webdiplomacy.net

GR2 can be viewed here.

Question for you guys - if the link is for view only, can you still use the filters?

Original post can on GR2 can be found here

Tweaks made in this version:

• 1v1 added - all 1v1 games are considered ranked

• V was changed from 50*A*P*R to 75*A*P*R. This was done to increase the range of ratings. This pushes the top players closer to 2000

• For the GB only list, GB games are no longer discounted

• For the 1v1 only list, press does not affect the weighting

I had to faff a bit to use the filters, and as it looked like Zultar was both 1st and 11th in the overall rankings when I finished I reckon I did it wrong .

Nope, looks like you did it right. Looks like a copy+paste error from when I added the new data. I'll redo it quickly...

Should be fixed now, I think.

...and I had forgot to actually import the new data. Can we all just forget how many mistakes I've made here?

No worries. Your efforts are much appreciated :)

For that, I've made sure you beat VI in both overall and classic. Though it looks like the two of you are pulling away in classic. Neck-and-neck but with a decent lead over ghug in 3rd.

I still don't have any statistics to prove whether GR or GR2 is better. Is it possible to get at least 1,000 games worth of expected results vs actual results?

Is there a reason VI isn't listed as active in classic?

@ghug, He is listed as active now. I had fucked up uploading things but it should be correct now.

@chippe, correct - nobody has run that analysis. I can send you the raw data and the GR2 code if you would like but I don't have time to do it myself right now.

ghug wrote: ↑

Sat Mar 03, 2018 4:53 pm

Is there a reason VI isn't listed as active in classic?

What are you looking at? He comes up as active when I open it.

Yonni wrote: ↑

Sat Mar 03, 2018 5:06 pm

@ghug, He is listed as active now. I had fucked up uploading things but it should be correct now.

@chippe, correct - nobody has run that analysis. I can send you the raw data and the GR2 code if you would like but I don't have time to do it myself right now.

Send me the raw data than. Just wondering, but if I'm getting the raw data, how exactly are you scoring it for ELO purposes?

If you get me the raw data I'll see if ghug can set me up with the GR raw data.

GR can actually be configured to spit out a massive debug file. I'm not sure if it'll have what you need, but I can email it to you if you want.

Yonni wrote: ↑

Sat Mar 03, 2018 12:42 pm

• V was changed from 50*A*P*R to 75*A*P*R. This was done to increase the range of ratings. This pushes the top players closer to 2000

• For the GB only list, GB games are no longer discounted

Those changes seemed to help my rank position a lot(!), for some reason, Yonni. Check's "in the mail".

Yonni, would you be able to add columns to show "changes" from prior month, like GR1 does? Maybe it doesn't make sense for change from Feb to Mar since you changed the formulas, but I think most people would love to see how they are rising or falling, month to month. Alternatively, could you save a document for each month, so we can go back to the prior month doc and compare?

We're some players dropped from the Feb version? (Can't find "serginss" in classic for example??)

Yeah, I should keep track monthly. I'll need to figure out the easiest way to do that. Maybe just make a new sheet every time.

Are you sure serginss is a user? Don't see that name anywhere

So it seems like 1v1 completely dominates these rankings, which imo decreases their meaningfulness, but that's your call. Also on the subject of 1v1 though is that your 1v1 ratings seem to differ from the straight Elo, which is odd to me, since theoretically this system should reduce to Elo in a 2-player game, right?

ghug wrote: ↑

Sat Mar 03, 2018 9:53 pm

So it seems like 1v1 completely dominates these rankings, which imo decreases their meaningfulness, but that's your call. Also on the subject of 1v1 though is that your 1v1 ratings seem to differ from the straight Elo, which is odd to me, since theoretically this system should reduce to Elo in a 2-player game, right?

I'd pretty strongly agree that 1v1 probably shouldn't be in the rankings.

ghug wrote: ↑

Sat Mar 03, 2018 9:53 pm

So it seems like 1v1 completely dominates these rankings, which imo decreases their meaningfulness, but that's your call.

Happy to go with the flow on this one. I'm happy to remove it or reduce it even further from a 0.25 weighting to 0.1 or even 0.05 if that's better.

ghug wrote: ↑

Sat Mar 03, 2018 9:53 pm

Also on the subject of 1v1 though is that your 1v1 ratings seem to differ from the straight Elo, which is odd to me, since theoretically this system should reduce to Elo in a 2-player game, right?

There are some changes from the 1v1 Elo ratings like weighting for games played and provisional ratings.

Yonni wrote: ↑

Sat Mar 03, 2018 10:23 pm

Happy to go with the flow on this one. I'm happy to remove it or reduce it even further from a 0.25 weighting to 0.1 or even 0.05 if that's better.

Why not put it in a separate rating? I think it's very much a different game.

1v1 is a large part of the site now. If you want a site ranking I don't see why you wouldn't include it.