|
Post by squirrel on Sept 27, 2018 14:42:14 GMT -5
Ran him through again....now -8.2!! Is the formula correct squirrel? I don't see him throwing picks at that high rate in games. Few things: 1. When I get home from the office I’ll run my GML QB through this to check the #s are right 2. The numbers aren’t fully predictive for one player. Over a population the correlation between the answer of his and TD pass / INT ratio is clearly significant 3. Gameplanning obviously matters too ie how many long passes is your guy throwing
|
|
|
Post by squirrel on Sept 27, 2018 14:43:02 GMT -5
Ran him through again....now -8.2!! Is the formula correct squirrel? I don't see him throwing picks at that high rate in games. Few things: 1. When I get home from the office I’ll run my GML QB through this to check the #s are right 2. The numbers aren’t fully predictive for one player. Over a population the correlation between the answer of this and TD pass / INT ratio is clearly significant 3. Gameplanning obviously matters too ie how many long passes is your guy throwing
|
|
|
Post by Nick on Sept 27, 2018 14:43:56 GMT -5
Greathouse has excellent AI...with a 21 SOL. Shows how much you can rely on combines to tell the truth.
|
|
|
Post by Bighouserulez on Sept 27, 2018 15:14:39 GMT -5
As we know Combines are a part of the equation that is what makes the game great.
I will never forget having the argument in the V (It was Civil) about a certain QB. MY spreadsheet showed we was going to be like 70 avoid AI and you really felt the opposite. I wish that forum was up for my sheets and the convo ug.
|
|
|
Post by ManWithNoName on Sept 27, 2018 15:18:13 GMT -5
As we know Combines are a part of the equation that is what makes the game great. I will never forget having the argument in the V (It was Civil) about a certain QB. MY spreadsheet showed we was going to be like 70 avoid AI and you really felt the opposite. I wish that forum was up for my sheets and the convo ug. Who was right?
|
|
|
Post by Bighouserulez on Sept 27, 2018 15:23:15 GMT -5
So that gives Scott in GML like a 1.9 and i would have guess his bar would have been in the 70 and it gives my QB i asked about a -10 avoid INT rating . I like it Squirell it gives some work we can use to make it work. My Qb Mr Krabs here in RZB is a -20. I think somehow you have to tweak in the sole score to help you find the missing component. Crabtree is a 43 Sole 7 intel and 45 rated kid so you have to guess the hidden avoid AI is at least decent in him. Neat Squirrel this gives me some headway
|
|
|
Post by Bighouserulez on Sept 27, 2018 15:26:52 GMT -5
As we know Combines are a part of the equation that is what makes the game great. I will never forget having the argument in the V (It was Civil) about a certain QB. MY spreadsheet showed we was going to be like 70 avoid AI and you really felt the opposite. I wish that forum was up for my sheets and the convo ug. Who was right? Ha i think that it was never really settled but his QB threw more picks then he wanted but was better then i thought. I try not to ever keep score but make steps with all this stuff. I made spreadsheets to show what combines led to what each bar would be. It has changed since the new game but it always gave you another view to go with DA to show how much a static bar was above where it was supposed to be. I had an avoid fumble one and an avoid interception one. The math was not dead on but you could really see what sole/intel/rd would make a good AI or bad AI.
|
|
|
Post by Nick on Sept 27, 2018 15:38:09 GMT -5
So that gives Scott in GML like a 1.9 and i would have guess his bar would have been in the 70 and it gives my QB i asked about a -10 avoid INT rating . I like it Squirell it gives some work we can use to make it work. My Qb Mr Krabs here in RZB is a -20. I think somehow you have to tweak in the sole score to help you find the missing component. Crabtree is a 43 Sole 7 intel and 45 rated kid so you have to guess the hidden avoid AI is at least decent in him. Neat Squirrel this gives me some headway From what I am reading I don't think this is being understood properly. The SOL score is totally irrelevant in this. It is a calculation that provides a true value. The SOL doesn't come in to it. Crabtree's AI is pretty solid. Without calculating exactly, I would estimate 60.
|
|
|
Post by Nick on Sept 27, 2018 15:41:07 GMT -5
You guys get the the +/- is in relation to the AI being over or under the overall rating? The confusing bit being a negative score is good and positive is bad.
|
|
|
Post by Bighouserulez on Sept 27, 2018 15:42:06 GMT -5
Yes at our eyes Nick you and i can figure it out. That is a hidden rating or Bar. If you looked at squirrel's sheet it showed has a missing component.
Sole does comes into it just on the mathmatics.
How would we calculate his AI rating without the sole score?
|
|
|
Post by Bighouserulez on Sept 27, 2018 15:50:43 GMT -5
I get what you are saying i guess. It is just taking the variance in the overall rating.
OK i was more working with the rough idea on drafting. My fault i misunderstood your post a second.
|
|
|
Post by Nick on Sept 27, 2018 15:53:50 GMT -5
The easiest way to move forward is to forget SOL. Combines are not accurate now as the change every year. We can also see from the calculation just how inaccurate they now are. As per Greathouse discussed earlier whose AI is way better than his SOL/INT would have you believe.
So as you know the basic rule is if you see a guy who appears to have too few red bars in general for his overall rating that illustrates he has high AI.
When we go in to this over/under calculation the base line figure is the players overall rating. So if Crabtree comes under by x, x+y (y being overall rating) = AI. So because Steigenbum comes under the same amount as Crabtree, thereby having the same x Steigenbum's AI is higher because his y is higher than Crabtree's.
So you can't just have a + or - because the + or - is in direct relation to his overall rating.
I think Squirrel's spreadsheet is going in the right direction. But unless I am missing something there is nothing in there that allows for the overall rating to be taken into consideration. I think this could be built in, and then at that point we have really have a nice tool.
|
|
|
Post by ManWithNoName on Sept 27, 2018 15:54:23 GMT -5
You guys get the the +/- is in relation to the AI being over or under the overall rating? The confusing bit being a negative score is good and positive is bad. When I was looking at it earlier it said positive was good and negative was bad.
|
|
|
Post by Nick on Sept 27, 2018 15:57:31 GMT -5
Seeing more red bars is bad. My definition of that is a + or I use the term "over".
|
|
|
Post by squirrel on Sept 27, 2018 15:59:53 GMT -5
Right, sorry again everyone, when I was at the office google docs was having trouble with the firewall. Now I'm home, I've compared the outputs of this against my GML QB and I'm OK with the calculation.
There's 5 people in the google doc, I'd encourage you all to hit refresh and get the latest version. Also, try not to change anything other than the green cells.
(in fact I'm sitting here watching someone update this in real time...there's probably a more stable way to do this, but I don't know what it is. Probably just upload an excel file)
Anyway, FWIW the way I look at it is...taking my GML QB as an example
1. If you multiply all his bars through you get a score of 33. That's what his overall rating 'should be' just based on the bars
2. But his actual overall rating is 32.
3. That means the element one can ascribe to hidden rating is minus 1 i.e. his hidden ratings are bad
4. His hidden ratings, I think, are for avoid INT and avoid fumbles. I have no idea whether that's right or in what proportion
5. Now, I have managed to get him to have a TD pass / INT ratio of just over 2x, which is good. A lot of that is me deliberately managing it to give the QB a chance because I know this is a weakness. If I wasn't doing various other things (nothing super clever, just obvious things like having good WRs) I would expect the ratio would be bad
|
|