New Sticky Note

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 setherick
6/01/2018 10:33 pm
mardn72 wrote:
Hey Setherick, I'm not trying to be argumentative, but I'm just not seeing the games that same way you are. I've watched every one of my games this season and I see a nice balance between offensive and defensive playmaking, my QB making dumpoff passes regularly, and my QB hitting 2nd and 3rd targets regularly. Maybe would you watch my latest game and see what you see happening? https://mfn1.myfootballnow.com/log/8153 I know you don't always have a lot of free time, so no worries if you don't want to spend what time have breaking down my film.


OK, did it. Seeing the same garbage in your games as I do in mine. Bad coverage being masked my inept QB play.

One drive in the game before the latest code change.

1) https://mfn1.myfootballnow.com/watch/8140#1476489
a) Bad B&R means that the WR toast BOTH PLAYERS THAT ARE SUPPOSED TO DOUBLE HIM.

2) https://mfn1.myfootballnow.com/watch/8140#1476490
a) QB locks on to the RB and never looks at the WIDE open TE.

3) https://mfn1.myfootballnow.com/watch/8140#1476492
a) Wide open pass gets dropped despite it being a 10 yard throw at most.

4) https://mfn1.myfootballnow.com/watch/8140#1476493
a) QB throws into double coverage despite the RB being WIDE open on the back side of the throw.

5) https://mfn1.myfootballnow.com/watch/8140#1476494
a) QB ignores the wide open dump off pass.

4 and 5 would have both gone to the RB in the pre-4/4 0.4.3 code. Now the RB isn't even considered.

Most recent game against Ray after the last code change.

1) https://mfn1.myfootballnow.com/watch/8153#1478818
a) QB locks on to the RB and makes a bad pass regardless.

2) https://mfn1.myfootballnow.com/watch/8153#1478823
a) Slot WR is wide open, but the QB throws a BAD check down pass to the RB. Both slot and RB are running short routes, so this isn't a route thing.

3) https://mfn1.myfootballnow.com/watch/8153#1478826
a) Yet another bad pass to the RB.

4) https://mfn1.myfootballnow.com/watch/8153#1478830
a) Awful B&R means the WR is wide open, but the QB still makes an bad pass on a ball that travels <20 yards through the air.

5) https://mfn1.myfootballnow.com/watch/8153#1478834
a) Oh, look at that, another bad pass to a RB.

6) https://mfn1.myfootballnow.com/watch/8153#1478835
a) This garbage should have been a pick six.

7) https://mfn1.myfootballnow.com/watch/8153#1478837
a) AND THEN THERE IS THIS RIDICULOUS BS. This is supposed to be a throwaway, but it ends up 20+ yards downfield. The WR stutters to break the coverage on the play and then runs under the throw away to catch it.

8) https://mfn1.myfootballnow.com/watch/8153#1478839
a) OMG another bad pass to a wide open TE. This should have been a TD.

9) https://mfn1.myfootballnow.com/watch/8153#1478840
a) But not to outdue ourselves let's throw an equally bad pass on the next play.

10) https://mfn1.myfootballnow.com/watch/8153#1478841
a) But on the next play, let's go back to ignoring the WIDE open RB and throw into double coverage because that's more fun.

11) https://mfn1.myfootballnow.com/watch/8153#1478844
a) Ray's QB was missing the fun of throwing into double coverage. Instead of, you know, running down field in the WIDE OPEN RUNNING LANE FOR AT LEAST 15 YARDS that he has.

12) https://mfn1.myfootballnow.com/watch/8153#1478845
a) Stutter breaks the play, but it's not in the PbP. Amazingly the QB actually throws a good pass here.

13) https://mfn1.myfootballnow.com/watch/8153#1478846
a) Another "throwaway" that ends up downfield and should be picked off.

14) https://mfn1.myfootballnow.com/watch/8153#1478855
a) It really looks like the QB is throwing to the defender here.
Last edited at 6/01/2018 10:33 pm

Re: Version 0.4.3 Release Candidate Discussion

By setherick
6/02/2018 7:02 am
I barely beat a CPU team in 87, but I should have lost the game.

1) https://beta87.myfootballnow.com/watch/1871#340841
a) WR stutters at the 30 yard line which means the 100 M2M CB stops covering on this play. The QB of course throws an awful pass because it's ANY yards downfield. It doesn't matter that the QB has 99 ARM STRENGTH, which should put him in the same class as Big Ben in rocketting passes downfield. Or that the WR has 92 catch. All the stupid QB nerfs make sure that this pass is incomplete. The problem, of course here, is that you can CLEARLY SEE THE ACTUAL PROBLEM IN THE CODE THAT STILL HASN'T BEEN SOLVED. That's why this code is such garbage. It has an obvious bug that should be fixed. That is being IGNORED in order for a series of increasingly illogical bandaids.

But wait there's more.

2) https://beta87.myfootballnow.com/watch/1871#340917
a) What happens on this play? I thought maybe I was in run key so the LB looked inside first, but I'm not. I don't know if this is failed B&R or a stutter. All I know is the LB has better M2M than the WR has Route. The LB has better B&R than the WR has Avoid. The LB also has the same SP and better AC. But the LB is NOWHERE on this play. Let's add another bandaid for this scenario.

But wait there's more.

3) https://beta87.myfootballnow.com/watch/1871#340962
a) Desperate drive by my CPU opponent cut short because his QB panics and throws a pass into the dirt for no reason.

Next Play

4) https://beta87.myfootballnow.com/watch/1871#340963
a) QB rolls away from the play so that he makes sure to get enough distance to botch this pass to a RB like all passes to RBs.
Last edited at 6/02/2018 7:54 am

Re: Version 0.4.3 Release Candidate Discussion

By raymattison21
6/02/2018 7:07 am
setherick wrote:
You're right. What you did in the second game was abuse the **** out of a few plays, but for the most part, all the plays that you called in the second game (except for the I Form 212 plays) were the same from the first game. And the strategy was the same.

Also, the most points I had allowed that season prior to that crapshow was 29. The simple fact of the matter was that coverage worked if you had good M2M and B&R guys. Plays just broke when WRs stuttered.

But...eff it...let's just release 0.4.3 tomorrow. I'm sure that no one will be mad.


Those were completely different plans. In the second game if we called anything besides 212 plays it was a glitch , and probably still is now. Those other plays were out of my matrix . I had to omit them completely in order for them not be be called. As far a man being broke. It always was and still is now. I could care less if a release happens. It's been forever now cause too much was put on the table I the first place. It took two seasons to scrap obfuscation but we're running this test for seasons now with nothing concrete imo. Still, it is better than 4.2

And the only people that will be mad if it is released are the ones that abuse the long pass. Which is every champ in every league. At least they still can blitz over 75% of the time and still be effective .

One big difference that really distorts rating match ups is qb ability . I don't think and 60 rated default qbs will be winning championships under this code. Just get an elite qb already .

Re: Version 0.4.3 Release Candidate Discussion

By setherick
6/02/2018 7:29 am
raymattison21 wrote:
One big difference that really distorts rating match ups is qb ability . I don't think and 60 rated default qbs will be winning championships under this code. Just get an elite qb already .


I'm sorry Ray, but this shows that you haven't actually been watching the film closely. I don't care about QB ratings - and I don't play a 60 overall QB in any league for that matter.

But since you mentioned it, I do care about the fact that QB attributes DO NOT MATCH game play.

Your QB has 90-100 Arm? Congratulations, he'll throw a wounded duck on ANY pass that goes over 10 yards.

Your QB has 100 FOV? Congratulations, he'll throw into double coverage on almost every pass while ignoring wide open WRs.

Your QB has 100 Accuracy? Congratulations, he'll never cleanly complete a pass that is < 5 yards and will throw almost every RB pass in the dirt.

Your DB has 100 M2M? Congratulations, he'll have effectively 0 M2M when a WR stutters.

Those are all code problems. NOT PERSONNEL PROBLEMS. And those are consistent problems across all of the games right now.

---

Went back to look at your game plan from last season: https://mfn1.myfootballnow.com/log/7965

Almost all of the TDs in this game came against zones, and many against the Short zones, which is an obvious exploit. The three TDs in this game against M2M all involved stutter:

1) https://mfn1.myfootballnow.com/watch/7965#1445069
2) https://mfn1.myfootballnow.com/watch/7965#1445087
3) https://mfn1.myfootballnow.com/watch/7965#1445103

I refuse to go through any more of your pass long games, but I'm willing to bet right now, that 90% of your TDs against M2M out of Long passes came from the stutter bug.
Last edited at 6/02/2018 7:38 am

Re: Version 0.4.3 Release Candidate Discussion

By raymattison21
6/02/2018 8:58 am
setherick wrote:
raymattison21 wrote:
One big difference that really distorts rating match ups is qb ability . I don't think and 60 rated default qbs will be winning championships under this code. Just get an elite qb already .


I'm sorry Ray, but this shows that you haven't actually been watching the film closely. I don't care about QB ratings - and I don't play a 60 overall QB in any league for that matter.

watching close, but stating the general as some guys are performing. 4.2 has risen that bar closer to 80, but that doesn't mean an 80 should be a 90 or higher in terms of playing like one

But since you mentioned it, I do care about the fact that QB attributes DO NOT MATCH game play.
it is a bit deeper that that. Try playing a bad qb and being successful . It won't work, so in my mind something correct is going on

Your QB has 90-100 Arm? Congratulations, he'll throw a wounded duck on ANY pass that goes over 10 yards.
alot of variables are at play, but Suttle nuances should fix it

Your QB has 100 FOV? Congratulations, he'll throw into double coverage on almost every pass while ignoring wide open WRs.
I don't see this with all qbs all the time.

Your QB has 100 Accuracy? Congratulations, he'll never cleanly complete a pass that is < 5 yards and will throw almost every RB pass in the dirt.
my 97 accuracy qb can't compete 40 % of passes , but his default is 54. So that's a wash as a comparison . Perhaps he should join the Professional Dart Championship and retire from football

Your DB has 100 M2M? Congratulations, he'll have effectively 0 M2M when a WR stutters.
that is steep, but generally we are in agreement here.

Those are all code problems. NOT PERSONNEL PROBLEMS. And those are consistent problems across all of the games right now.
if anyone takes anything personal here. They have to remember it's a virtual world. Anything I say is an opinion and should be taken so. My efforts are to be happy as should be everyones. Change doesn't accommodate that at all. Finding the balance between the two within the constraints of whats possible is the key to a successful release here.

---
to me the latter is mute. I was just talking about the plans installed not the outcomes. I just proved that code was broke


Went back to look at your game plan from last season: https://mfn1.myfootballnow.com/log/7965

Almost all of the TDs in this game came against zones, and many against the Short zones, which is an obvious exploit. The three TDs in this game against M2M all involved stutter:

1) https://mfn1.myfootballnow.com/watch/7965#1445069
2) https://mfn1.myfootballnow.com/watch/7965#1445087
3) https://mfn1.myfootballnow.com/watch/7965#1445103

I refuse to go through any more of your pass long games, but I'm willing to bet right now, that 90% of your TDs against M2M out of Long passes came from the stutter bug.

Re: Version 0.4.3 Release Candidate Discussion

By jdavidbakr - Site Admin
6/02/2018 10:13 am
Making a couple of changes to address the commentary above:

* I'm changing the logic that determines whether a DB will bite on the stutter to hopefully significantly reduce its occurrence for high M2M players.
* I'm making a tweak to the pressure/distance accuracy algorithm to have a probability based on the QB's attributes of whether those are applied against a blitz. My goal with this adjustment is to push the blitz more toward feast or famine, instead of all feast, in a more realistic way than the hot read did.

I'm not going to change the QB scramble logic in this version, but that might be something to look at for the next version. I just wanted to address that I did see that comment and agree that it needs to be dealt with.

A couple of comments to the 'pre 4/1' references, the code that was close to release was the code where I think it was Ray that had a huge long pass game. I refuse to knowingly release this next version with code that will produce 900-yard passing games and 90+ scores to owners who have a dedicated long passing game. Also, the issue with the QB not checking to other receivers was not introduced after 4/1, in fact it exists in 0.4.2 and just is not likely getting hit as often because of other progressive changes in the logic elsewhere. And, you are putting too much weight into the distance penalty. It hasn't even applied to passes that were less than 10 yards downfield over the last few days, and the curve doesn't kick in significantly until it hits the 20-yard mark, so the fact that players are dropping short passes is most likely a result of other causes. Also, a couple of your examples of the QB not looking at other receivers, I have to say, kind of make me wonder what kind of a FOV you expect a QB to have that he can see out of the air holes in the back of his helmet. The only way he's going to check down is if he decides not to throw to the receiver he's looking at or anyone else in his FOV at that moment, and a spot check of a few of your examples shows a range of less than 35% probability of him deciding to throw, which means 2 times out of 3 he's going to check down to another WR in that exact same moment.
Last edited at 6/02/2018 10:20 am

Re: Version 0.4.3 Release Candidate Discussion

By setherick
6/02/2018 12:32 pm
jdavidbakr wrote:
Also, a couple of your examples of the QB not looking at other receivers, I have to say, kind of make me wonder what kind of a FOV you expect a QB to have that he can see out of the air holes in the back of his helmet. The only way he's going to check down is if he decides not to throw to the receiver he's looking at or anyone else in his FOV at that moment, and a spot check of a few of your examples shows a range of less than 35% probability of him deciding to throw, which means 2 times out of 3 he's going to check down to another WR in that exact same moment.


That's a fair point. On the examples from Ray and Mardn's game that I pulled out. The problem is more that the QB never looks at any other receiver. The QB stays locked on the first read throughout the entire play. If the QB quickly skims the play, then he should see the rest of the receivers. If there are no other receivers in the FOV, then it's the only receiver that he will throw to. Here's a good example: https://mfn1.myfootballnow.com/watch/8140#1476494

The RB is beyond 170 degrees from the QB, but the reason that the RB is beyond the 180 degree mark is because the QB is squared with the WR instead of being squared with the field. QBs should be squared with the field 99% of the time (I mean come on, we coach that in junior high), but they aren't. Instead, they square with their first read and limit their FOV unnecessarily.

The exact same problem is on the previous play: https://mfn1.myfootballnow.com/watch/8140#1476493

The QB takes the snap and immediately squares with his first read. This is particularly disastrous because there is one and only one read when he squares the field that way in his FOV. If the QB drops straight back and squares with the field, then all of the reads would have been within a 170 degree FOV at the time the ball is thrown.

So while you are perfectly right that the QB can't see someone that is behind him. If he were squared correctly, he'd see the player anyway. I should have taken the time to explain this before.

Re: Version 0.4.3 Release Candidate Discussion

By jdavidbakr - Site Admin
6/02/2018 12:42 pm
setherick wrote:
jdavidbakr wrote:
Also, a couple of your examples of the QB not looking at other receivers, I have to say, kind of make me wonder what kind of a FOV you expect a QB to have that he can see out of the air holes in the back of his helmet. The only way he's going to check down is if he decides not to throw to the receiver he's looking at or anyone else in his FOV at that moment, and a spot check of a few of your examples shows a range of less than 35% probability of him deciding to throw, which means 2 times out of 3 he's going to check down to another WR in that exact same moment.


That's a fair point. On the examples from Ray and Mardn's game that I pulled out. The problem is more that the QB never looks at any other receiver. The QB stays locked on the first read throughout the entire play. If the QB quickly skims the play, then he should see the rest of the receivers. If there are no other receivers in the FOV, then it's the only receiver that he will throw to. Here's a good example: https://mfn1.myfootballnow.com/watch/8140#1476494

The RB is beyond 170 degrees from the QB, but the reason that the RB is beyond the 180 degree mark is because the QB is squared with the WR instead of being squared with the field. QBs should be squared with the field 99% of the time (I mean come on, we coach that in junior high), but they aren't. Instead, they square with their first read and limit their FOV unnecessarily.

The exact same problem is on the previous play: https://mfn1.myfootballnow.com/watch/8140#1476493

The QB takes the snap and immediately squares with his first read. This is particularly disastrous because there is one and only one read when he squares the field that way in his FOV. If the QB drops straight back and squares with the field, then all of the reads would have been within a 170 degree FOV at the time the ball is thrown.

So while you are perfectly right that the QB can't see someone that is behind him. If he were squared correctly, he'd see the player anyway. I should have taken the time to explain this before.


I see your point. I'm going to call that out of scope for this version, though.

Re: Version 0.4.3 Release Candidate Discussion

By setherick
6/02/2018 12:55 pm
jdavidbakr wrote:

I see your point. I'm going to call that out of scope for this version, though.


I'm fine with that. I'll stop recording those until 0.4.4.

To be fair, I was only recording them because I though that QB squaring was a visual glitch.

EDIT

And now that this exchange has had the stupid mantra stuck in my head the entire time I mowed the front yard. Here's the mantra that you use to teach your beginning QBs (and really all positions).

It's faster to move your eyes than your head.
It's faster to move your head than your hips.
It's faster to move your hips than your feet.

Eyes, head, hips, square, step, throw....

Now everyone repeat it.
Last edited at 6/02/2018 1:52 pm

Re: Version 0.4.3 Release Candidate Discussion

By raymattison21
6/03/2018 10:18 am
jdavidbakr wrote:
Making a couple of changes to address the commentary above:

* I'm changing the logic that determines whether a DB will bite on the stutter to hopefully significantly reduce its occurrence for high M2M players.
* I'm making a tweak to the pressure/distance accuracy algorithm to have a probability based on the QB's attributes of whether those are applied against a blitz.

in specific what are the attributes we should look for in a qb who can beat the blitz ?

My goal with this adjustment is to push the blitz more toward feast or famine, instead of all feast, in a more realistic way than the hot read did.

I'm not going to change the QB scramble logic in this version, but that might be something to look at for the next version. I just wanted to address that I did see that comment and agree that it needs to be dealt with.

A couple of comments to the 'pre 4/1' references, the code that was close to release was the code where I think it was Ray that had a huge long pass game.

I refuse to knowingly release this next version with code that will produce 900-yard passing games and 90+ scores to owners who have a dedicated long passing game.

nobody want huge passing games, but 500 yards should be possible in the right situation


Also, the issue with the QB not checking to other receivers was not introduced after 4/1, in fact it exists in 0.4.2 and just is not likely getting hit as often because of other progressive changes in the logic elsewhere.

what is the check down logic? Perhaps it was explained later in this post? Does it have anything to do with the auto setting of throwing to backs and TEs instead wrs? Or is that different as I wish we could have a slider in the game planning section we could change game to game.

And, you are putting too much weight into the distance penalty. It hasn't even applied to passes that were less than 10 yards downfield over the last few days, and the curve doesn't kick in significantly until it hits the 20-yard mark, so the fact that players are dropping short passes is most likely a result of other causes.

Also, a couple of your examples of the QB not looking at other receivers, I have to say, kind of make me wonder what kind of a FOV you expect a QB to have that he can see out of the air holes in the back of his helmet.

an elite , experienced, max play familiar, look off QB like brady has checked down to rbs without having them in his FOV. Based on coverage looks and intial movement of the defense at the snap or soon after one could guess if a checkdown is open or not and make the play.

The only way he's going to check down is if he decides not to throw to the receiver he's looking at or anyone else in his FOV at that moment, and a spot check of a few of your examples shows a range of less than 35% probability of him deciding to throw, which means 2 times out of 3 he's going to check down to another WR in that exact same moment.


So my main questions are....

What the blitz beating attributes for qb?

Is there an auto setting which decides passes are throws to wrs (intial reads) or TEs and backs (check down reads ? )? I think there two logics going on here. What are they?