GR Calculator

[TemplarsBane]

We all love seeing the points go big, but the real measure of skill on this site is Ghost Rating. Sometimes I wanted to know how much GR I could gain or lose if I took over a position for someone else. Or how much this press variant or that map might get me. How much more valuable is a 3 player draw than a 4 player draw in this specific game?

I noticed myself doing some napkin math over and over and wanted to create a calculator where I could just change some variables and it would spit out the answer I needed. So I made one!

Currently there is only functionality for DSS and not for SoS, but it could be added. Feel free to take this, make a copy, use it for yourself. If you notice some errors please let me know and I'll correct them! Enjoy!

https://docs.google.com/spreadsheets/d/ ... sp=sharing

[sweetandcool]

Does taking over a position affect your ghost rating less than playing a full game?

[TemplarsBane]

Not to my knowledge. But I'm also not a site admin or anything. I have no behind the scenes knowledge of how any of the formula works beyond what's posted publicly. I'd assume that taking over a position is the same as starting a game from the beginning.

[sweetandcool]

Good to know. I thought taking over positions was a nice thing to do, to let a game continue naturally, but now I know that I'm actually just tossing away my rating points

[TemplarsBane]

Good to know. I thought taking over positions was a nice thing to do, to let a game continue naturally, but now I know that I'm actually just tossing away my rating points

Only if you lose! Good luck lol

[A_Tin_Can]

Does taking over a position affect your ghost rating less than playing a full game?

No. One of the problems with the system is that it only counts the players at the end of the game. This is a pretty hard problem to solve for a general case without allowing exploits.

I've posted about this before, but personally I wouldn't put much stock in GR. Not because of the practical problems, but because the formula is pretty wonky.

If you're interested in looking into it, have a look at what happens to the rating in a solo win in a game of highly rated peers vs what happens in a solo win of a game of low rated peers. Similar problems exist when you have one very highly rated player playing with low rated players.

(For rating system nerds familiar with Elo-variants: the problem is because GR uses your rating inappropriately in the K factor, and then again in the same way in the expected outcome - which means they cancel out, and you don't get the properties you're supposed to get in a rating system - there's a post I made in the old forum somewhere with much more details and examples).

There are other multiplayer rating systems that would be more appropriate for Diplomacy.

People have a big emotional attachment to GR though. FWIW, it's definitely better than points.

[sweetandcool]

Does taking over a position affect your ghost rating less than playing a full game?

No. One of the problems with the system is that it only counts the players at the end of the game. This is a pretty hard problem to solve for a general case without allowing exploits.

I've posted about this before, but personally I wouldn't put much stock in GR. Not because of the practical problems, but because the formula is pretty wonky.

If you're interested in looking into it, have a look at what happens to the rating in a solo win in a game of highly rated peers vs what happens in a solo win of a game of low rated peers. Similar problems exist when you have one very highly rated player playing with low rated players.

(For rating system nerds familiar with Elo-variants: the problem is because GR uses your rating inappropriately in the K factor, and then again in the same way in the expected outcome - which means they cancel out, and you don't get the properties you're supposed to get in a rating system - there's a post I made in the old forum somewhere with much more details and examples).

There are other multiplayer rating systems that would be more appropriate for Diplomacy.

People have a big emotional attachment to GR though. FWIW, it's definitely better than points.

As a chess player, I am very familiar with the ELO system. I assumed GR was more or less the same, but had a strong suspicion that something about it was "off". Do other websites use the same Ghost Rating system, or something else?

Personally, I think filling in for a game should make it treated as unranked for the substituting player, or at least make it affect the player's rating less. Or perhaps rate the player depending on some sort of criteria like "played 80% or more of the game".

[JECE]

It's the same thing with "Game Stats" like # and % of wins and losses.

You used to be able to track how well you did after joining games in progress with the points system (back when you had to bet to take over positions and PPSC was a thing), but even then I only was able to keep track manually:
https://webdiplomacy.net/userprofile.php?userID=17421