Re: POWER rankings, new addition ?
Posted:
Aug 13, 2014, 12:48 AM
well, chef man, I would have no idea, its just an idea I had. that's for rainwolf or you to figure out. It would be cool if a power rating could be next to the points rating. something like a 50 point system, from 0 to 50 ! some times a player has an 1800 rating, but when you look at that players completed games, all they have been playing is 1200 rated players. It would seem that you look at how many games they have played, and how week or strong there points rating is ?
Posts:
435
Registered:
Jan 23, 2009
From:
north carolina
Age:
48
Re: POWER rankings, new addition ?
Posted:
Aug 13, 2014, 3:44 AM
well for me... i like players seeing me as green a power rating may skew that tho i assume that a power rating can be manipulated as the rating system so i will be a lowly green w/ a low power rating.. like kermit says... it aint easy being green... but its me"
Re: POWER rankings, new addition ?
Posted:
Aug 13, 2014, 8:45 AM
whats wrong the existing rating system, it seems similar Elo rating system used by many
from wikipedia "The Elo rating system is a method for calculating the relative skill levels of players in competitor-versus-competitor games such as chess. It is named after its creator Arpad Elo, a Hungarian-born American physics professor. The Elo system was invented as an improved chess rating system and is also used in many other games. It is also used as a rating system for multiplayer competition in a number of video games,[1] and has been adapted to team sports including soccer (association football), American college football, basketball, Major League Baseball, competitive programming, and esports.
Message was edited by: madmike at Aug 13, 2014 11:21 AM
Posts:
2,220
Registered:
Mar 4, 2007
From:
San Francisco
Age:
45 Home page
Re: POWER rankings, new addition ?
Posted:
Aug 14, 2014, 12:50 AM
this post is not to boost my ego, but merely to use my self as example to answer your question. And maybe I'm not answering your question, but merely offering data to consider when posing such a question.
My win lose ratio i think is around 58% on record.
1- That data is from all of my rated games.
2- Most of those rated games were from before I was any good at the game.
3- Of all my games both rated and unrated, most of my games are unrated, because many players decline to play me rated, or what ever reason.
4- If you take only the games I played from when I hit around a rank of 1800 or so, and include from that point forward both all of my rated games and all of my unrated games,... my ratio will say i win over 95% of the time.
So,.. am i a 58% player? or a 95% player,... or what..
Thanks for reading,
~Z
Scire hostis animum - Intelligere ludum - Nosce te ipsum - Prima moventur conciliat - Nolite errare
Posts:
435
Registered:
Jan 23, 2009
From:
north carolina
Age:
48
Re: POWER rankings, new addition ?
Posted:
Aug 14, 2014, 6:44 AM
when i imagine it its like a strength bar and a stamina bar in many video games. personally i think that would be cool! have no idea how to rate such a thing... but as z said previously... mebbe a job for jayhawklov (sounds a lot like 'this is a job for superman!;lol) i even think the 'power bars' would be cool as in opposed to colored squares mebbe a meter that is colored? i know im getting carried away but fun to consider ...
Posts:
435
Registered:
Jan 23, 2009
From:
north carolina
Age:
48
Re: POWER rankings, new addition ?
Posted:
Aug 14, 2014, 7:06 AM
question tho... would unrated players be subject to the terms of the power ratings? i know many excellent players who play only unrated and prefer to keep it that way wouldnt that in a sense be 'outing' them? on the other hand if they were not included then im not sure how accurate the 'power rating' would be just a thought (and yes... it hurts when i get them...not used to it...lol)
Posts:
66
Registered:
Dec 13, 2008
From:
New York City
Age:
58
Re: POWER rankings, new addition ?
Posted:
Aug 15, 2014, 2:52 AM
Might be an interesting idea to have this in a box below or near the traditional Elo rating. It may result in more rated game challenges for higher ranked players.
It should not be difficult to calculate the average (mean) rating of each players' opponents in rated games only.
As per zoeyk's point about games very long ago not really reflecting a players' current form and level of skill, games more than 5 years ago, or some such cutoff date should be omitted.
Re: POWER rankings, new addition ?
Posted:
Aug 16, 2014, 6:41 PM
Let's try to come up with a good metric to represent what you want it to represent, which is not yet clear to me. The current rating metric has the property of giving out big rewards if the rating gap between winner and loser is big and if the winner was lower rated, and vice versa. This in itself, to me at least, makes current ratings a pretty good reflection of a player's current skill, ... if he's willing to play anyone.
There's also the winning streak, but that does not account for the ratings of both players.
My question is, what would you specifically like to measure? - A winning percentage in the past 50 games? (But that does not reflect ratings) - Anything that reflects ratings has to deal with the following: when a much lower ranked player loses to a higher ranked player, the changes in rating are small, but the changes in rating are also small when the difference in ratings is small too, regardless of who wins or loses then. It's not so clear to me how to deal with that.
@xtraclassy: if I understand you correctly, you'd calculate the mean rating of (the last 50 or so) opponents, and compare that to a player's current rating. If that mean is lower, then that power bar would be closer to red (red = bad in video games, unlike here on pente.org ), if it's much higher, then it's closer to green, and if it's close to his current rating, it would be somewhere in the middle. Is that what you had in mind? - That would seem to encourage players to player higher rated players, but at the same time discourage players to play lower rated players, I see the benefit in the former, but the latter not so much.
I may be misunderstanding though. So my question to anyone interested, what is it you'd like this new metric to represent? Layman's terms or something conceptual is perfectly fine, let's put our brains together to figure something out.
Posts:
66
Registered:
Dec 13, 2008
From:
New York City
Age:
58
Re: POWER rankings, new addition ?
Posted:
Aug 17, 2014, 2:49 AM
Hi Rainwolf,
I have no problem with the current rating system whatsoever, and have no desire to see it go, but I do understand the "strength of schedule" argument made in this thread.
The current rating system would remain untouched and unchanged, and still show the true number one player and champion since it is quite possible that the player with the toughest schedule may have lost all the games played against higher ranked players. Note also that it cannot exist without the current system since it is computed from opponents ranking under the current system.
Note that this power ranking would have nothing to do with a players' color.
This is not a high priority, in my opinion, but would be an interesting metric alongside the current top 10.
Regards,
PS If you are implementing it, 50 games may be a rather small sample size. A player's last 100 games may reveal a bit more.
Posts:
2,220
Registered:
Mar 4, 2007
From:
San Francisco
Age:
45 Home page
Re: POWER rankings, new addition ?
Posted:
Aug 17, 2014, 4:08 AM
i agree with Clyde/xtraclassy.
Ok just for fun I'll throw out an idea here with out over thinking it.
Criteria:
1- player has played at least 500 games of rated and unrated combined.
2- the 500 games of rated and unrated combined must have been completed within the last 2 years
3- of those 500 games of rated and unrated combined use only games vs players with a ranking of 1700 or greater
4- the number of those games must be at least 100 or greater. if more than 100, then of those only use the most recent 100 and disregard the games prior to that.
5- if the ratio of P1 to P2 is uneven, then remove the uneven remainder, and include same number of opposite seat games from the omitted list in order to bring the ratio to even. if this is not possible within the criteria due to lack of games then the criteria is not met.
4- of those games divide into groups of player was P1 and player was P2
5- of the P1 group, for every win, give 1 point. then for every loss remove 2 points
6- of the P2 group, for every win, give 3 points. then for every loss remove zero points.
7- add up all points and use that yield to determine rating. perhaps assuming that average is all P1's win and all P2's lose then the average number to compare the players yield to would be 50.
lowest yield would be -100 , and yes i do mean negative 100 below zero.
highest yield would be 200
so low= -100, middle= 50, high= 200
which if low were to start at zero then would translate to:
low= 0, middle= 150, high= 300
not sure if any one will understand what i just did there
~z
Scire hostis animum - Intelligere ludum - Nosce te ipsum - Prima moventur conciliat - Nolite errare
Posts:
2,220
Registered:
Mar 4, 2007
From:
San Francisco
Age:
45 Home page
Re: POWER rankings, new addition ?
Posted:
Aug 17, 2014, 4:29 AM
one of the main things to walk away with in my proposal is this, what the value of P1 wins/loses compared to P2 wins/loses could be considered. which i have clearly explained my formula for above.
from there perhaps to make it more complex to become more accurate, is consider adjusting those values depending on the rating of the other player playing,.. but not sure if that would be good or bad to tamper with, because how accurate is their rating any ways right? There simply is no way to perfectly do this... it is only for fun.
the most accurate way is for a "HUMAN MASTER" to review the player using human thought. even that will not be perfect, but it will be far more reliable than any formula you will ever come up with, in my humble opinion.
Scire hostis animum - Intelligere ludum - Nosce te ipsum - Prima moventur conciliat - Nolite errare
