Suggestion (ping: PatsFanFrogMan)

[lytic]

My ratings are obviously lower with CA, why not add my brilliant CA into my rating and only do a 2x on my mid... or something to that fact.
I don't need a super mid, all I need is a supernatural CA and I'll kick anyone ass. See what I'm saying?

so 1X defensive and offensive ratings
2x my mid
1x my CA rating

[vex]

That would be a LOT of fun for Froggy to search out and see EVERY team that played CA.

[FrogMan]

lytic, I see what you are saying, but as vexroid is hinting, it's not a rating clearly shown on the ratings list, on the match report. It would be a b*tch to retrieve and would end up being only people who want it to be done this way that end up with it, too much manual changes on a week to week basis...

FM

[lytic]

well, when I get to Iceland, I'd take the time to do the ratings for the CA and pressing peeps. It's not a lot of us...

[lytic]

dola and p.s.

How about them #1 in the NFL Pats!!! AFC EAST CHAMPS!!!!!!!!

:D

[KevinNU7]

You could also make it so that if the person wants it included they have to post it in the weekly thread.

[FrogMan]

Originally posted by lytic
dola and p.s.

How about them #1 in the NFL Pats!!! AFC EAST CHAMPS!!!!!!!!

:D

Heh, heh :D

As posted in the general discussion forum:

GO PATS!!!!

http://cache.boston.com/sports/galleries/2003patriots/game13/17.jpg

http://cache.boston.com/sports/galleries/2003patriots/game13/18.jpg

WOOHOO!!!

[FrogMan]

Originally posted by KevinNU7
You could also make it so that if the person wants it included they have to post it in the weekly thread.

Yes, this could be done, but would need manual adjustments on my end. I'd probably also get kinda frustrated to see people only posting their ratings when they want me to use the CA or pressing rating...

FM

[The Afoci]

I still think that if they don't post their ratings, they probably don't care about the rankings much anyway. I would just take what is posted and leave it at that.

[BishopMVP]

If you were to do it for CA, you'd have to do it for other tactics too. For AIM, do you double the center attack and cut in half the wing attacks? For AOW, do you multiply each wing by 1.5 and cut the center in half?

I think the NSI is over-rated as a tool for measuring team strength and has many problems, for instance I think that Midfield should be multiplied by 4 or 4.5. If it is so much more important than offense or defense, why does it only get put in 3 times as they are? I also think that offensive ratings should be worth 1.5 times as much as their defensive counterparts to reflect Hattrick's (perceived) offensive advantage.

But in the end, it really makes sense to keep the current system because it is so easy to calculate. As long as the ratings don't go above world class, I can look at any match and tell you the NSI in 10 seconds and get a fairly good idea of how strong a team is.

Almost everyone thinks they get screwed somehow by the ratings system, I mean, I sold my brilliant GK and greatly improved my midfield and forwards, but while I feel my team is better, I get lower NSI's than I did before, because I've shifted them from Defense to Offense, where they are harder to get.

And, uh, Go Pats! :)

[lytic]

Like I said, I'll start a CA/Pressing ratings post and I'll do the ratings myself. It wouldn't be too hard to edit your NSI xls (it's real niiiice) for CA/Pressing and I'll pass your the number. It will have to wait until like March though, so maybe next season.

Up to you Frog!

[lytic]

Originally posted by BishopMVP
If you were to do it for CA, you'd have to do it for other tactics too. For AIM, do you double the center attack and cut in half the wing attacks? For AOW, do you multiply each wing by 1.5 and cut the center in half?

And, uh, Go Pats! :)

My point is... midfield is 3x. CA and pressing have a rating to it (Wretched, solid, etc). My mid is a lot lower than it was last season because of it, which lowers my NSI tenfold. Not that it matters that much... but seems like a good way to make up a few points for sacrificing a 3x area.
AIM AOW is more of a choice of attacking and has no rating assigned to it. Also, you ratings are already changed to what AIM/AOW does for your team, so no reason to mess with them for NSI ratings.

Just a thought... I would rather have it 50% mid and 50% rating, but that would get complicated and give partial points.


... and yes, the Pats are cool.

[BishopMVP]

Originally posted by lytic
My point is... midfield is 3x. CA and pressing have a rating to it (Wretched, solid, etc)......... Also, you ratings are already changed to what AIM/AOW does for your team, so no reason to mess with them for NSI ratings.

I had the impression that AIM/AOW only changed the percentage of attacks given to your center or wings rather than their ratings. I think normal it is something like 40% center, 30% each wing, and choosing AIM would up the center and decrease the wings, vice versa for AOW. If I'm wrong or someone has the actual percentages, I'd love to see them.

But my whole point was really where do you stop with this? Your method may be closer than NSI to the real strength, and mine still further, but you are sacrificing the simplicity, and IMO, if you are going to eliminate the easy calculations, you might as well go all-out and see how close you can get your formula. I'd love to see some discussion on my complaints with NSI, and how to tweak these even more to more accurately reflect team strength.

[lytic]

It does change... though it's more noticable on the defense.
AIM will lower wing defense ratings and vice versa.

I don't know the %, I don't think they ever announced it or even gave a clue...

Maybe your right, maybe I'm trying to complicate a simple calculation.

[lytic]

dola

at least it's somthing somewhat interesting to discuss :p

[TargetPractice6]

lytic, you are wrong :p AM and AW changes the amount of attack to a certain area, not the attackr atings itself.

[daedalus]

I do agree with one thing in this thread and, frighteningly enough, it's what The Afoci said [I'll be in the corner crying, if anyone needs me]: skip the folks who doesn't post their ratings in the ratings thread. I don't mean, like, folks who missed it once or twice since that happens sometimes. But, like, folks who consistently doesn't bother.

Back to topic, I don't really see the point of it. As Bishop said, yes, it'd be more precise but that translate to more work for Frogger and I just don't see the point of it. This is just a general way to rank just to do for fun. Tacking on extra work for Frogger just seems silly to me.

[sterlingice]

Originally posted by daedalus
I do agree with one thing in this thread and, frighteningly enough, it's what The Afoci said [I'll be in the corner crying, if anyone needs me]: skip the folks who doesn't post their ratings in the ratings thread. I don't mean, like, folks who missed it once or twice since that happens sometimes. But, like, folks who consistently doesn't bother.

Or, if makes less work for ya, FrogMan, we can all kindof "adopt a delinquent team". If 20 or 30 of the more active people get 1 team each, that's 20 or 30 you don't have to look up.

SI

[The Afoci]

Originally posted by daedalus
I do agree with one thing in this thread and, frighteningly enough, it's what The Afoci said [I'll be in the corner crying, if anyone needs me]:


As I always say, don't wipe your butt and it won't hurt.

[BishopMVP]

Originally posted by TargetPractice6
AM and AW changes the amount of attack to a certain area, not the attack ratings itself.

Woot! Any chance you know the percentages and how they change it? Or is that classified material available to National Coaches only? ;)

Originally posted by daedalus
Back to topic, I don't really see the point of it. As Bishop said, yes, it'd be more precise but that translate to more work for Frogger and I just don't see the point of it. This is just a general way to rank just to do for fun. Tacking on extra work for Frogger just seems silly to me.

Actually, a couple of my suggestions could be implemented as just changes in the spreadsheet rather than more work getting them from Hattrick. These are the making midfield/attack more valuable. Just for an example, keep the defense as is, 1x (passable=6), make attack ratings 1.5x (passable=9) and midfield 6x (passable=36.) Right now, attack and defense are broken into L/R/C and midfield isn't, so I never understood how making Midfield 3x represented how important it is. This only made it equal to defense/attack.

Disadvantages I can see are:
- doesn't take into account tactics (CA/AIM, etc.)
- would force people to re-think how good a rating is as they would increase across the board
- doubles the amount of difference between mid ratings (i.e. 5.9 inadequate vs. 6.0 passable)

Any one else have thoughts on trying a sytem like that?

[TargetPractice6]

No, I don't know the percentages. Apoc might have a rough idea since he has been building the new tactics into his match calc.

[FrogMan]

For now I don't plan on changing much to the rankings. Not always because it would be a whole lot of work, but more because they are easy to refer to. The formula is the same as HATStats, which makes it kinda easy to find out where your team stands.

Not saying Bishop's idea isn't good, or that the idea of using CA & pressing isn't good, but I just don't feel like it, at least not right now...

FM

[BishopMVP]

Originally posted by FrogMan
Not saying Bishop's idea isn't good, or that the idea of using CA & pressing isn't good, but I just don't feel like it, at least not right now...

I wasn't suggesting that at all. I was thinking more along theoretical lines/maybe leading into a match calculator like Apoc's old one. I think that the discussion would be more rewarding than any formula that came out.

[TargetPractice6]

The 3x does weight the midfield though. There is no left or right side midfield to the match engine, so multiplying it by 3 doesn't really reflect how things work in Hattrick. It just gives it a larger portion of the rating that any other single rating.

It gives the midfield much more weight than anything else. You can't just say that it's because the attack as three ratings and so does the defense. That may be so, but you can't just group the attack as a whole because that isn't how the game engine looks at it. There are three separate attacks and each one does something different.

So as it is calculated now, the midfield is given more weight than the center and side attacks.

[Mr. Wednesday]

Something that one of the LOTA members suggested was taking into account variation in the ratings (by subtracting the standard deviation, IIRC) -- I believe the theory was that more balanced ratings tended to work better. Any thoughts for or against?

[finkenst]

Yeah.. I have a thought:

I hated statistics.

[daedalus]

Originally posted by BishopMVP
Woot! Any chance you know the percentages and how they change it? Or is that classified material available to National Coaches only? ;)According to the Help, it's 20-40% depending on your proficiencies. For more detailed numbers, you could probably be informed but the National Coaches and their enforcers would be forced to deal with you in a mostly lethal manner.Originally posted by daedalus
Back to topic, I don't really see the point of it. As Bishop said, yes, it'd be more precise but that translate to more work for Frogger and I just don't see the point of it. This is just a general way to rank just to do for fun. Tacking on extra work for Frogger just seems silly to me.Actually, a couple of my suggestions could be implemented as just changes in the spreadsheet rather than more work getting them from Hattrick. These are the making midfield/attack more valuable.Oh, I don't have a problem with what you're suggesting. [I can even see a lot of the points you're making.] I'm referring more to the fact that adding conditional things like changing values and multipliers for particular tactics and what nots just makes far, far more work for Frogger than really necessary.

[BishopMVP]

Originally posted by daedalus
Oh, I don't have a problem with what you're suggesting. [I can even see a lot of the points you're making.] I'm referring more to the fact that adding conditional things like changing values and multipliers for particular tactics and what nots just makes far, far more work for Frogger than really necessary.

Sorry if I came off that way. Adding in specialties or tactics, basically anything not in that right column would mean more work. I just figured I'd point it out because my first post was kind of rambling and some suggestions would have meant more work, others not. Thanks for the attack %, although I'd love if someone else had a more precise breakdown (And didn't have to kill me if I'm told ;) ).

Originally posted by TargetPractice6
The 3x does weight the midfield though. There is no left or right side midfield to the match engine, so multiplying it by 3 doesn't really reflect how things work in Hattrick. It just gives it a larger portion of the rating that any other single rating.

It gives the midfield much more weight than anything else. You can't just say that it's because the attack as three ratings and so does the defense. That may be so, but you can't just group the attack as a whole because that isn't how the game engine looks at it. There are three separate attacks and each one does something different.

So as it is calculated now, the midfield is given more weight than the center and side attacks.

I see where you're coming from, but just because the match engine doesn't list midfield 3 times doesn't mean it's not there. When I think of the ratings, I kind of divide the field into a 3x3 square, with each defense and attack rating taking up 1 box and the midfield being the remaining three. This could be off, but that's how I look at the ratings.




As for Mr. Wednesday's point about using standard deviation, I agree partway, but if its taken out to the extreme (all excellent ratings, except right attack is world class and left defense is wretched) it starts to fall apart. I like the thought, but it just seems even harder than tactics to get an accurate read on.

[TargetPractice6]

But that's not how it works. There is only one midfield that feeds all three attacks.

[BishopMVP]

Originally posted by TargetPractice6
But that's not how it works. There is only one midfield that feeds all three attacks.

The way I read this statement is that you have 3 combinations.

1. Midfield - Right Attack
2. Midfield - Center Attack
3. Midfield - Left Attack

I look at this and think that it means the midfield is already counted 3 times in the game engine, and thus making it 3x in NSI calculations doesn't reflect any added importance.



We may just be at an impasse where you and I look at the same thing and read two different things, but I'd love to hear more thoughts on this, especially if other FOFC'ers want to get involved.

[TargetPractice6]

No, it is only counted once. The midfield distrbutes a number of scoring chances relative to you opponent's attacks. Those chances are then distributed to one of the three attacks, but the level of midfield has nothing to do with which one it goes to. By no means are there three separate midfield ratings for each one.

The midfield multiplied by 3 gives it more importance than any other rating. All three attacks grouped together don't really tell you anything related to the game engine because all three opperate imdependently.

[Mr. Wednesday]

If we knew the actual attack distribution, it would problably make the most sense to do a weighted average of the attack ratings. As it is, maybe it would make more sense to you if you viewed it as averaging the attack ratings, rather than multiplying midfield by 3.

[BishopMVP]

I think Mr. Wednesday is saying what I was trying to. Let's assume for now that each attack gets 1/3 of the possible chances. Then rather than saying that midfield is counted 3 times in the game engine, I should have said that midfield is counted once and then each attack is worth 1/3 this.

In my last post, I wasn't trying to say there were 3 different mid ratings, but one that fed all three. I wanted to have a tree with all 3 attack ratings branching off from the midfield, but it's hard to do that on a message board.

[TargetPractice6]

Each attack ratings is worth 1/3 of the midfield because the midfield is already weighted in NSI. The three attack ratings act independently of each other, so there is no one attack that equals the midfield multiplied by three. So in essence, you have the midfield worth three times as much as any other rating when figuring NSI.

[Mr. Wednesday]

The attack ratings act independently, but they only act 1/3 of the time (for the sake of argument, because we don't know what the actual distribution is, or at least I don't). The midfield generates the opportunities, then 1/3 go to right attack, 1/3 to central attack, and 1/3 to left attack. They certainly don't act independently, in the sense that there's a finite number of chances to distribute and if one goes to one attack, it doesn't go to the other two. Right? That's what I understand from what I've read in the conferences, some of it from Bjorn.