League Rule Changes for 2014 AAG40 2.0

[boldfresh]

why no retroactive tier changes?  i don’t get it.

[MrRoboto]

@Boldfresh:

why no retroactive tier changes?  i don’t get it.

Well I don’t like the idea of losing PPG, just because one of my earlier opponents messed up a game or got diced, thus dropping a tier.

Or I don’t like getting free PPG, just because one of my earlier opponents improved a lot, thus gaining a tier. I didn’t defeat him, when he was strong - only, when he was still weak - so I don’t deserve these extra PPG.

[boldfresh]

i guess that could apply - but just as often it might be that you beat a player better than you when he was a tier 2 that had been diced in a couple of games.  but over time, he rises back to tier 1 where he really should have always been so you should get the credit for beating a tier 1.

[lexgreat]

Very constructed example, I think, Boldfresh.

In sports, you beat a player on day x and get the points the player is worth at that time, no? Not what he has been worth 2 weeks before that, when he was ranked 10 places better or worse.

If retroactive changes are kept, nobody will complain, but it feels counter-intuitive. If you play a beginner, you have it easier. 20 games on and this beginner will be better, but still the earlier win was against a “worse” player and should always be counted like that.

@Seth: I can see why tiers are practical, but do they dont feel very objective…who is tier 2, who is tier 3, who draws the borders and where, which exceptions should apply etc. If tiers are necessary to decide how much a win is worth, then there should at least be an objective way to create them, easy to understand for all players.
Surely not easy to find a better system than the one we have, though, and I dont claim I have one. More sophisticated systems also would likely need much more work from the ones doing the rankings, and you cannot ask more from them with the current commitment so high already. Using the current ranks of both players as modifiers of base points and bonus points for beating the qualifying positions surely sounds like a lot of work, for example.

[boldfresh]

@alexgreat:

Very constructed example, I think, Boldfresh.

In sports, you beat a player on day x and get the points the player is worth at that time, no? Not what he has been worth 2 weeks before that, when he was ranked 10 places better or worse.

If retroactive changes are kept, nobody will complain, but it feels counter-intuitive. If you play a beginner, you have it easier. 20 games on and this beginner will be better, but still the earlier win was against a “worse” player and should always be counted like that.

@Seth: I can see why tiers are practical, but do they dont feel very objective…who is tier 2, who is tier 3, who draws the borders and where, which exceptions should apply etc. If tiers are necessary to decide how much a win is worth, then there should at least be an objective way to create them, easy to understand for all players.
Surely not easy to find a better system than the one we have, though, and I dont claim I have one. More sophisticated systems also would likely need much more work from the ones doing the rankings, and you cannot ask more from them with the current commitment so high already. Using the current ranks of both players as modifiers of base points and bonus points for beating the qualifying positions surely sounds like a lot of work, for example.

gamer draws the lines very clearly between tiers - so that is objective no?

[gamerman01]

The two issues are related:
Retroactive changes and clear boundaries at tiers.

Retroactive changes have a clear purpose.  An unknown entity comes along and loses a couple of games.  He’s now tier 3, but really, he’s a tier 1 player - he just lost twice to tier 1’s.

Now this player is 0-2 with a 2.00 average.  A win against him under MrRoboto and alexgreat’s thinking is not worth much.  In my system it’s worth 4 points, but this is a tier 1 player.
After starting out 2-0, he reels off 6 straight wins and is 6-2 and 4.00 average, tier 1.

All 8 players who played him should get credit for playing a tier 1 player.  But if there are no retroactive changes, then some players only get credit for beating a tier 3, and some for losing to a tier 2, but all along this is a tier 1 player.

Also, without retroactive changes, then how many points does the first guy to beat him get?
See, with my system, if it turns out that newbie you played was a star, you will later get credit for playing a star.
If he’s a hopeless newbie, that also will be reflected later.
Under the proposals I’m reading, you would just get some average prize for defeating this new player to the league, but there is a very wide disparity in competence levels among new players!

Hobo is a good example.
He started in the league this year.  Hobo has a lot of A&A experience, but had a lot to learn about G40.  Bold got an easy win off him, but Hobo came back and beat Bold twice and is now solidly a tier 1 player.  Bold gets credit for beating a tier 1 and losing twice to a tier 1.  Now I agree with you that his first win against Hobo was a relatively easy 6 points.  No system is perfect.  But there are strong merits to retroactive changes in A&A (please don’t compare to sports - it’s not the same IMO), and OVERALL it seems to average out and work out.

My challenge to you - name a few players who are significantly under-rated or over-rated, who have played at least 10 games.

There are only 2 players I can think of who had significant changes in ranking positions after their first few games.
They are Snakeeyes and Hobo.  These 2 players had high A&A IQ all along (I played Snake one of his earliest games, and while I beat him easily, I also recognized that he was very talented and told him he would win a lot of games, at that time), and they started out a bit low because they were early on the learning curve.  But it only took a few games, and they were on the rise, and just kept winning.  A couple of players played these 2 on their rise up, and now have credit for playing a tier 1, and not a tier 2 or tier 3.  You can only play a couple games against a given opponent anyway.  The difference in a single game between playing a tier 1 or a tier 2 is only 1 point.  Yes, sure you can find situations where the amount of points awarded may not be exactly appropriate, but these things tend to average out.  Again, find me a few players who are inaccurately ranked.  I don’t know of any.

I don’t think I’ve seen any players ranked highly who dropped precipitously.  I have seen few games where a player who was “supposed” to win, actually lost.  The system is definitely working.  MrRoboto, you need to hang around longer - respectfully, I think you are giving out too much opinion for not enough observation.  I predict that your ranking will settle where it should be by the time you complete 5-6 games.

[gamerman01]

Actually, Snake hasn’t cracked tier 1 yet.
So while I played Snake when he probably wasn’t as good, I only have 2 extra points as a result of retroactive changes.

Now the question is:
Was Snake really tier 3 quality when I played him twice? 
I’m not sure that he was - he was maybe tier 2 quality even at that time.  And even if he was, retroactive points only give me 2 more points.

The purpose of retroactive points is to correct for earlier inaccurate ratings.  I understand the downside is that you could be getting “free” points by the fact that a guy you beat earlier has improved.  As far as losing points because your past opponent later did something stupid or got diced - I say that will work itself out.  If he’s really tier 1 quality and dipped to tier 2, then he will win and get back to tier 1 and you will get your points back.  This is one of the reasons I am slow to change tiers on somebody, like when Bold was tier 1 all year but reported some losses and got all the way down to about 3.3.  Again, this is where human beats computer.  I figured he would win some games and get back to tier 1, and also he told me he was winning some games.  He is already up to 3.43.

Again, I appreciate the thoughts and I appreciate constructive criticism, but that doesn’t mean I agree with it  :-)
I will be re-reading all these posts and suggestions at a future date when deciding on the system for next year, so keep the ideas coming.  I do really like some of the ideas posited already, so don’t take by my responses that I am stone-walling everything, because I am not.

[gamerman01]

@Gamerman01:

As far as losing points because your past opponent later did something stupid or got diced - I say that will work itself out.  If he’s really tier 1 quality and dipped to tier 2, then he will win and get back to tier 1 and you will get your points back.

Another possibility is that your opponent was actually over-rated when you beat him.  Which would be a point in favor of the retroactive changing and a point against you earning 6 points and not having any taken away from you when his true colors are revealed!

There are 2 sides to the coin.

[seth4god]

Ok, yeah that all makes sense and you’ve convinced me that the retro point/tier changes are the way to go.

[lexgreat]

I think that retroactve changes dont reflect the learning curve. You played Snake when he was starting, it was easier to beat him then than now. Dont you agree? Why would you get the same points than, say, someone who plays him now that Snake has learned a lot and is the harder to beat? If someone is a genius all along, then it would be correct to change retroactively, but I think thats not the norm. People dont have the same strenght from first to last game played. You have, Gamerman, but everyone starting compete here likely improves.

I feel this is a valid side of a coin, to be discussed as we do, and if not seen that way by the majority, not enacted, with no one left disappointed. Easy.

[Jennifer]

Which is part of why I suggested that your standing is determined by your ranking last year.  Perhaps all new players should be ranked tier 3 instead of tier 4 and they can earn a lower ranking or a higher ranking based on their performance that first year?  (those playing this year are earning their tier for next year.)

Just a suggestion

[boldfresh]

Ok so if I end as a tier 2 this year then everyone beating me next year gets only credit for tier 2.  :roll:

[Jennifer]

Yes, but everyone who plays you knows you are worth, say, 4 points or whatever a tier 2 is worth then.  There’s no ambiguity.

[Guest]

@seththenewb:

And as Gamer said, he’s not a computer.

He isn’t? I see him as the early version of skynet. Learning how to best wipe out the human race.

[gamerman01]

:evil: :evil:

Must…. find… John… Connor…

[boldfresh]

@Cmdr:

Yes, but everyone who plays you knows you are worth, say, 4 points or whatever a tier 2 is worth then.  There’s no ambiguity.

jenn, that statement is just silly.  have you been talking to cow too much?  seems it may be scrambling your brain.  :-P

[gamerman01]

Whatever system, exactly, we use next year, things will be reset.  We are towards the end of the year when “learning curve” effects are the greatest.  I’m currently thinking about implementing Jenn’s suggestion, at least in part.  Players who have played many games and we know where they stand, will have a fixed tier based on 12-13.
One of the problems with non-retroactive changes to points, is that highly ranked players will have a disincentive to play a newcomer.  As it is now, if the newcomer is good, they will rise in the rankings and the player who played them early on will eventually get the credit they deserve for playing a new player.
There are +'s and -'s to whatever system you use.  I will probably make some tweaks for next year, but no matter what I do there will be criticism, and some of it will be valid.

Still, like I said, my 2 wins against Snake only got me 2 extra points, IF he was truly a tier 3 when I played him.  Like I said, I told him he was going to win a lot of games, which I wouldn’t say to your average tier 3, because I could see that he had high A&A IQ - just inexperienced with G40.

But the opposite can happen, too.  You play someone who is 2-0 and has a 4.00 rating, but all he did was beat 2 tier 3’s and is currently over-rated.  You beat him and it takes him down to tier 2, so you get 5 points and not 6 because of the retroactivity.  Works both ways.  Tends to average out.
One more time…… show me a few players who are significantly under or over-rated.  I don’t think you can.  Which means the system is working very, very well, as several accomplished players have spoken out and said over the past several months.

[gamerman01]

@Gamerman01:

One more time…… show me a few players who are significantly under or over-rated.  I don’t think you can.

Who have played 6+ games

[gamerman01]

@Boldfresh:

@Cmdr:

Yes, but everyone who plays you knows you are worth, say, 4 points or whatever a tier 2 is worth then.  There’s no ambiguity.

jenn, that statement is just silly.  have you been talking to cow too much?  seems it may be scrambling your brain.   :-P

It made sense to me….
She’s saying that players know if they beat you they will gain the points for beating a tier 2 and it won’t change on them.

Anyway, the idea behind retroactive changes is that we now have a better picture of how good the player actually was all along, and so people who played him before should get credit added/taken away for that…

Either way, I think the effect is pretty minor.  It’s not like a 4.00 player is going to be a 3.50 if I did it different.  :roll:

[boldfresh]

i think that once a player is a known quantity to gamer he can assign a tier value to that player.  if gamer has a record to go on for the previous year, he can feel pretty comfortable making that determination.  I think a good compromise would be to make 5 tiers.  give new players tier 1 status for the first 5 games, then have gamer adjust the tier as appropriate.  no retroactive changes in points after the new tier change.  then if there is further improvement over the next 5 games the player could be moved up again - and again, no retroactive points.  but someone who beat that player in their second set of 5 games would get the points for beating a tier 2.

remember, the goal for all of this is to give the best approximation of skill levels in the league so good matches can be found and a rightful champion can be crowned right?

if someone thinks gamer would make changes in tiers without good reason, i challenge them to show me once that it happens.