I'm with Lando that a non-live gunboat list would rock.

@Cachimbo, I'm not on the monthly one either.

I'm not on the monthly breakouts either. I think if you didn't have a game finish last month, you didn't get on the breakout lists at all.

*spit take*

@CS: You are right, the gunboat list shouldn't have the results quarter-weighted. This would make gunboat ratings and full press ratings comparable.

However, unless I currently fail at maths, this shouldn't have any effect on the convergence of scores or the ranking of players, no?

@Alderian: Do live-games get the same weight as non-live games in the full GR list?

WTA, Classic Map, Full Press, Non-Live:
Justice has been done.

Ok, I'll try to answer as many of these at the same time as I can since I've worked with Alderian on this and deciding which ones to calculate.

There is NO WEIGHTING when it comes to specific ratings. Moreover, the convergence there is as fast as the convergence for the regular overall ghostrating, which is based on ELO. Ghost is in the process of working on a different rating that would converge faster, but this will be a long time before it's done.

Live games are weighted the same as non-live games.

When I did the regression analysis for the different scores, there was a very high correlation between gunboat scores, meaning your rating for all of them are very much related or comparable. For example, the correlation between wta-gb and all gb games were 96%, meaning that if you know one of those two scores, you can accurate predict the other scores. In real life situations, this is very rare as a 60-70% correlation is considered to be highly correlated. So to skip through the math mumbo-jumbo, if you know one score, then you pretty much know the other. Your relative skill is not going to change much at all by varying the types of gunboat games you play.

After the top 783 the quality seems to drop drastically

@zultar: Alderian has stated above that the gunboat results are still quarter weighted for the gunboat-only list which, in case it is true, would make gunboat and full press ratings uncomparable. Objections?

Nigee, I think there are still some competitors as far back as 1018.

if you're not in the top 1000 talk to the hand :-)

Oh, I did not realize that he still had that on. Ooops. My apology. Smile sheepishly.

Geez, those separated stats really emphasize how my wins come nearly exclusively from live games.

Hello, Hand, I would like to beseech you to recognize that 1018 a quite magnificent ranking.

I was looking at the new WTA-Classic-FullPress-Non-Live-GhostRatings and found it odd that it had (3) Survives for me, since I know for a fact I looked into this in the past few months and I only have (1) Survive of that sort, so I was wondering what was wrong with these new rankings... Turns out I played two WTA gunboat games before that was a feature on this site, back when gunboats were played on the honor system, so that explains that...

I thought these GR ratings were rubbish but as I'm 103 Gunboater for May I now know they are brilliant

Keep in mind that the Live, Gunboat, and WTA lists all have different numbers of active players, so being 200 in one and 300 in another doesn't mean you are better at the first if the competition is stiffer in the second. Maybe in addition to a rank number there should be a percentile as well.

@Cachimbo, yep, three months of no activity and you drop off the lists. But you'll still show up on the all time lists.

Is there a reason why the all-time list and the monthly list show a different number of games (but the same number of points)?

nm, I'm an idiot.

Hmm. Now I know how much playing gunboat hurts my GR... Why do I stink at gunboats?? Oh well, its not like I'll stop playing them any time soon.

I do have a real question though. Can you not drop below 100 GR? I'm shown as having 100 even in the GB list with now decimals. Seems odd.

Awesome, though i do seem to be missing from the may list (at least i'm in the top 100 of the all-time 'classic' list)

@Yonni, you are not in the Gunboat May '12 list because you haven't played a gunboat game in a few months. In the Gunboat All Time list you have a Peak ghost rating of 100 because that is your peak rating, you've never gone above 100.

Peak never drops, it only goes up when your current rating is > than your old peak rating.

@orathaic, on the all-time 'classic' list, it shows your last game date as 12/11/2011. That is the last time you completed a game of that category, so you do not show up in the May list because that is more than 3 months ago.

Ah, of course your peak can't be lower than 100... And, again, I'm an idiot. Also terrible at GB evidently..

No worries. I'll be updating this page with a FAQ section now that we've got some frequently asked questions. :)

lies, i've lost gameID=77257 since then... but it's not over yet.

@ora - you know damn well the game doesn't show up on GR until it is over. Don't be a putz.

Fun stat of the day.
The Czech has played 886 live games. Let's say, on average, he plays for 2 hours for each game. That 1772 hours, or 74 DAYS of playing diplomacy online. Holy Christ this game is addictive.

@Mr A: The reason is has an effect on convergence is that you can think of processing each game as calculating a GR for that game alone, and then your new GR is, say, 93% your old GR and 7% your new one. Or, to put it another way, if in all of your games, you behave like a GR 200 player, and you start at GR 100, then it takes ~15 games to get a decent fraction of the way to 200.

For gunboat, instead of 93%/7%, it's more like 98%/2%. So it takes several times as many games in which you behave as a GR 200 players to trend to 200.

This is the reason I was noting that playing more games pushes your GR higher; with the current low weighting, there are a lot more people who play 200+ GR gunboat than who have had enough games to push their rating close to the value it will stabilize at.