

Sorry for the faulty code!

Edit: I've now added more. Here's an output from my league:

Code: Select all

Results
-------
------------------Parity---------------
Average wins by division champ:  11.75
(NFL average                  :  11.25)
---------------------------------------
# of teams with < 4 victories :  3
(NFL norm                     :  3)
---------------------------------------
Number of blowouts (28+ points):  19
(NFL norm                      :  22)
---------------------------------------
Number of shutouts             :  12
(NFL norm                      :  9)
---------------------------------------

------------Playcall balance-----------
# Pass-heavy teams (> 59% pass):  1
(NFL norm                      :   )
---------------------------------------
# Run-heavy teams (< 41% pass) :  7
(NFL norm                      :   )
---------------------------------------
# Pass/run balanced (41%-59%)  :  24
(NFL norm                      :   )
---------------------------------------
Total pass:  14748
Total run :  16670
Pass %    :  47
---------------------------------------
Number of 4000-yard passers    :  0
(NFL norm                      :  7)
---------------------------------------
Pass completion %              :  59
(NFL norm                      :  )
---------------------------------------
Number of 1000-yard rushers    :  31
(NFL norm                      :  16)

--------------Fumbles------------------
WR/TE Fumble per touch         :  0.025670775924583
(NFL norm                      :  )
---------------------------------------
RB Fumble per touch            :  0.012512140775867
(NFL norm                      :  )
---------------------------------------


Off the bat, a significant fact is that there are no 4000-yard passers, most of the ones in my league top off at around 3300. And there are 31 RBs with 1000 yards rushing, when the pro football norm is about half that. Then the fact that receivers are twice as likely to fumble as RBs.....I still have to complete some of the research of real-life stats, but if I recall correctly the RBs fumble % should be around 4%. And a receiver on average would be more-sure-handed and less likely to get stripped of the ball.
Attachments
Release.rar
draft 3 of league analyzer
(64.35 KiB) Downloaded 223 times

User avatar
superben21
MVP
Posts: 622
Joined: Fri Aug 05, 2011 1:44 pm
Location: Ohio

Re: Analyzing a league for parity, balance and realism

Postby superben21 » Sat Jan 14, 2012 11:25 am

Well here are the new results from the first year of my chise.

Code: Select all

Results
-------
------------------Parity---------------
Average wins by division champ:  10.875
(NFL average                  :  11.25)
---------------------------------------
# of teams with < 4 victories :  1
(NFL norm                     :  3)
---------------------------------------
Number of blowouts (28+ points):  13
(NFL norm                      :  22)
---------------------------------------
Number of shutouts             :  2
(NFL norm                      :  9)
---------------------------------------

------------Playcall balance-----------
# Pass-heavy teams (> 59% pass):  2
(NFL norm                      :   )
---------------------------------------
# Run-heavy teams (< 41% pass) :  0
(NFL norm                      :   )
---------------------------------------
# Pass/run balanced (41%-59%)  :  30
(NFL norm                      :   )
---------------------------------------
Total pass:  16355
Total run :  14931
Pass %    :  52
---------------------------------------
Number of 4000-yard passers    :  0
(NFL norm                      :  7)
---------------------------------------
Pass completion %              :  58
(NFL norm                      :  )
---------------------------------------
Number of 1000-yard rushers    :  27
(NFL norm                      :  16)

--------------Fumbles------------------
WR/TE Fumble per touch         :  0.0220349967595593
(NFL norm                      :  )
---------------------------------------
RB Fumble per touch            :  0.0113584137614077
(NFL norm                      :  )
---------------------------------------

I need to adjust coach sliders to make more passing i think. :roll:

Thanks for the work! This should really help me get a better understanding of what I need to change in my chise. I like your work a lot :D

EDIT: FWI the reason for the low amount of less than 4 win teams is probably due to fantasy draft...
~ Superben21 | An active moderator at one point or another.

forhuskers
Rookie
Posts: 9
Joined: Mon Nov 14, 2011 10:27 pm

Re: Analyzing a league for parity, balance and realism

Postby forhuskers » Sat Jan 14, 2012 12:04 pm

I'm not sure you'll ever get what you are looking for. I was part of Red Wolf's project over on FF. I think we, as a group, got everything about as close as it can get. This was for coach mode and sims. We ran a bunch of seasons (actually the same one over and over) After each weeks games that stats would be analyzed looking for odd results and how the stats compared to the NFL actuals. I really think the sliders and settings we (mostly Red Wolf) came up with were the best possible. I've got a word document showing our 3 year averages compared to the NFL. It just wasn't possible to get everything balanced to the NFL.

Another thing we found is that it seemed that games that were played (watched games either CPU-CPU or human-CPU) used a different engine than pure sims. Simming really seemed to throw the stats out of whack. To get the best results you would need to run every game every week without simming any. It would take a long time to complete a season!

Someone posted Red Wolf's sliders on here somewhere. I hope this may give you a little perspective on what you are getting in to! :D

User avatar
torontogrudlies
MVP
Posts: 837
Joined: Fri Jul 15, 2011 4:56 pm

Re: Analyzing a league for parity, balance and realism

Postby torontogrudlies » Sat Jan 14, 2012 12:23 pm

forhuskers wrote:I'm not sure you'll ever get what you are looking for. I was part of Red Wolf's project over on FF. I think we, as a group, got everything about as close as it can get. This was for coach mode and sims. We ran a bunch of seasons (actually the same one over and over) After each weeks games that stats would be analyzed looking for odd results and how the stats compared to the NFL actuals. I really think the sliders and settings we (mostly Red Wolf) came up with were the best possible. I've got a word document showing our 3 year averages compared to the NFL. It just wasn't possible to get everything balanced to the NFL.

Another thing we found is that it seemed that games that were played (watched games either CPU-CPU or human-CPU) used a different engine than pure sims. Simming really seemed to throw the stats out of whack. To get the best results you would need to run every game every week without simming any. It would take a long time to complete a season!

Someone posted Red Wolf's sliders on here somewhere. I hope this may give you a little perspective on what you are getting in to! :D


Thanks! I guess if the two things (simming vs. played/watched games) end up being too disconnected, another option might be the total control sim. In my franchise, I don't play daily....in fact, in duration, the season I just finished lasted about as long as a real NFL season. In between, I had practice, playbook design, doing the journal and doing GM stuff.... so if I just had to total control sim every CPU-CPU game, it wouldn't be horrible. For someone who plays a lot more, if they're completing an entire season in a few weeks' time, this might be cumbersome though.

Hopefully I'm not entirely reinventing the wheel with this stuff, as you'd mentioned the research done on FF. I guess one new thing which I'm (hopefully) bringing to the table is this utility, which spits out all these stats instantly in one place, and we can then make adjustments then re-simulate.

Naturally, lots of the stuff from FF is gone, but if you still have any data (such as the word doc) I would love to take a look! Thanks....


Return to “Sliders and adjustments”

Who is online

Users browsing this forum: No registered users and 4 guests