Now that Norbert has told us that the beta period is getting closer, We will be changing or hats from wishful game idealists to critical examiners of the new engine. I would like to ask the other coaches - What things are you going to focus on to determine how the new engine is working? How are you going to do it? What factors are the most important to get right? What things do we need to work together on to test?

Sonce I won't feel that winning a NC in the beta means much, maybe we will need to share strategies during our games against each other to see what works and what doesn't. What do y'all think?

I think that we should share as much as possible our setups vs each opponent and try various things with knowledge beforehand.

For me, I'll be sticking with D3 and D2 teams where I have most of my current experience.  I plan on having two teams I hope.  One of them will definitely be wishbone and 4-3/Nickel based.  The other will be Trips/I-Formation and 4-3/Nickel based.

My primary goal will be to test the wishbone in all combinations available.  My secondary goal will be to test defensive settings.  I figure there will be plenty of folks testing Trips, I-formation and defenses....so my contribution will be mostly the wishbone stuff, since there are not that many coaches that run it significantly.

I have tons of data from past seasons and I'll likely be measuring a new data set against my historical data to see how it compares.  The biggest thing I want is that when I choose to run inside.....I want the engine to run it inside (at least way more often than it does now).  Same thing with outside running.

It strikes me that we need LOTS of data to tell us any information.  To do that, I would suggest that coaches set up simple game plans at first....like Pro Set Always Run Outside....or something like that....with the opposing coach running Nickel, blah, blah, blah....then the next game change it to Pro Set always pass vs...Nickel, blah, blah, blah

We probably need some coaches who are willing to do tests like that along with some other coaches who would play a more realistic game approach on both sides of the ball.  We need to see what the engine will do at the individual play level as well as how it performs in an overall complete game setting.

The #1 thing I have got to see, or I'm out the door, is a true cause and effect relationship between attributes and results.

I don't wanna see a 50/50/50 RB outgain a 70/70/70 RB in a game.  Or a team of 50/50 OL pushing around a team of 70/70 DL.

I believe that there will be a page where you can play anyone you want and set the offensive and defensive sets.  I think you can simulate as many games as you want in that page.

I have not seen the page, and it will not show 100% of the PBP (said norbet in a forum post), but I bet you would be able to import the results to Excel or some other spreadsheet (Maybe Google Apps, OpenOffice, LibreOffice, etc.).

Posted by bhouska on 2/4/2013 5:35:00 PM (view original):
The #1 thing I have got to see, or I'm out the door, is a true cause and effect relationship between attributes and results.

I don't wanna see a 50/50/50 RB outgain a 70/70/70 RB in a game.  Or a team of 50/50 OL pushing around a team of 70/70 DL.

+1

Posted by bhouska on 2/4/2013 5:35:00 PM (view original):
The #1 thing I have got to see, or I'm out the door, is a true cause and effect relationship between attributes and results.

I don't wanna see a 50/50/50 RB outgain a 70/70/70 RB in a game.  Or a team of 50/50 OL pushing around a team of 70/70 DL.

I'm not trying to be funny or PITA with this question, but I would like to know.

Is there any room in this comment for an "upset"?  We all know they happen in real life, what would be required in the game for a team with the 50/50 OL to have the RB break a few and the team win the game?

I do totally agree on needing more cause/effect...I would just like to clarify how we understand the cause/effect.

Hope that makes sense.  If not, I'll try again.  LOL

I want to see a cause/effect relationship more intricate than the 70/70/70 RB outgaining the 50/50/50. I think bhouska means this as well, although I won't put words in his mouth.

If the 50/50/50 guy (Player A) outgains 70s guy (Player B), then there needs to be a correlation. Was Team B's DLine tired most of the game? Was Player A playing behind a much superior OL? 

It needs to be a cause effect relationship with intricacies.

Posted by harriswb3 on 2/4/2013 7:17:00 PM (view original):

Posted by bhouska on 2/4/2013 5:35:00 PM (view original):
The #1 thing I have got to see, or I'm out the door, is a true cause and effect relationship between attributes and results.

I don't wanna see a 50/50/50 RB outgain a 70/70/70 RB in a game.  Or a team of 50/50 OL pushing around a team of 70/70 DL.

I'm not trying to be funny or PITA with this question, but I would like to know.

Is there any room in this comment for an "upset"?  We all know they happen in real life, what would be required in the game for a team with the 50/50 OL to have the RB break a few and the team win the game?

I do totally agree on needing more cause/effect...I would just like to clarify how we understand the cause/effect.

Hope that makes sense.  If not, I'll try again.  LOL

I think there is room in Bhouska's comment for an upset.  It would come in the form of coaching and how you set your team up to play.  Generally upsets happen when a team is poorly prepared.  But those are upsets.  All other things being equal, the more talented players usually win in real life (and even at times will overcome poor coaching).  

So here are 2 examples. 

1) Coach A (ranked 5) and Coach B (ranked 105) both run trips and the 4-3.  All the settings are the same.  Coach A has more talented players.  Coach A should win.

2) Coach A (ranked 5) and Coach B (ranked 105) run different sets and the settings are different.  Coach A has more talented players but runs between the tackles using the iformation every game.  Coach B happens to run a 5-2 or a 4-4 defense and plays the defense close to the line which keeps the run damage to a minimum despite facing "better players".  On offense, Coach B uses the shotgun as his only offensive formation and Coach A has the defense set to 4-4 with his defense close to the line.   Coach B is able to take advantage of the thin secondary despite facing better players on defense than the offensive players he is coaching and gets a few deep plays during the game.  Coach B pulls off the upset.  When analyzing the game, both coaches are able to see why the less talented team won.  It is still cause-effect, and the upset as a feature is intact.   

caesari, I totally agree...I suppose what I'm asking is:  how do we validate the "correlation"?  I'm asking because I don't know.  When the game results "look good" then we will be happy, but when the upset happens, what amount and what type of information do we need from the PBP to demonstrate the reasons for the upset?  The current engine clearly does NOT give us enough information.

Maybe its too early for me to ask that, without having seen the new engine in operation.  

Does anyone have some old PBP from GD 1.0?  I never saw that, but I was under the impression that it gave a lot more information. 

Of course that example I gave is a gross oversimplification, but what I was trying to say but most likely didn't make very clear was that ideally the upsets should make some sort of sense...  They don't have to be a mystery that gets written off to the "upsets happen in real football" excuse for strange outcomes.  

Better analytics would be helpful to understand each play better for those with the interest and time to make sense of everything.   My left brain is not nearly strong enough for that type of undertaking but I can tell from some of the posts here that there are pletny of players who can and will make sense of the data if it is given.      

Posted by scrodz on 2/4/2013 7:10:00 PM (view original):

Posted by bhouska on 2/4/2013 5:35:00 PM (view original):
The #1 thing I have got to see, or I'm out the door, is a true cause and effect relationship between attributes and results.

I don't wanna see a 50/50/50 RB outgain a 70/70/70 RB in a game.  Or a team of 50/50 OL pushing around a team of 70/70 DL.

+1

+1

What we will really need from Norbert is the attributes he is using to compare each decision point - and what will modify those attributes and by how much. He has said that such things as athleticism, fatigue and technique will cause some variability. And how much effect will formation IQ play. bhouska's point is the main concern that I run into. SIM teams with 60 pt per player disadvantages playing the better team to a standstill may happen a few plays in a game, but they seem to happen too often for the entire game. I would hope the PBP would give information on which player won the comparison and why. I also hope that the variability doesn't dilute the true attribute values so that again we end up with coain toss outcomes.

Sharing and comparing data and information during the update will help us analyze the new engine. We are the beta TEST, and we need to run this new engine through the wringer.  I am going to do my best to make sure each PBP has an understandable set of information to determine which players are involved and who did well and who did poorly. As I have seen recently - I don't want DE tackling a WR 15 yards downfield on a pass play or OLB covering WR 31 yards downfield on the sideline (both on nickle D). These are engine errors and I hope I can't find any in the update.

Oh yeah, and I definitely don't want to see a team of WR's with HND's in the teens and twenties dropping 1-2 balls per season!  That is ridiculous.

i agree with the correlation comment as long as there is still some variability for upsets.  on the other hand, i'd like to see the new 'player roles' give some extra coaching flexibility for managing the game.  right now it's kinda vanilla, straightforward, with the variability out of our hands.  i'd like a little more say in how the variability turns out =].

Pay no attention to the man behind the curtain (which is the "harriswb3 factor:  if wishbone run, then yards gained = 8).  LOL