Thoughts on the CPU Passing Game

[Sven Draconian]

To this point, I have sliders that I like for the HUM run game, pass game and are solid for the CPU run game (playing on Heisman level). I'm getting great player differential based on ratings and I'm getting pretty realistic play from the lines (again, based on differences in ratings).

My issue is with the CPU pass game. With HUM coverage on 50 and CPU QBA at 65 I get abused in the pass game by poor QBs (to the tune of 70% comp). This is playing as Utah State (and some other awful teams), so I do expect good stats, but I don't see the differential.

However, when I drop the slider for QBA, the CPU gets neutered and every drive they have stalls out.

The other options are playing around with CPU catch, but I have a similar issue. Once it gets too low, the computer becomes ineffective on offense because they drop wide open passes. I want them to drop passes when they get hit so the over the middle drag route game becomes a bit less effective, but dropping the WR catching they drop the open, deeper passes that they need to be effective.

The pass blocking/pass rush game is pretty decent. With CPU pass block at 0 and HUM pass rush at 70 I get realistic pressure (maybe a touch much). If I drop it lower the pass rush gets a bit low, if I rasie it I lose any difference in player ratings (and my guys turn into ROBO lineman).

And, I have too many turnovers (in general). Especially fumbles (with HUM tackling at 50, hit power at 50). Changing either of those ratings will further enhance the computers ability to throw drag routes all over the place.

Cliff notes:
I think 65 QBA is a "magic" number for good QB decisions.
I can't get proper WR drops (I can get good numbers, but poor timing).

Basically, I want to combat poor AI decisions (mainly their desire to run ineffective plays...), without making Robo QB appear. Thoughts/opinions/research is appreciated.

[StudiousSquid]

Quote: Originally Posted by Sven Draconian

To this point, I have sliders that I like for the HUM run game, pass game and are solid for the CPU run game (playing on Heisman level). I'm getting great player differential based on ratings and I'm getting pretty realistic play from the lines (again, based on differences in ratings). My issue is with the CPU pass game. With HUM coverage on 50 and CPU QBA at 65 I get abused in the pass game by poor QBs (to the tune of 70% comp). This is playing as Utah State (and some other awful teams), so I do expect good stats, but I don't see the differential. However, when I drop the slider for QBA, the CPU gets neutered and every drive they have stalls out. The other options are playing around with CPU catch, but I have a similar issue. Once it gets too low, the computer becomes ineffective on offense because they drop wide open passes. I want them to drop passes when they get hit so the over the middle drag route game becomes a bit less effective, but dropping the WR catching they drop the open, deeper passes that they need to be effective. The pass blocking/pass rush game is pretty decent. With CPU pass block at 0 and HUM pass rush at 70 I get realistic pressure (maybe a touch much). If I drop it lower the pass rush gets a bit low, if I rasie it I lose any difference in player ratings (and my guys turn into ROBO lineman). And, I have too many turnovers (in general). Especially fumbles (with HUM tackling at 50, hit power at 50). Changing either of those ratings will further enhance the computers ability to throw drag routes all over the place. Cliff notes: I think 65 QBA is a "magic" number for good QB decisions. I can't get proper WR drops (I can get good numbers, but poor timing). Basically, I want to combat poor AI decisions (mainly their desire to run ineffective plays...), without making Robo QB appear. Thoughts/opinions/research is appreciated.

I think the key this year is the speed threshold/pass coverage combo. The problem this year is the AI plays man coverage in their psychic mode, with the defenders running the routes before the receivers do.

As a result you don't see separation. I would lower speed threshold to 45 or 40 to start out. The next step would be to lower human pass coverage to 45 or 40.

What ends up happening in man coverage or any coverage is that if your defender makes a mistake the CPU receiver can exploit it and get separation in his route.

The slight drop in pass coverage means the defender doesn't recover as fast from his mistake. The extra speed from the speed threshold slider being dropped allows the receiver the ability to separate a bit.

This allows the CPU QB to find open receivers and move the chains. This is helpful as most humans probably blitz or press a lot. So this makes blitzing/jamming more risk/reward as it should be.

I'd start out with speed threshold at 45, and human pass coverage at 45 and see how it goes. You can tinker with each to see what you like.

Also lower the human interceptions so the cpu QB doesn't average 2-3 picks a game.

[Sven Draconian]

Quote: Originally Posted by StudiousSquid

I think the key this year is the speed threshold/pass coverage combo. The problem this year is the AI plays man coverage in their psychic mode, with the defenders running the routes before the receivers do. As a result you don't see separation. I would lower speed threshold to 45 or 40 to start out. The next step would be to lower human pass coverage to 45 or 40. What ends up happening in man coverage or any coverage is that if your defender makes a mistake the CPU receiver can exploit it and get separation in his route. The slight drop in pass coverage means the defender doesn't recover as fast from his mistake. The extra speed from the speed threshold slider being dropped allows the receiver the ability to separate a bit. This allows the CPU QB to find open receivers and move the chains. This is helpful as most humans probably blitz or press a lot. So this makes blitzing/jamming more risk/reward as it should be. I'd start out with speed threshold at 45, and human pass coverage at 45 and see how it goes. You can tinker with each to see what you like. Also lower the human interceptions so the cpu QB doesn't average 2-3 picks a game.

I've had the threshhold at 35. I've thought of putting it even lower (down to 25). The threshold works both ways though, if my DB is faster he will now make up ground on a slower WR..which can actually fool the AI QB into throwing another bad ball.

My INT's are at 30.

The "psychic" AI is fine. DB's should jump routes. The problem is I want the AI QB to miss receivers...without turning into a ******, which is the dilema. With QBA at 65 they don't miss many passes. However, if you put it below 65 they start making really dumb passes (lobbing a pass to the a covered receiver in the flats). QBA is tied into awareness (or, it feels that way).

[RoyalBoyle78]

Quote: Originally Posted by StudiousSquid

I think the key this year is the speed threshold/pass coverage combo. The problem this year is the AI plays man coverage in their psychic mode, with the defenders running the routes before the receivers do. As a result you don't see separation. I would lower speed threshold to 45 or 40 to start out. The next step would be to lower human pass coverage to 45 or 40. What ends up happening in man coverage or any coverage is that if your defender makes a mistake the CPU receiver can exploit it and get separation in his route. The slight drop in pass coverage means the defender doesn't recover as fast from his mistake. The extra speed from the speed threshold slider being dropped allows the receiver the ability to separate a bit. This allows the CPU QB to find open receivers and move the chains. This is helpful as most humans probably blitz or press a lot. So this makes blitzing/jamming more risk/reward as it should be. I'd start out with speed threshold at 45, and human pass coverage at 45 and see how it goes. You can tinker with each to see what you like. Also lower the human interceptions so the cpu QB doesn't average 2-3 picks a game.

I'm using 55 TH now, I noticed anything lower then 50 Receivers are beating CB with a higher speed rating and that should not be happening.