User Pass
Home Sign Up Contact Log In
Forum > Suggestions > Provide a Refund to the Nashville Team in USA AA#1
Page:
 
Ken1
offline
Link
 
Originally posted by SCACE
Thanks Ken, good convo. I know wrap up is getting nerfed, which is perfect...I still think it will be an overly effective way to tackle someone, but I hope the fumbles caused by it will become almost non-existant unless the levels/builds are so far out of whack a fumble should probably occur even on wrap up

Not sure if I recall seeing anything about the passing game getting tweaked, but if the overall offense gets a boost + the new AI, hopefully that will be enough to balance it out.


Yes, this has been very productive, at least as long as the people who need to see it do, and wade through the less productive parts to see the more productive ones.
 
SCACE
offline
Link
 
Originally posted by jdbolick
Originally posted by SCACE

While we do both seem to agree something is up here, we completely disagree on what the problem and what the solution is.

Actually you've repeatedly acknowledged that the cascade effect of confidence penalties is real and is a problem. Only you've said that you want ways to hopefully avoid that problem, whereas I'd just rather fix the problem directly. Adding the elements you describe wouldn't fix the underlying issue, they'd only reduce its frequency of occurring. While that would be a step in the right direction, it's a bit like using "Run Flat" on a tire puncture. It helps, but it's not fixing the issue. I would support the changes you propose, but something also needs to be done to prevent morale penalties from completely overwhelming every other determinant.


That's where we disagree. I only think the confidence/morale penalties are "out of whack" because the real life options to avoid them don't yet exist in the game. I believe that with these options, those types of results would almost completely disappear, and would probably only show up when people used overly risky settings, which isn't a bad thing, in my opinion.

I don't believe you need to nerf the negative effect of low confidence and its effect on morale, because I believe that will severely lessen the important of the attribute -- an attribute that people were already ignoring / neglecting -- and just allows them to continue on the same path, boosting other attributes at the expense of confidence.

Because of tweaks to the system based on complaints, not unlike yours, GLB has to go back and tweak tackling again, because complaints led from one extreme to another. Bort has to be careful when he implements new tweaks to season four and hopefully we'll see a better, more realistic and balanced game. But any tweaks now, in the games first run, could turn out to need a tweak in the other direction again in a season or two.

 
jdbolick
offline
Link
 
Originally posted by Ken1
I hope that Bort can come to understand-- if he hasn't already-- that ypa is the key stat. It's okay if completion percentage is "too high" if yards per completion (ypc) is "too low," the combo of which would still lead to a fairly realistic ypa.

Very insightful stuff, and it reminds me of an article by Dr. Z a couple of years back regarding the formula for QB rating being obsolete due to an over-reliance on completion percentage as opposed to YPA. Dr. Z was explaining why QB ratings have increased so dramatically over previous decades, with the answer being that teams are throwing more short, high percentage passes than in prior years.

 
SCACE
offline
Link
 
one more think bolick - there are currently no penalties in the game, yet if I'm remembering right, the SIM takes into account that there will be penalties. There are probably times where there is a pass defend or an INT where there should be a pass interference penalty. However, these don't exist yet, inflating INT numbers. Just something else to consider, as there are many real life options missing from the game right now, and if they were in the game, with confidence/morale as it stands, we would not see these outliers you aren't fond of, again, unless players/teams really muffed their tactics. In which case, those are the games where outliers should be reached.
 
SCACE
offline
Link
 
Originally posted by jdbolick
Originally posted by Ken1

I hope that Bort can come to understand-- if he hasn't already-- that ypa is the key stat. It's okay if completion percentage is "too high" if yards per completion (ypc) is "too low," the combo of which would still lead to a fairly realistic ypa.

Very insightful stuff, and it reminds me of an article by Dr. Z a couple of years back regarding the formula for QB rating being obsolete due to an over-reliance on completion percentage as opposed to YPA. Dr. Z was explaining why QB ratings have increased so dramatically over previous decades, with the answer being that teams are throwing more short, high percentage passes than in prior years.



Makes sense - I couldn't stand to watch Gannon run that Raiders offense a few years back...would much rather see the long pass, but obviously there has been a move to the safer options, avoid mistakes to try and sustain drives and win the time of possession / turnover battle. Not all coaches gameplan this way, but I've definitely seen more of it in my lifetime vs. watching archived footage.
 
jdbolick
offline
Link
 
Originally posted by SCACE
That's where we disagree.

Because you don't understand the mathematics. I'm not intending to disparage you when I say that. This same thing has happened to me countless times when fellow owners in fantasy leagues wanted to change settings and couldn't recognize the mathematical consequences of doing so. The fact is that most people don't readily recognize those issues. I'll quote my earlier attempted explanation again to see if it can help us reach some sort of understanding.

"While I have no idea about what is the actual performance penalty of throwing an interception, let's indulge in a hypothetical and say that it's -5% to stats. So after QB Joe's first interception, his stats drop to "0.95 x X." The penalized attributes increase the likelihood of a second interception, dropping him to "0.95^2 x X," or 0.9025 x X. With each passing interception, we see more drastic results. 0.857, then 0.815, then .774 and so on. By the end of this chain, the original AA level QB now has the effective attributes of someone much lower level.

A more appropriate design would be to have declining penalties where the fourth interception doesn't hurt as much as the first. You'd still see morale drop with every bad play, but instead of being three-fourths of the beginning stats, you could be something more along the lines of 85-90%. That's more realistic and would prevent these absurd results where quarterbacks throw 60% of their regular season interceptions total in one playoff game.
"


Having flat penalties as a result of "bad plays" creates this cascade issue, and it's one that just gets worse and worse with each iteration. Does it really make sense that a fourth interception should damage a quarterback's morale and thus performance as much as the first or second one? Wouldn't you think that by that point the damage was pretty much done?

Mathematically the problem is that flat penalties compound with each point in the sequence, which is why you end up with exponential applications instead of progressive ones. A progressive model would still have performance suffer from making bad plays, and it would still absolutely make confidence critical in mitigating the effects of those plays, but it would not produce a cascade that transforms the average AA quarterback into a D-league one during the course of a game.
 
LionsLover
offline
Link
 
Why isn't this thread locked yet?
 
jamyp
offline
Link
 
Originally posted by LionsLover
Why isn't this thread locked yet?


I really hope it isn't locked until after the Miami-Brainerd game tomorrow. I have a hunch there might be a little discussion going on if anyone watches.
 
Zeke0123
offline
Link
 
Originally posted by LionsLover
Why isn't this thread locked yet?



Because its turned very productive
 
Ken1
offline
Link
 
Originally posted by LionsLover
Why isn't this thread locked yet?


I'd honestly say that the signal to noise ratio in this thread has become very good.

Many people have made many good points about many things. There's really no good reason to lock a thread when most of those posting are not only being civil, but posting truly worthwhile stuff.
 
ress5609
offline
Link
 
Originally posted by jamyp
Originally posted by LionsLover

Why isn't this thread locked yet?


I really hope it isn't locked until after the Miami-Brainerd game tomorrow. I have a hunch there might be a little discussion going on if anyone watches.


?
 
Zeke0123
offline
Link
 
Originally posted by jamyp
Originally posted by LionsLover

Why isn't this thread locked yet?


I really hope it isn't locked until after the Miami-Brainerd game tomorrow. I have a hunch there might be a little discussion going on if anyone watches.


As a spectator what are you expecting to see ?

 
sjmay
offline
Link
 
Originally posted by jdbolick
Originally posted by SCACE

That's where we disagree.

Because you don't understand the mathematics. I'm not intending to disparage you when I say that. This same thing has happened to me countless times when fellow owners in fantasy leagues wanted to change settings and couldn't recognize the mathematical consequences of doing so. The fact is that most people don't readily recognize those issues. I'll quote my earlier attempted explanation again to see if it can help us reach some sort of understanding.

"While I have no idea about what is the actual performance penalty of throwing an interception, let's indulge in a hypothetical and say that it's -5% to stats. So after QB Joe's first interception, his stats drop to "0.95 x X." The penalized attributes increase the likelihood of a second interception, dropping him to "0.95^2 x X," or 0.9025 x X. With each passing interception, we see more drastic results. 0.857, then 0.815, then .774 and so on. By the end of this chain, the original AA level QB now has the effective attributes of someone much lower level.

A more appropriate design would be to have declining penalties where the fourth interception doesn't hurt as much as the first. You'd still see morale drop with every bad play, but instead of being three-fourths of the beginning stats, you could be something more along the lines of 85-90%. That's more realistic and would prevent these absurd results where quarterbacks throw 60% of their regular season interceptions total in one playoff game.
"


Having flat penalties as a result of "bad plays" creates this cascade issue, and it's one that just gets worse and worse with each iteration. Does it really make sense that a fourth interception should damage a quarterback's morale and thus performance as much as the first or second one? Wouldn't you think that by that point the damage was pretty much done?

Mathematically the problem is that flat penalties compound with each point in the sequence, which is why you end up with exponential applications instead of progressive ones. A progressive model would still have performance suffer from making bad plays, and it would still absolutely make confidence critical in mitigating the effects of those plays, but it would not produce a cascade that transforms the average AA quarterback into a D-league one during the course of a game.


Except for the FACT that he wasn't an average AA Quarterback, you can take that one to the bank, you can claim you know football, math, though that his sketchy now, see the 16x16 post if you have any doubt.

You have NO IDEA about builds in this game, that is a given, it is clear that you don't know how they work, how build affects outcome all the fun stuff, you think that because this QB was ranked #9 in something, he is average, when that couldn't be farther from the truth, in fact, I am willing to venture he is probably one of the worst QB in the AA loop, for any country, but that would be pure speculation as A. I ain't gonna look at all the AA teams and QB's there.
and B. they wouldn't have their builds open to begin with.
 
bignasty31
offline
Link
 
Originally posted by jdbolick
Originally posted by SCACE

That's where we disagree.

Because you don't understand the mathematics. I'm not intending to disparage you when I say that. This same thing has happened to me countless times when fellow owners in fantasy leagues wanted to change settings and couldn't recognize the mathematical consequences of doing so. The fact is that most people don't readily recognize those issues. I'll quote my earlier attempted explanation again to see if it can help us reach some sort of understanding.

"While I have no idea about what is the actual performance penalty of throwing an interception, let's indulge in a hypothetical and say that it's -5% to stats. So after QB Joe's first interception, his stats drop to "0.95 x X." The penalized attributes increase the likelihood of a second interception, dropping him to "0.95^2 x X," or 0.9025 x X. With each passing interception, we see more drastic results. 0.857, then 0.815, then .774 and so on. By the end of this chain, the original AA level QB now has the effective attributes of someone much lower level.

A more appropriate design would be to have declining penalties where the fourth interception doesn't hurt as much as the first. You'd still see morale drop with every bad play, but instead of being three-fourths of the beginning stats, you could be something more along the lines of 85-90%. That's more realistic and would prevent these absurd results where quarterbacks throw 60% of their regular season interceptions total in one playoff game.
"


Having flat penalties as a result of "bad plays" creates this cascade issue, and it's one that just gets worse and worse with each iteration. Does it really make sense that a fourth interception should damage a quarterback's morale and thus performance as much as the first or second one? Wouldn't you think that by that point the damage was pretty much done?

Mathematically the problem is that flat penalties compound with each point in the sequence, which is why you end up with exponential applications instead of progressive ones. A progressive model would still have performance suffer from making bad plays, and it would still absolutely make confidence critical in mitigating the effects of those plays, but it would not produce a cascade that transforms the average AA quarterback into a D-league one during the course of a game.


I agree with this. They already get a reduction in their points if they throw a pick so they'll be performing at a lower level and even though the 2nd, 3rd, etc. INT hurts less because it's taking a penalty from a lower number then the original, the fact remains that the relationship should be inversely proportional, in other words, the more mistakes you make, the less it hurts you so 10 INT games don't happen. It won't save the team from losing, but it will provide a better and more realistic sim.
 
Ken1
offline
Link
 
Originally posted by jdbolick
Originally posted by Ken1

I hope that Bort can come to understand-- if he hasn't already-- that ypa is the key stat. It's okay if completion percentage is "too high" if yards per completion (ypc) is "too low," the combo of which would still lead to a fairly realistic ypa.

Very insightful stuff, and it reminds me of an article by Dr. Z a couple of years back regarding the formula for QB rating being obsolete due to an over-reliance on completion percentage as opposed to YPA. Dr. Z was explaining why QB ratings have increased so dramatically over previous decades, with the answer being that teams are throwing more short, high percentage passes than in prior years.



Very interesting.

The NFL QB rating formula isn't bad, but I think I would give ypa and interceptions double weight compared to completion % and TD %.

Completion percentage still does matter, as it measures whether the QB did the minimum of what he was asked to do on a given play. But ypa and interception % are the really important metrics. Ypa measures the reward level and interception % the risk level.
 
Page:
 


You are not logged in. Please log in if you want to post a reply.