NOTICE: This league is using the BLEEDING EDGE game engine. For more information, click here.

The new user interface is in preview!

Want to check it out? Click here! (If you don't like it, you can still switch back)

League Forums

Main - Beta Chat

Re: Version 0.4.3 Release Candidate Discussion

By raymattison21
5/21/2018 3:01 pm
setherick wrote:
Enjoy being the only voice on the boards Ray. You're criticizing me playing a player more than you are realizing that the bad players are highlighting more problems with the game than the good ones are. The broken coverage above would have happened if the player would have had 90 SP and 50 M2M. Stutter still breaks coverage code.

To that end, let's just release 0.4.3 because it's going to be fun to exploit.


Sorry, if I was over critical here. I was presenting another possible thought in order to create some dialog or other experiences on the subject . I respect you as a person/ gamer and enjoy your thoughtful critiques as well.

Still, I don't think a 90 speed 50 man coverage db or lb should be any good at coverage . Perhaps recovering , but that's impossible if the stutter leaves them 15 yards behind. 5 maybe a person different story.

Re: Version 0.4.3 Release Candidate Discussion

By raymattison21
5/21/2018 3:27 pm
jdavidbakr wrote:
mardn72 wrote:
Broke down film for my first two games. It still feels like the stutter step is too powerful. In particular, I watched this WR embarrass people all game, despite having very average ratings. https://mfn1.myfootballnow.com/player/13269


I'll turn down the stutter impact a tad more. I'm trying to make small tweaks to it, so let me know if it continues to be too strong.

mardn72 wrote:
The other thing I'm concerned with now, and it's a feeling that's been growing for a number of games now, it the logic that causes QBs to have bad passes. The ones that are flagged "slipped out of his hands", "threw it behind the receiver", etc. Even elite QBs seems to have that trigger way too much, and in big bunches it seems.

I wonder if we need to make poor-to-average WR ratings behave worse, while letting high QB skills have even more of a positive impact.


Some of those color comments are because of the WR ratings and not the QB ratings; also the longer the pass, the less likely it will be caught, and some of the color is to help those apparent 'drops' to have more of a reason. (A long pass is more likely to be just out of reach, even for an elite QB, for example) I'm really just trying to give an idea of which factor in the probability calculation seems to be having the greatest impact, but there are a lot of variables that goes into the calculation and it's only picking one. I'm fine taking it out or reducing it if it's too verbose or too confusing.


Talking about wr ratings and qb out put My defenders gave up alot of catch overs to wrs who's route wasn't good at all. One guy had 40 and went for 100 yards most of them catch overs and alot vs. double coverages .

This makes me think the verbage is not using the wrs route skill very much. Yes, they have good hands and qb skill, but they were clearly covered tight all game , but it seemed to make a little difference whether it was a catch or not.

Later in the game stuff seemed to balance out. Either due to pressure or fatigue factors but the beginning was catch over after catch over. I chalked it up to RNG , but this post has me thinking other wise.

Re: Version 0.4.3 Release Candidate Discussion

By mardn72
5/21/2018 3:40 pm
jdavidbakr wrote:
mardn72 wrote:
The other thing I'm concerned with now, and it's a feeling that's been growing for a number of games now, it the logic that causes QBs to have bad passes. The ones that are flagged "slipped out of his hands", "threw it behind the receiver", etc. Even elite QBs seems to have that trigger way too much, and in big bunches it seems.

I wonder if we need to make poor-to-average WR ratings behave worse, while letting high QB skills have even more of a positive impact.


Some of those color comments are because of the WR ratings and not the QB ratings; also the longer the pass, the less likely it will be caught, and some of the color is to help those apparent 'drops' to have more of a reason. (A long pass is more likely to be just out of reach, even for an elite QB, for example) I'm really just trying to give an idea of which factor in the probability calculation seems to be having the greatest impact, but there are a lot of variables that goes into the calculation and it's only picking one. I'm fine taking it out or reducing it if it's too verbose or too confusing.


Ok, it's good to know more of the logic behind those calls. In looking closer, Baltimore has a pretty insanely talented set of DBs, so that may have added to the feel of my QB being bad. I'd say that's a good thing! Insane talent should have a visible impact.

In general, I'm feeling like things are getting solid. Maybe not perfectly tuned, but I haven't felt game-breaking holes yet this season. It'd be interesting to see if anyone can really exploit things the way they are now.

Re: Version 0.4.3 Release Candidate Discussion

By mardn72
5/21/2018 3:45 pm
raymattison21 wrote:

Talking about wr ratings and qb out put My defenders gave up alot of catch overs to wrs who's route wasn't good at all. One guy had 40 and went for 100 yards most of them catch overs and alot vs. double coverages .

This makes me think the verbage is not using the wrs route skill very much. Yes, they have good hands and qb skill, but they were clearly covered tight all game , but it seemed to make a little difference whether it was a catch or not.

Later in the game stuff seemed to balance out. Either due to pressure or fatigue factors but the beginning was catch over after catch over. I chalked it up to RNG , but this post has me thinking other wise.


I don't mind WR with low route making plays, as long as they are the right kind of plays. Route should allow a WR to stutter step well or make good cuts to create separation. If you're seeing low route WR making tough catches in tight coverage, that's fine if they have good hands and courage. If you're seeing WR with low route wide open from stuttering/cutting, that's an issue to me.

Re: Version 0.4.3 Release Candidate Discussion

By raymattison21
5/21/2018 4:40 pm
mardn72 wrote:
raymattison21 wrote:

Talking about wr ratings and qb out put My defenders gave up alot of catch overs to wrs who's route wasn't good at all. One guy had 40 and went for 100 yards most of them catch overs and alot vs. double coverages .

This makes me think the verbage is not using the wrs route skill very much. Yes, they have good hands and qb skill, but they were clearly covered tight all game , but it seemed to make a little difference whether it was a catch or not.

Later in the game stuff seemed to balance out. Either due to pressure or fatigue factors but the beginning was catch over after catch over. I chalked it up to RNG , but this post has me thinking other wise.


I don't mind WR with low route making plays, as long as they are the right kind of plays. Route should allow a WR to stutter step well or make good cuts to create separation. If you're seeing low route WR making tough catches in tight coverage, that's fine if they have good hands and courage. If you're seeing WR with low route wide open from stuttering/cutting, that's an issue to me.


No doubt thats why I didn't link any plays as I am in agreement about catch and punish, but separation the ones we can't see is what I am talking about. You have to remember the shoulder pads we see are the size of a real players wing span. So if a guy is right next to him he's really 3 to 6 feet away. 6 feet I am in agreement but 3 or less?

I see my guys are on top of them , within three feet but cannot at least knock it down ? And noting ratings yes thier catch was better with equivalent or less punish and definitely less man and zone comparison to route.

I am assuming here but catch could be a bit overpowered on shorter to meduim routes. My backs have little to not catch but still make plenty of catch overs themselves . If that wasn't happening I wouldn't say anything .

But Lbs are getting torched by TES as well and I am not seeing much of a difference in elite ones aND average ones(TEs). As I think a similar effect is happening there as well. As TES are heavier and slower making it harder for them to get separation without a stutter .

Also, i Saw a few wide open drops by elite wrs on long routes. Seems catch is devalued there to fit our changes to deep passing and turned up for shorter and medium routes to fit there. My response here are to make that part of the game visually better along with stats making more sense .

Re: Version 0.4.3 Release Candidate Discussion

By raymattison21
5/21/2018 5:27 pm
raymattison21 wrote:
mardn72 wrote:
raymattison21 wrote:

Talking about wr ratings and qb out put My defenders gave up alot of catch overs to wrs who's route wasn't good at all. One guy had 40 and went for 100 yards most of them catch overs and alot vs. double coverages .

This makes me think the verbage is not using the wrs route skill very much. Yes, they have good hands and qb skill, but they were clearly covered tight all game , but it seemed to make a little difference whether it was a catch or not.

Later in the game stuff seemed to balance out. Either due to pressure or fatigue factors but the beginning was catch over after catch over. I chalked it up to RNG , but this post has me thinking other wise.


I don't mind WR with low route making plays, as long as they are the right kind of plays. Route should allow a WR to stutter step well or make good cuts to create separation. If you're seeing low route WR making tough catches in tight coverage, that's fine if they have good hands and courage. If you're seeing WR with low route wide open from stuttering/cutting, that's an issue to me.


No doubt thats why I didn't link any plays as I am in agreement about catch and punish, but separation the ones we can't see is what I am talking about. You have to remember the shoulder pads we see are the size of a real players wing span. So if a guy is right next to him he's really 3 to 6 feet away. 6 feet I am in agreement but 3 or less?

I see my guys are on top of them , within three feet but cannot at least knock it down ? And noting ratings yes thier catch was better with equivalent or less punish and definitely less man and zone comparison to route.

I am assuming here but catch could be a bit overpowered on shorter to meduim routes. My backs have little to not catch but still make plenty of catch overs themselves . If that wasn't happening I wouldn't say anything .

But Lbs are getting torched by TES as well and I am not seeing much of a difference in elite ones aND average ones(TEs). As I think a similar effect is happening there as well. As TES are heavier and slower making it harder for them to get separation without a stutter .

Also, i Saw a few wide open drops by elite wrs on long routes. Seems catch is devalued there to fit our changes to deep passing and turned up for shorter and medium routes to fit there. My response here are to make that part of the game visually better along with stats making more sense .



Could be the reason ints are up on short routes as well

Re: Version 0.4.3 Release Candidate Discussion

By setherick
5/22/2018 8:01 pm
I can't stand breaking down more film than this. Here's what I think is going on: QBs complete way too few passes UNLESS the WR stutters and breaks the play. Otherwise, QBs throw into double coverage WAY too much and have STOPPED checking down or completing their checkdowns. Basically, all the good things about 0.4.3 up until say April 1 are not there, and what's left is all the broken garbage that had been fixed (except for stutter which was always broken).

1) https://beta87.myfootballnow.com/watch/1791#326382
a) There is no reason for the QB to not complete this pass or the RB to not catch the ball here. If I hear "play knowledge" as an answer, I may throw my laptop off the balcony.

2) https://beta87.myfootballnow.com/watch/1791#326387
a) Howell is a stud, but I don't see why the QB makes this pass anyway. At the point of the interception, both outside WRs are breaking free. Monroe has 97 FOV, 100 Look Off, and 100 Accuracy to boot.

3) https://beta87.myfootballnow.com/watch/1791#326396
a) The WR cuts his route too shallow here, and almost causes a bad play. He has 100 route.

4) https://beta87.myfootballnow.com/watch/1791#326397
a) Two WIDE open WRs on this play. Elite players gotta Elite sometimes.

5) https://beta87.myfootballnow.com/watch/1791#326401
a) Stutter throws off the QB here. He throws it way behind the receiver because of how the receiver breaks after the stutter. The QB is throwing the ball to the spot the WR would have been if he had run straight after the stutter instead of cut right. Not sure if the first stutter or the second stutter causes this bug.

6) https://beta87.myfootballnow.com/watch/1791#326403
a) The WR running the deep cross is WIDE open. Elite players gotta Elite.

7) https://beta87.myfootballnow.com/watch/1791#326404
a) Outside top WR is WIDE open. RB is WIDE open. QB throws it into double coverage. Elite players gotta Elite.

8) https://beta87.myfootballnow.com/watch/1791#326405
a) This play shows why stutter should be pulled completely from the 0.4.3 release. The DB that falls down here has 71 functional coverage. Sure he's playing a 100 Route WR, but he's nowhere near the play when the WR catches the ball. This will be all over the place in standard leagues since there are NOT ENOUGH 100 M2M cover DBs and WAY TOO MANY 100 Route WRs.

9) https://beta87.myfootballnow.com/watch/1791#326406
a) Two WRs open IN THE MIDDLE OF THE FIELD. QB throws into double coverage on the outside.

There are too many unaccountable for factors right now to make this even enjoyable to test. When you can't match player play to attributes, what's the point in playing at all?
Last edited at 5/22/2018 8:03 pm

Re: Version 0.4.3 Release Candidate Discussion

By Ragnulf-le-maudit
5/23/2018 3:52 am
I see the same things than Setherick, and I have to say it's sometimes frustrating. At some point, i think the passing game was nice. Now, it's not and it's difficult to undersrtand what went wrong.

No WR near the pass. First time i see this.
https://mfn1.myfootballnow.com/watch/8101#1469426
Same
https://mfn1.myfootballnow.com/watch/8101#1469536


Last edited at 5/23/2018 4:12 am

Re: Version 0.4.3 Release Candidate Discussion

By setherick
5/24/2018 6:22 am
This is just from the first half of my 87 game. Passing is worse than it was in 0.3.8 - the gunslinger years.

But it is also easily fixed:

* The QB FOV algorithms are obviously wrong or player attributes are way overvalued. 80-100 FOV QBs throw into double coverage almost every play and miss wide open WRs. This suggests that there is a logical error with that code.

* QBs throw "bad" passes to wide open WRs when they can find them. This suggests that either there is a mathematical problem with determining accuracy when the WR is wide open. Or there is an unnecessary limitation being placed on QBs in order to prevent big gains.

* Coverage still sucks. The first two bullet points above suggest that the root cause of all of the QB madness has not been addressed, and instead, the only position looked at in the passing equations was the QB. More below.

* DBs still stop covering their man. Remember two months ago when I said that 0.4.3 was one problem away from release? And that problem was making sure that DBs covered their man. Well, it's still not happening and now QB decision making code is awful. Let's revert the code back to a pre-April 1 state and start over.

* Stutter is still breaking plays badly despite the DBs M2M score, and the only reason it isn't more apparent are all the unnecessary nerfs being placed on QBs right now. The stutter problem is related to the DBs stop covering their man problem. But, really, I don't see why solving the simple problem of a DB covering a WR once the WR has stopped running a route caused the code to end up in the state that it is in.

1) https://beta87.myfootballnow.com/watch/1804#328717
a) Just another 100 FOV QB throwing into double coverage while everyone else is man up and the outside WR on the bottom of the screen has a step on his man.

2) https://beta87.myfootballnow.com/watch/1804#328719
a) Another STUPID throw into double coverage while a number of players are WIDE open. If the QB throws the the right slot this is a TD.

3) https://beta87.myfootballnow.com/watch/1804#328725
a) Hey, a checkdown... into double coverage... while the outside WR is starting to break free.

4) https://beta87.myfootballnow.com/watch/1804#328726
a) Let's throw to the RB INTO DOUBLE COVERAGE on this play like it's v.0.3.8 while the outside WR breaks the jam and sprints up field untouched. It's cool. Passing sucks.

5) https://beta87.myfootballnow.com/watch/1804#328731
a) At least the QB doesn't throw into double coverage here, but he throws to the worse possible receiver given the play. The QB needs 15 yards for the first down. When he makes this throw, both outside WRs are beyond the marker and the TE is just starting to break free. A QB on a good high school team is going to throw to one of those three players. A pro QB is going to recognize the TE has broken coverage and drop the ball over the top for a long gain. But it's cool for MFN QBs to just randomly throw short into coverage. It's more work for the punters.

6) https://beta87.myfootballnow.com/watch/1804#328733
a) Slot WR breaking free to the outside right at the first down marker or RB who is tightly covered with two other LBs right behind the coverage? Oh, gotta throw to the RB here.

7) https://beta87.myfootballnow.com/watch/1804#328735
a) Throw it away? Take a sack? No! Let's throw into triple coverage because that's the best plan here.

8) https://beta87.myfootballnow.com/watch/1804#328737
a) So Glenn makes a **** of a move and scores a TD, but he is still the worst receiver to throw this pass to. The TE AND the slot WR are WIDE OPEN and could have walked into the end zone. Good play by the RB. Stupid throw.

9) https://beta87.myfootballnow.com/watch/1804#328741
a) Checkdown to the RB is wide open for a nice gain. But let's panic for no reason and throw this into double coverage knowing there is a trail guy behind that coverage.

10) https://beta87.myfootballnow.com/watch/1804#328745
a) This isn't that bad, but the slot WR is WIDE open when the QB throws. Why even have a FOV attribute?

11) https://beta87.myfootballnow.com/watch/1804#328749
a) Just watch this one. Really, just watch it. My 2nd grade flag QB made better decisions.

12) https://beta87.myfootballnow.com/watch/1804#328750
a) Hmmm...the outside WR on the bottom of the screen is breaking free, so let's throw into double coverage and hope for an interception. Yes!

13) https://beta87.myfootballnow.com/watch/1804#328756
a) TE is running a wide open seam route and is past the first down marker on this 3 and 9 play, so let's THROW THE BALL SHORT into tight coverage so we can kick the FG. Stupid.

14) https://beta87.myfootballnow.com/watch/1804#328758
a) Slot WR and TE are WIDE OPEN for a TD, so let's throw the ball to the outside.

15) https://beta87.myfootballnow.com/watch/1804#328762
a) Why is this a "diving catch"? Why isn't a TD?

16) https://beta87.myfootballnow.com/watch/1804#328763
a) The slot WR is WIDE open. The RB checkdown is WIDE open. Let's throw into double coverage.

17) https://beta87.myfootballnow.com/watch/1804#328767
a) QB throws to the only WR that is double covered on this play.

18) https://beta87.myfootballnow.com/watch/1804#328772
a) When the QB throws, the TE is past the marker and WIDE OPEN.

19) https://beta87.myfootballnow.com/watch/1804#328774
a) The proble with this OPI is who the eff does he interfer with?

20) https://beta87.myfootballnow.com/watch/1804#328775
a) When this ball is knocked down, both CBs are 8 and 4 yards away from the play. This can't be a knockdown.

21) https://beta87.myfootballnow.com/watch/1804#328781
a) The WR is WIDE OPEN and the QB throws it there. So of course he throws it BADLY and the WR has to stop.

22) https://beta87.myfootballnow.com/watch/1804#328784
a) The RB is open here, but so is the slot WR who has the biggest gain potential.

23) https://beta87.myfootballnow.com/watch/1804#328792
a) Let's talk about this one for a second because it shows a lot of the bad things with the passing game right now:
i) Stutter straight up breaks the play from the outset.
ii) QB is not under pressure, but throws a bad pass to the WIDE OPEN WR who has to adjust to make the catch.
iii) These two things show how you are using unnecessary code to limit the QB ability to throw the ball rather than actually fixing coverage. It's pretty awful.

Re: Version 0.4.3 Release Candidate Discussion

By jdavidbakr - Site Admin
5/24/2018 8:23 pm
setherick wrote:
* The QB FOV algorithms are obviously wrong or player attributes are way overvalued. 80-100 FOV QBs throw into double coverage almost every play and miss wide open WRs. This suggests that there is a logical error with that code.


One point I do want to make here is that the QB's FOV is only one factor in him choosing an alternate receiver than his initial target. Another huge part is his experience. The algorithm first decides if he is even going to consider another WR; more experienced QBs will choose this more often. Then, and only then, does the FOV impact what WRs he considers. Yes, 100 FOV will give him basically a 170 degree range of view, but if he's a rookie he's unlikely to take advantage of that.

That's not the only purpose of FOV, another usage is deciding whether he sees a DB in coverage.

setherick wrote:
* QBs throw "bad" passes to wide open WRs when they can find them. This suggests that either there is a mathematical problem with determining accuracy when the WR is wide open. Or there is an unnecessary limitation being placed on QBs in order to prevent big gains.


I'm not going to argue this point, as there is an artificial impact to reduce the completion rate especially of longer passes. I've been closely monitoring the completion percentage and interception percentage as well as the yards per attempt and have been working to get them close to NFL averages. I want a pass heavy game to be able to do very well but just as likely bite you - live by the sword, die by the sword. Do I feel like this is the way it should be forever? No, I agree with you that I'd like the coverage to impact it better; but, this is an incredibly complex system and I do think we need to keep plugging away at it in small pieces to get it better and better. Overall, I do feel like the changes we have here currently have improved the passing game.

setherick wrote:
* DBs still stop covering their man. Remember two months ago when I said that 0.4.3 was one problem away from release? And that problem was making sure that DBs covered their man. Well, it's still not happening and now QB decision making code is awful. Let's revert the code back to a pre-April 1 state and start over.


I haven't changed anything about the QB decision making code, in fact really since April 1 I've tweaked the stutter impact and worked on normalizing the completion rates, and that's about it.

setherick wrote:
* Stutter is still breaking plays badly despite the DBs M2M score, and the only reason it isn't more apparent are all the unnecessary nerfs being placed on QBs right now. The stutter problem is related to the DBs stop covering their man problem. But, really, I don't see why solving the simple problem of a DB covering a WR once the WR has stopped running a route caused the code to end up in the state that it is in.


I'm pushing another tweak to the stutter to see if I can keep the DB from getting beat so badly.

setherick wrote:
1) https://beta87.myfootballnow.com/watch/1804#328717
a) Just another 100 FOV QB throwing into double coverage while everyone else is man up and the outside WR on the bottom of the screen has a step on his man.


I disagree. Both outside receivers are blanketed by their DBs at the moment the pass is thrown. The passed to target has some separation from both his coverage players (although I won't argue it was a great decision to make that throw)

setherick wrote:
2) https://beta87.myfootballnow.com/watch/1804#328719
a) Another STUPID throw into double coverage while a number of players are WIDE open. If the QB throws the the right slot this is a TD.


Also a rookie QB locking on to his #1 target.

setherick wrote:
3) https://beta87.myfootballnow.com/watch/1804#328725
a) Hey, a checkdown... into double coverage... while the outside WR is starting to break free.


Actually that's his primary target (you can tell by the fact that he's following him from the snap). I'm not going to defend that point, but it's not a check-down.

setherick wrote:
4) https://beta87.myfootballnow.com/watch/1804#328726
a) Let's throw to the RB INTO DOUBLE COVERAGE on this play like it's v.0.3.8 while the outside WR breaks the jam and sprints up field untouched. It's cool. Passing sucks.


Locking on to his primary target here. Should he have switched to the receiver on the bottom running the deep route? Probably; clearly the logic fired to decide that he wasn't going to look at another receiver.

setherick wrote:
5) https://beta87.myfootballnow.com/watch/1804#328731
a) At least the QB doesn't throw into double coverage here, but he throws to the worse possible receiver given the play. The QB needs 15 yards for the first down. When he makes this throw, both outside WRs are beyond the marker and the TE is just starting to break free. A QB on a good high school team is going to throw to one of those three players. A pro QB is going to recognize the TE has broken coverage and drop the ball over the top for a long gain. But it's cool for MFN QBs to just randomly throw short into coverage. It's more work for the punters.


You're also dealing again with a rookie QB who probably was locking on to his #1 target.

I'm not going to comment on the other plays, although I do appreciate the feedback and am looking at them, but it's a lot of the same. There are a couple of points I want to make, though:

1) The sim is not perfect, and I do feel like you are getting frustrated at points that are several versions away from solving.
2) I think you are also getting frustrated at logic where a player is making the bad decision because, well, he's a simulated human and part of the simulation is to make bad decisions. Even Tom Brady makes bad decisions. If you want to just have the best skills win every time then why bother to play the game?
3) I'm trying to keep the changes in scope, and have this 0.4.3 be an iteration better than 0.4.2. It's not going to be perfect, and I'm not trying to make it perfect, especially at this stage in development.

Let's fix what is making 0.4.3 worse than 0.4.2, and get on to 0.4.4 to fix what needs to be fixed next.