Chris
United States
New Jersey
-
Does anybody know of a good method for scoring in a long term power grid tournament? I would want each player to have a "rating" but I am not quite sure the best way to calculate said rating. Thanks!
-
Moshe Callen
Israel
Jerusalem
I like to exchange ideas but I have no interest in a pissing contest.
If you want me to review your game, just GM me and send me a copy. Abstracts, wargames and euros equally welcome. No party or dexterity games please.
-
TheRakeman wrote: Does anybody know of a good method for scoring in a long term power grid tournament? I would want each player to have a "rating" but I am not quite sure the best way to calculate said rating. Thanks!
My off the cuff proposal would be a points system as follows:
1. Winner gets +50 pts. Ties split these points if not broken.
2. +5 points for the number of cities powered in the last round
3. +1 for each elektro left over
-
W M Shubert
United States
Portland
Oregon
KGS is the #1 web site for playing go over the internet. Visit now!
Yes, I really am that awesome.
-
whac3 wrote: TheRakeman wrote: Does anybody know of a good method for scoring in a long term power grid tournament? I would want each player to have a "rating" but I am not quite sure the best way to calculate said rating. Thanks! My off the cuff proposal would be a points system as follows: 1. Winner gets +50 pts. Ties split these points if not broken. 2. +5 points for the number of cities powered in the last round 3. +1 for each elektro left over
I'd do this:
Winner: +50
2nd place: +25
3rd: +20
4th: +15
etc.
That way you are rewarded only for beating other players. If you manage to do it with 1 city (to use an incredibly unlikely hypothetical), you get the same reward as doing it with 21 cities.
-
David Corbin
United States
Cumming
Georgia
-
TheRakeman wrote: Does anybody know of a good method for scoring in a long term power grid tournament? I would want each player to have a "rating" but I am not quite sure the best way to calculate said rating. Thanks!
Not quite what you're asking for but perhaps it will benefit you. Here is the site for my Power Grid League http://www.dcorbin.com/pgl . The League Rules explains the how rankings will be determined.
-
Chris
United States
New Jersey
-
dcorbin13 wrote: TheRakeman wrote: Does anybody know of a good method for scoring in a long term power grid tournament? I would want each player to have a "rating" but I am not quite sure the best way to calculate said rating. Thanks! Not quite what you're asking for but perhaps it will benefit you. Here is the site for my Power Grid League http://www.dcorbin.com/pgl . The League Rules explains the how rankings will be determined.
That's very similar to the kind of thing I'm looking for, in fact what I'm doing is more of a "league" than a tournament I suppose so this might come in handy. However I don't like the idea of going solely by win percentage because mine isn't going to last quite as many games as this one is.
-
United States
Norwood
Massachusetts
-
Ratings on games usually use the Elo System. Google is your friend.
-
Chris
United States
New Jersey
-
markgravitygood wrote: Ratings on games usually use the Elo System. Google is your friend.
I looked up the Elo system as my original idea but I'm not sure how well it works for multiple players?
Also, since this isn't a long term tournament, should there just be a static arbitrary K Value?
-
United States
Norwood
Massachusetts
-
Yes, problematic. You could adjust the K-factor, for example, in a 5-player game, where "third" nets zero points, 2nd hay have K1 and 1st K2, with 4th and 5th negative and opposite.
It may not work, and probably too complicated.
-
juan
Argentina
Ciudad Autónoma de Buenos Aires
-
If you want to start having a record of your plays, i recomend you use Access data base. I use it for almost all my games. If you are intrested send me an inbox
-
Randall Bart
United States
Granada Hills
California
Red October
Earth is one of my favorite planets
-
TheRakeman wrote: Does anybody know of a good method for scoring in a long term power grid tournament? I would want each player to have a "rating" but I am not quite sure the best way to calculate said rating. Thanks!
IMO Power Grid is not well suited to such tournaments. Whatever system you choose will alter the endgame, and players need to know this.
-
Chris
United States
New Jersey
-
Barticus88 wrote: TheRakeman wrote: Does anybody know of a good method for scoring in a long term power grid tournament? I would want each player to have a "rating" but I am not quite sure the best way to calculate said rating. Thanks! IMO Power Grid is not well suited to such tournaments. Whatever system you choose will alter the endgame, and players need to know this.
As this is a pretty casual tournament setting, I'm not sure anybody will care in the least - but this is a good point and something I am trying to take into consideration at least. In just about any multiplayer game I imagine games will play out very differently when every game is not in its own little vacuum.
-
Russ Williams
Poland
Wrocław
Dolny Śląsk
-
What my old group used to use:
http://russcon.org/RussCon/ratings.html
Like wmshub, I believe margin of victory etc should be irrelevant. The only relevant things are how many people I beat and how many people beat me.
-
Jamie Jensen
United States
Boise
Idaho
-
It really depends on what you mean by "long term". Ideally*, since Power Grid was designed as a "winner take all" game where 1 player wins and everybody else loses (there are no 2nd, 3rd, etc. places), you would assign points only to the winner of the game using something like Microsoft's TrueSkill (there's a C# implementation accompanied by a great article here) and after perhaps a dozen rounds (depending on how many players you have) you'd have a really good idea of who the best players are. However, if you want a system with decent accuracy over fewer numbers of games, you might end up having to take into account the ranks of everyone else who didn't finish in 1st, which as Barticus88 rightly notes this will affect people's strategies in the game (I don't mind this per se, but it does make player strategy more complicated since maximizing one's expected number of points means taking into account how each move affects everyone's chances to end up in 1st, 2nd, etc. rather than just understanding its effects on finishing 1st). Also, as Russ said, I wouldn't worry about margin of victory when calculating points (I wouldn't say it's irrelevant, but this just adds a not insignificant amount of complexity to considerations when trying to maximize one's points while adding little in the way of an useful payoff since how exactly this maps to the degree of difference in skill is anything but obvious).
If you're sold on the idea of TrueSkill (or at least some other ELO-like system), you can skip the rest of this. If you're willing to sacrifice accuracy for simplicity, then I have an alternative. Russ' system above may very well be better than what I'm about to suggest, it's been awhile since I've read over it so I'm not certain. His system is more complex, taking into account the uncertainty involved when players play differing numbers of games IIRC, whereas the below does not, though I suspect this won't matter for your purposes. I remember thinking I was somehow dissatisfied in how his system aggregated scores for games played with differing numbers of players, but the following is also flawed in this regard and I stopped being interested in how to fix it once I discovered TrueSkill. Anyway, here you go:
Assuming no ties are possible (the World Boardgaming Championships add a 3rd tiebreaker, highest-numbered power plant, to ensure this), if the same number of Players are always in each game then I'd just assign points to each Rank equal to Players - Rank. Because the number of players per game might vary, points can be normalized by dividing them by 1 less than the number of players, i.e. (Players - Rank) / (Players - 1). If you don't add that third tiebreaker and instead allow for the possibility of ties, tied-players' ranks should be counted as equal to the average of the ranks they would have had if they'd been ranked sequentially, e.g. two players "tied for 1st" would actually have a rank of 1.5.
*Actually, I think the best algorithm is probably TrueSkill Through Time, but the only implementation of that I'm aware of uses F#.
-
|
|
