Posted by norbert on 5/9/2012 2:51:00 PM (view original):
Let me "clue" you in to how the engine actually works.    A low rating doesn't trigger a poor performance from a player just because it is low.  When checking ratings of a player for performance, it factors in all the ratings against all the ratings of the defender, or in case of team versus team checks, the average of the players.  So for instance, if a formula factors in STR, SPD, ELU, and GI, it averages those ratings into that check so one low rating can be insignificant if the other ratings are higher.

This is something I'd like to change.  I'd like to see a high SPD, low STR guy perform as you would expect, and perform differently than a high STR, low SPD guy.   To do that requires rebuilding the engine to break out those checks so I can compare a player's or team's STR against the opponent's STR when necessary, or SPD versus SPD.  That will allow us to check if a player is evading tackles or breaking through them.  However, in order to do this, I pretty much have to gut how the engine is currently simulating the plays and that means a lot more work than just trying to tweak the current engine.  I can't totally eliminate random results, but I'm really really going to try to limit them.  The key will be to allow everything to work through the beta period and get some good testing in, allowing us to push out the new engine with some new expectations, but not so different that people that have been playing are put off.

Anyway, you don't have to insult me to get my attention.

Posted by llad_lu76 on 5/9/2012 3:06:00 PM (view original):

Posted by norbert on 5/9/2012 2:51:00 PM (view original):
Let me "clue" you in to how the engine actually works.    A low rating doesn't trigger a poor performance from a player just because it is low.  When checking ratings of a player for performance, it factors in all the ratings against all the ratings of the defender, or in case of team versus team checks, the average of the players.  So for instance, if a formula factors in STR, SPD, ELU, and GI, it averages those ratings into that check so one low rating can be insignificant if the other ratings are higher.

This is something I'd like to change.  I'd like to see a high SPD, low STR guy perform as you would expect, and perform differently than a high STR, low SPD guy.   To do that requires rebuilding the engine to break out those checks so I can compare a player's or team's STR against the opponent's STR when necessary, or SPD versus SPD.  That will allow us to check if a player is evading tackles or breaking through them.  However, in order to do this, I pretty much have to gut how the engine is currently simulating the plays and that means a lot more work than just trying to tweak the current engine.  I can't totally eliminate random results, but I'm really really going to try to limit them.  The key will be to allow everything to work through the beta period and get some good testing in, allowing us to push out the new engine with some new expectations, but not so different that people that have been playing are put off.

Anyway, you don't have to insult me to get my attention.

*Thank you* for the response, Norbert, and it would be great if there could be greater differentiation amongst attributes that we could take advantage in game planning, setting the depth chart, etc.  For instance, if we have a high spd, low str RB, wouldn't we want him running aggressive out of trips, but the low spd, high str RB could run conservative from the I or ND Box.  To test this empirically, say both RBs average 5 yds/carry in the above scenario, but we could switch them, and see performance of *both* decline.  The point is that we're looking for an interaction between attributes and gameplanning, and that is difficult to discern with the 'random' noise in the current engine.  Anyhow, it would be great if we could set the depth charts specifically for this....

Posted by llad_lu76 on 5/9/2012 3:06:00 PM (view original):

Posted by norbert on 5/9/2012 2:51:00 PM (view original):
Let me "clue" you in to how the engine actually works.    A low rating doesn't trigger a poor performance from a player just because it is low.  When checking ratings of a player for performance, it factors in all the ratings against all the ratings of the defender, or in case of team versus team checks, the average of the players.  So for instance, if a formula factors in STR, SPD, ELU, and GI, it averages those ratings into that check so one low rating can be insignificant if the other ratings are higher.

This is something I'd like to change.  I'd like to see a high SPD, low STR guy perform as you would expect, and perform differently than a high STR, low SPD guy.   To do that requires rebuilding the engine to break out those checks so I can compare a player's or team's STR against the opponent's STR when necessary, or SPD versus SPD.  That will allow us to check if a player is evading tackles or breaking through them.  However, in order to do this, I pretty much have to gut how the engine is currently simulating the plays and that means a lot more work than just trying to tweak the current engine.  I can't totally eliminate random results, but I'm really really going to try to limit them.  The key will be to allow everything to work through the beta period and get some good testing in, allowing us to push out the new engine with some new expectations, but not so different that people that have been playing are put off.

Anyway, you don't have to insult me to get my attention.

*Thank you* for the response, Norbert, and it would be great if there could be greater differentiation amongst attributes that we could take advantage in game planning, setting the depth chart, etc.  For instance, if we have a high spd, low str RB, wouldn't we want him running aggressive out of trips, but the low spd, high str RB could run conservative from the I or ND Box.  To test this empirically, say both RBs average 5 yds/carry in the above scenario, but we could switch them, and see performance of *both* decline.  The point is that we're looking for an interaction between attributes and gameplanning, and that is difficult to discern with the 'random' noise in the current engine.  Anyhow, it would be great if we could set the depth charts specifically for this....

Posted by norbert on 5/9/2012 3:24:00 PM (view original):

Posted by llad_lu76 on 5/9/2012 3:06:00 PM (view original):

Posted by norbert on 5/9/2012 2:51:00 PM (view original):
Let me "clue" you in to how the engine actually works.    A low rating doesn't trigger a poor performance from a player just because it is low.  When checking ratings of a player for performance, it factors in all the ratings against all the ratings of the defender, or in case of team versus team checks, the average of the players.  So for instance, if a formula factors in STR, SPD, ELU, and GI, it averages those ratings into that check so one low rating can be insignificant if the other ratings are higher.

This is something I'd like to change.  I'd like to see a high SPD, low STR guy perform as you would expect, and perform differently than a high STR, low SPD guy.   To do that requires rebuilding the engine to break out those checks so I can compare a player's or team's STR against the opponent's STR when necessary, or SPD versus SPD.  That will allow us to check if a player is evading tackles or breaking through them.  However, in order to do this, I pretty much have to gut how the engine is currently simulating the plays and that means a lot more work than just trying to tweak the current engine.  I can't totally eliminate random results, but I'm really really going to try to limit them.  The key will be to allow everything to work through the beta period and get some good testing in, allowing us to push out the new engine with some new expectations, but not so different that people that have been playing are put off.

Anyway, you don't have to insult me to get my attention.

*Thank you* for the response, Norbert, and it would be great if there could be greater differentiation amongst attributes that we could take advantage in game planning, setting the depth chart, etc.  For instance, if we have a high spd, low str RB, wouldn't we want him running aggressive out of trips, but the low spd, high str RB could run conservative from the I or ND Box.  To test this empirically, say both RBs average 5 yds/carry in the above scenario, but we could switch them, and see performance of *both* decline.  The point is that we're looking for an interaction between attributes and gameplanning, and that is difficult to discern with the 'random' noise in the current engine.  Anyhow, it would be great if we could set the depth charts specifically for this....

I agree.  The first hurdle is to actually have there be differences in performance in those different situations for different ratings.  If a high STR back doesn't run between the tackles better than a low STR back, then everything else is meaningless.  So the huge task for this update is to make that happen in the engine.  I'll always be fighting randomness on single plays, but I should be able to provide expected results when looking at games overall.  If we can get the ratings and match ups to produce more effect in the engine, then it opens up a lot more game play.

Posted by sammyboy4022 on 5/9/2012 4:06:00 PM (view original):
I think a depth chart for all formations with all positions being on each depth chart is unrealistic. i.e if a team goes from 2nd and 7 I formation to third and 6 trips they dont sub out Olineman. However I would agree a basic depth chart with the whole team and a formation depth chart with skill positions only being changeable would work well and be following real life situations as well.  

Posted by mal247 on 5/9/2012 7:30:00 PM (view original):
Norbert,

I apologize for insulting you. My post was emotionally charged to say the least. I finished #1 in press and coaches but #4 in WIS, so I had the toughest draw of the #1 seeds. That being said, my team was MUCH better than his, and I lost. This low STR guy didn't lose me the game...I think the 6 turnovers did.

That said, it seems as if your post indicates that the running game is determined by the average speed, strength, elusiveness, GI and technique of ALL players on the field? Or just the OL + RB? Can you clarify?

It would help mitigate some of my frustration with this engine and could change the way I recruit certain players.

Posted by kilgo on 5/10/2012 10:35:00 AM (view original):
this might be single most informative thread ever in regards to the engine

Posted by muckraker on 5/10/2012 7:28:00 PM (view original):
Norbert, I appreciate what you are doing and know you're doing all that you can, but you say the update is still so far away.... I'm having a hard time staying excited with the current randomness of the game and the thought of paying for another five or so seasons before the update takes place, well........ Is there anything we can do to get the attention of your bosses to make this more of a priority? I realize this isn't a big revenue generator for them, but i would think this has been a pr nightmare at times.  Basically, is there anything we can do other than try to be patient? The continued drop in the #s isn't a good sign.