Re: [mepbmlist] KISS (Player Ratings)

[corsairs game 101]

>I'm sorry to be negative,

I, on the other hand, have no problem being negative of what I see as a
really bad idea. I have no problem being negative on my ideas, and don't
even require people tell their name to anyone on the planet that wants to
sign up for the group.

>and I do appreciate the large amount of work that
>someone's put in. BUT, I think you've drowned the baby in a bathtub of
>arcane algebra. For a player ratings system to work, it needs to be
>popular, and to be popular, it needs to be easy to understand.

Understand this. You'll gain about 35-55 points for each game you win, and
lose that many points for each game you lose. Win against high ranking
opponents, gain at the top end. Lose to weak ranked opponents and you lose
at the top end. Losing to high ranked players, or beating low ranked,
results in the low end.

The result is, if you win 4 of 6 games in a year against similar opponents,
and you'll have a score of about 1590(1500 start + 4*45 - 2*45). If you
only completed 2 games in that amount of time, and won them both, you would
also have a 1590 rank (1500+2*45).

So, the highest ranked players will be the players that have the highest
difference in wins and losses (wins-losses).

So, you can be the best player, winning every game, but if you only play 1
game at a time, each taking a year, you'll have a horrid score. If you win
2/3rds of your games, but play 4-5 at a time, stomping weak opponents in a
dozen turns, you'll have a great score.

Unlike some, I did see "stacking" going on in the bad old days of VCs.
Heck, I even wrote an article about it way back in the days of Mouth of
Sauron. There was lots of talk of how many players would drop Woodmen on
turn 1. I had several games where I'd request Corsairs or an elven nation
(being tired of Rhudaur and Woodmen), only to be greated with another sucky
nation setup.

VCs did effect how players selected nations, which side neutrals chose, and
how they played the mid to end games. I prefer how it is now, and would
hate to see anything come along that changes the current "what would be fun"
way of selecting nations, choosing sides, and playing a nation.

As I said in prior emails, at best, I'll ignore these ranking. At worst,
they will adversly effect the game. Instead of "fun", some people would be
using the game as a tool to effect their player ranking.

Signed,
Darrell B. Shimel Jr. (He who refuses to give his real name, becuase ME
Games knows who I am, and it should not be anyone elses business who I am.)

[Jason Bennett]

corsairs game 101 wrote:

> So, you can be the best player, winning every game, but if you only play 1
> game at a time, each taking a year, you'll have a horrid score. If you win
> 2/3rds of your games, but play 4-5 at a time, stomping weak opponents in a
> dozen turns, you'll have a great score.

But that's a good thing! The only way to prove that you're the best
player is to win consistently. There's no rating system in the world
that can give you an accurate score with an incomplete sample space.

You're looking at it in just the first year. Over five years, those
players' scores will approach appropriate numbers. There's no way to
find the correct rating, though, without enough games having been played.

> VCs did effect how players selected nations, which side neutrals chose, and
> how they played the mid to end games. I prefer how it is now, and would
> hate to see anything come along that changes the current "what would be fun"
> way of selecting nations, choosing sides, and playing a nation.

There's no evidence, though, that such a thing will happen with these
rankings. There's no money at stake for a high individual rating. Clint
has already said that the VC ratings are there mostly for backwards
compatibility, and the team rating is what matters.

jason

--
Jason Bennett, jasonab@...
E pur si muove!

[corsairs game 101]

>You're looking at it in just the first year. Over five years, those
>players' scores will approach appropriate numbers. There's no way to
>find the correct rating, though, without enough games having been played.

Incorrect. A player that wins 6 of 8 games a year, will continue to have
the same score as a player that wins 4 of 4 every year.

A player that wins 5 of 8 will have the same score as someone that wins 2 of
2. Multiplied out 5 years, 25 wins in 40 games is equal to 10 wins in 10
games.

It will be better to drop a game that is looking grim, and jump into a game
that you have a better chance of winning. You will get what you reward, and
this system will reward more/shorter games against sub-par opponents.

Corsairs (ooops, Dwarves and Cardolan 101)
AKA Darrell Shimel (just so that my opinion is more some how valid)

[Middle Earth PBM Games]

At 07:14 16/09/02, you wrote:
> >You're looking at it in just the first year. Over five years, those
> >players' scores will approach appropriate numbers. There's no way to
> >find the correct rating, though, without enough games having been played.
>
>
>Incorrect. A player that wins 6 of 8 games a year, will continue to have
>the same score as a player that wins 4 of 4 every year.
>
>A player that wins 5 of 8 will have the same score as someone that wins 2 of
>2. Multiplied out 5 years, 25 wins in 40 games is equal to 10 wins in 10
>games.
>
>It will be better to drop a game that is looking grim, and jump into a game
>that you have a better chance of winning. You will get what you reward, and
>this system will reward more/shorter games against sub-par opponents.

Not sure how it does that. But if you don't like the system what would you
suggest as an alternate system? I've not checked your maths here, but
assuming your case is valid what weighting method would work?

>AKA Darrell Shimel (just so that my opinion is more some how valid)

*** Just so that I will take into account your opinion. I won't
otherwise. Your name has no correlation to the validity of your point. :-)

Clint

[Jason Bennett]

corsairs game 101 wrote:
>>You're looking at it in just the first year. Over five years, those
>>players' scores will approach appropriate numbers. There's no way to
>>find the correct rating, though, without enough games having been played.
>
> Incorrect. A player that wins 6 of 8 games a year, will continue to have
> the same score as a player that wins 4 of 4 every year.
>
> A player that wins 5 of 8 will have the same score as someone that wins 2 of
> 2. Multiplied out 5 years, 25 wins in 40 games is equal to 10 wins in 10
> games.

All of that is true, assuming you always play exactly the same level of
opposition. While Clint's formula is vastly simplified over chess
ratings (although apparently still not enough for some), if you always
play the same opposition level in chess, you can never hope to have a
rating much higher than your opponent.

With these ratings, I agree that the player who is 6/8 against 1600
point opposition should have a lower rating than a player who goes 4/4,
but only slightly. And the 4/4 player should never have much more than a
1600 rating himself, since he only plays that level of opposition.

> It will be better to drop a game that is looking grim, and jump into a game
> that you have a better chance of winning. You will get what you reward, and
> this system will reward more/shorter games against sub-par opponents.

It certainly won't reward play against sub-par opponents, as your rating
will increase slowly or even decrease if you play low-rated opponents.

I agree that players should be penalized for dropping (either in
ratings, or in a drop count).

jason


--
Jason Bennett, jasonab@...
E pur si muove!

[Middle Earth PBM Games]

>VCs did effect how players selected nations, which side neutrals chose, and
>how they played the mid to end games. I prefer how it is now, and would
>hate to see anything come along that changes the current "what would be fun"
>way of selecting nations, choosing sides, and playing a nation.

Cool a postive remark... :-) In all seriousness I am trying to understand
why you would choose nations that give you more VPs. The nations that are
already picked are such nations. 14,10,21,22,9 are such nations and they
are almost always 1st to go (oops missed 20). Whereas the low VP scoring
nations such as 1,2,3,16,18,24 are last to go. (Note I am aware that this
is not 100% accurate but it's pretty close). I don't see how this would
change things. Let's say that we scrapped the Istari rating (the ONLY one
which has VPs as an aid to your overall Rating) what are players thoughts
on the others?

>As I said in prior emails, at best, I'll ignore these ranking.

*** Sure thing - that would be fine. I think they will be the closest
present approximation to what is a good player though...

>At worst,
>they will adversly effect the game. Instead of "fun", some people would be
>using the game as a tool to effect their player ranking.

** ... and lots of fun (based on my experience where I have played in other
such Ratings related games). When I moved from Board 3 to board 2 of my
school chess team it was great... :-) When I beat an opponent in MtG who
had a higher rating it was such a high, and losing to a spod was such a
low.... :-) Same here I would say when your team does well against a good
team - and loses badly to a bad team. (I compare this with the groups of
ME players that when we get together there's not much "bragging/enjoyment"
in saying my economy improved by 2% this turn, but "heh I killed Din Ohtar
in a 3-1 challenge! Yeh!!!" both might have the same dramatic impact on
the game but you know which one is the more exciting.) I think this type
of rating would add to the excitement (as seen in the World Champ games and
the GM team playing against various other teams as a point of honour [on
both sides]) of the game overall.

Clint

[corsairs game 101]

In response to one of my many positive emails, ME Games wrote
>Cool a postive remark... :-)

WOW, I'm surprised ME Games would make so emotive and combative comment to a
paying customer. Guess putting my name to my emails hasn't made my money
any more valuable to ME Games.

Too bad more time wasn't spent in finding the VERY BIG flaw in my prior
comments, and less time reacting violently to the message.

>Let's say that we scrapped the Istari rating (the ONLY one
>which has VPs as an aid to your overall Rating) what are players thoughts
>on the others?

Still gives too much weight to how many games are completed. Places the
emphasis on winning. Discourages people from playing in randomly assembled
games.

As stated above, there was a big flaw in my very first email on this topic.
Had that flaw been found by the proponents fo the system, it would have
helped the conversation move along.

I mad an error in my math by skipping a step.

Assuming all 1500 rank players in the game, the number of points awarded for
a 14 v. 11 game is 45 + [(11*1500 - 14*1500)/150]. I forgot the 45+. I
calculated a change of 30, when in fact, you would only get a change in
15.....

So, while gaining ranking for lots of very short, unbalanced games, the
gains would come a lot slower than I'd previously calculated.

Still, the rankings DO reward victories over lesser opponents. They reward
lots of games vs. long games against great opponents.

I'd much prefer to see no ranking system over a ranking system. I'd prefer
a ranking system that ranked a player that is 10 of 10 over a player that is
25 of 40.

I'd prefer a ranking system that truely ranked the skill of a player, over
one based on wins and losses. I've done poorly in games and won, then had
games where I personally played well, but lost. I have no idea of how to
design the ideal (or even a good) ranking system, and thus I prefer not
having one over having one.

Darrell Shimel

[Middle Earth PBM Games]

At 18:01 16/09/02, you wrote:
>In response to one of my many positive emails, ME Games wrote
> >Cool a postive remark... :-)
>
>WOW, I'm surprised ME Games would make so emotive and combative comment to a
>paying customer. Guess putting my name to my emails hasn't made my money
>any more valuable to ME Games.

Sorry I thought that the smiley face was attempting to decrease
tensions. My apologies if it did not do that. Note I have responded to
your emails where I felt I had a valid point to make where you have signed
them. Thanks for joining the debate.

>Too bad more time wasn't spent in finding the VERY BIG flaw in my prior
>comments, and less time reacting violently to the message.

*** Sorry - I don't think I reacted violently. T'was not my intention.

I mad an error in my math by skipping a step.

>*** Sorry as I mentioned I haven't had time to check that. This is a hot
>topic and going over someones maths is time consuming and it was on my
>agenda to work on.



>Assuming all 1500 rank players in the game, the number of points awarded for
>a 14 v. 11 game is 45 + [(11*1500 - 14*1500)/150]. I forgot the 45+. I
>calculated a change of 30, when in fact, you would only get a change in
>15.....
>
>So, while gaining ranking for lots of very short, unbalanced games, the
>gains would come a lot slower than I'd previously calculated.
>
>Still, the rankings DO reward victories over lesser opponents. They reward
>lots of games vs. long games against great opponents.

*** Why are long games more important? Long games generally mean DS
victories so one would be biasing the score in favour of DS wins over FP
wins in such circumstances. I agree that game endings is a major factor of
this. So the more you play the more opportunity to alter your score you
have. I don't see that as being necessarily a bad thing. (Professional
sportspeople put more effort into a game - ie play more - than amateurs and
the "rewards" are greater for it.)

>I'd much prefer to see no ranking system over a ranking system. I'd prefer
>a ranking system that ranked a player that is 10 of 10 over a player that is
>25 of 40.

How to do the second?

>I'd prefer a ranking system that truely ranked the skill of a player, over
>one based on wins and losses. I've done poorly in games and won, then had
>games where I personally played well, but lost. I have no idea of how to
>design the ideal (or even a good) ranking system, and thus I prefer not
>having one over having one.


*** How to do that allowing for what I think will happen in that players
won't vote. (From past experience). Note bringing attention to the fact
that players won't vote will certainly motivate some to do so - but only
for a short period of time (unless they get into the habit of doing so. We
had a long haul to get players to put their account number with their
emails - most of the time they do now, but still...)

BTW thanks for the input it's help clarify a number of points in my mind
and helped a lot to sort out some niggly bits. We have our disagreements
on many aspects of this but I have found your input invaluable.

Clint

[Middle Earth PBM Games]

>
>With these ratings, I agree that the player who is 6/8 against 1600
>point opposition should have a lower rating than a player who goes 4/4,
>but only slightly. And the 4/4 player should never have much more than a
>1600 rating himself, since he only plays that level of opposition.

We can modify the equation so that there is more bias to the scores between
players over that of the fixed points score of Win/Loss. (At present it's,
for Valar, 45 for a win modified by a variant score depending on the number
of nations on your side, Ratings of players on both sides with a
normalising factor. We can change that normalising factor (of 150) to
something else if you feel that would be more appropriate. What would that be?


>I agree that players should be penalized for dropping (either in
>ratings, or in a drop count).

*** Okay what score? The score of a factor x the Eventual Losing
score? We could do that? Say 50% of the Losing score? (Eg If I drop the
game then I lose 20 points as the Losing team loses 40 points when the
numbers are worked out?)

I quite like that.

Clint