Ticket to Ride New York Ticket to Ride New York

Forums

Search
Forums » T2R Competitive Play - English » Tournament ELO: Update Dec 28th- 2p USA after NC 2016
Show: Today's Posts 
  
AuthorTopic
RUS Docent CMEPTb
Senior Member
Romanov Train

User Pages
Posts: 413
Registered:
March 2011
Re:Tournament ELO: Update Dec 28th- 2p USA after NC 2016 Mon, 15 May 2017 18:37
In both cases Champion should win all matches Very Happy
So I think no need to reseed, let's keep it as is.
      
Truckerteller
Senior Member
Conductor Level 1

User Pages
Posts: 947
Registered:
October 2007
Re:Tournament ELO: Update Dec 28th- 2p USA after NC 2016 Mon, 15 May 2017 19:46
The amount of skill that goes into a game of chess is much higher than a game of ttr. Or to put it better, one game of ttr is far more reliant on chance than a game of chess. Another reason we chose match telo. 6,7 or 9 games will give a match result similar to chess. So k=30 is fine for match telo and consistent with most of the previous posts in this forum.

For game telo k=8-10 might be ok. If you search the forum, you'll find that drake initially did game telo. He wrote the vba code for Excel sysy uses.

As for new players, everyone started at the same 1600 back in 2006.
For tournament seeding players with less than x games played get a deduction per game under x. That's been done for league at least.
      
VojFromMars
Senior Member

Posts: 164
Registered:
January 2016
Re:Tournament ELO: Update Dec 28th- 2p USA after NC 2016 Mon, 15 May 2017 20:04
CIA Truckerteller écrit le Mon, 15 May 2017 19:46

The amount of skill that goes into a game of chess is much higher than a game of ttr. Or to put it better, one game of ttr is far more reliant on chance than a game of chess. Another reason we chose match telo. 6,7 or 9 games will give a match result similar to chess. So k=30 is fine for match telo and consistent with most of the previous posts in this forum.

For game telo k=8-10 might be ok. If you search the forum, you'll find that drake initially did game telo. He wrote the vba code for Excel sysy uses.

As for new players, everyone started at the same 1600 back in 2006.
For tournament seeding players with less than x games played get a deduction per game under x. That's been done for league at least.


Ok I'll run a match Telo with k=30. Ok for seeding with this results ?

Obviously when your rating system is brand new you have to start somewhere. So every body had to start out soemwhere and that had to be even.

Using performance as a first rater when your rating system is established is a no-brainer. If a very strong or a very weak opponent starts at 1600 he will obviously unbalance others players ratings. And this can be avoided, and is not an injustice for new players (it's rather fairer tbh).

Ok with this seeding ? If so I start the topic today. If not, epxect it for tomorrow... If no answer I'll open it in a couple of hours with this temporary seeding...
      
FLOP Hecki
Senior Member

User Pages
Posts: 1088
Registered:
October 2005
Re:Tournament ELO: Update Dec 28th- 2p USA after NC 2016 Tue, 16 May 2017 01:01
Intresting stuff. But 2nd place in Div A = place 49 in your TELO list. You're really sure that this makes sense? Smile

But i won't care as long as i can still compete in the tournaments. Smile
      
VojFromMars
Senior Member

Posts: 164
Registered:
January 2016
Re:Tournament ELO: Update Dec 28th- 2p USA after NC 2016 Tue, 16 May 2017 01:26
Here are the EU calculations for EU using match calculations and k=30 :

1 Truckerteller 1931
2 Sivorro 1853
3 Morientes ZL 1839
4 nicolas21 1826
5 SuperPello 1755
6 Marschall Vorw 1746
7 CMEPTb 1728
8 yaelka 1721
9 VojFromMars 1717
10 335d 1715
11 Sysyphus 1710
12 Hecki 1708
13 Megasiv 1705
14 Val 1679
15 Lexx 1678
16 Patterson 1673
17 cannonade 1656
18 GSV3 1646
19 RMarkes 1643
20 xbomanx 1639
21 Stormwarrior 1633
22 IL2T 1628
23 toddcalu 1617
24 Enix - Poulsard 1609
25 Alautre 1605
26 Will 1603
27 allan 1601
28 Alexkey 1599
29 thesteamwistle 1598
30 vigulisam 1594
31 The Tinman 1593
32 benny 1590
33 Lucullupus 1589
34 michelecannone 1587
35 mikola 1586
36 Marviktori 1585
37 TheBendite 1584
38 chekist13 1584
39 Knockando 1583
40 Cromze 1583
41 dandee 1580
42 DrakeStorm 1580
43 lirex 1579
44 Diplodokus 1574
45 dizz 1571
46 PopeRoma 1570 *
47 The Man Without 1568
48 Pammes 1567
49 DN 1565
50 Elric 1559
51 OllieB 1556 *
52 dea1 1554
53 Julichka 1551
54 Ironhorse 1545
55 Doc 1542
56 belelule 1536
57 Qorlas 1535
58 Loca 1531
59 Ukky 1527
60 Dgèdge34 1523
61 onyx puffin 1518
62 MaximouR 1510
63 Photios 1493
64 LeGoupil 1488
65 tiploup 1488
66 Cat 1487
67 Tyrana 1463
68 JenAck 1455
69 Tempest0 1425
70 Beefy 1422
71 Aurelie in wonderland 1365 *

As you see ... it's more violent than the previous one

Just to make it clear I copy/paste the Python code used at the end of the message...

There are only four options possible if you find it too strange :

1/ we've still not understood each others regarding what formula should be used
2/ my code isn't ok ; Python is pretty easy to read and you shall be able to pinpoint my mistakes
3/ I don't have the right datas
4/ there was a mistake in the code formerly used.


# -*- coding: utf-8 -*-
"""
TELO Version 0.1

@author: Vojislav Petrov 

Date : 13 May 2017

Coded in Python 3.4
"""

import numpy as np

def proba_win(EloA,EloB):
    """returns probability of a win for player A"""
    return 1/(1+10**((EloB-EloA)/400))

def variation_Elo(EloA,EloB,k,score):
    """returns variation of elo for player A who scored score ; same k"""
    return k*(score - proba_win(EloA,EloB))

def parse_former_Telo(filename):
    """parses former Telo
    returns a list of lists : player,name,ELO
    a list of players with provisionnal Telo"""
    f=open(filename)
    s=f.read()
    f.close()
    T=s.split('\n')
    players=[]
    Provisionnal=[]
    for line in T:
        if line!='':
            s=line.split('\t')
            players.append([s[1].strip(' '),int(s[2])])
            if len(s)>3 and s[3]=='*':
                Provisionnal.append(s[1].strip(' '))
    return players,Provisionnal

def parse_results(filename):
    """parses results of usual form ; e.g. those posted by Docent 
    returns a list of games 
    a game is a list [Player A ; Player B ; Score A ; Score B]"""
    f=open(filename)
    s=f.read()
    f.close()
    T=s.split('\n')
    results=[]
    for line in T:
        if line!='':
            s=line.split('-')
            results.append([s[1].strip(' '),s[2].strip(' '),float(s[3]),float(s[4])])
    return results

def parse_results_2(filename):
    """ Parse my results ; similar to Docent form ; variation on v1"""
    f=open(filename)
    s=f.read()
    f.close()
    T=s.split('\n')
    results=[]
    for line in T:
        if line!='':
            s=line.split('-')
            results.append([s[0].strip(' '),s[1].strip(' '),float(s[2]),float(s[3])])
    return results

def get_id(players,P):
    """returns the id of a player P or returns None if he doesn't exist
    takes account of the fact those damn players don't care of uppercase"""
    for i in range(len(players)):
        if players[i][0].upper()==P.upper():
            return i

def updates_ELO(players,Provisionnal,game,k=30):
    """ updates players ELO and Provisionnal list after a game"""
    if get_id(players,game[0]) is None:
        players.append([game[0],1600])
        Provisionnal.append(game[0])
    a=get_id(players,game[0])
    if get_id(players,game[1]) is None:
        players.append([game[1],1600])
        Provisionnal.append(game[1])
    b=get_id(players,game[1])
#    EloA,EloB,score=players[a][1],players[b][1],game[2]/(game[2] +game[3])
    if game[2]>game[3]:
        match=1
    elif abs(game[2]-game[3])<1e-6:
        match=0.5
    else:
        match=0
    EloA,EloB,score=players[a][1],players[b][1],match
    DeltaElo=variation_Elo(EloA,EloB,score,k)
    players[a][1]+=DeltaElo
    players[b][1]-=DeltaElo

def EloFile(filename,players,Provisionnal):
    """ writes a file of new elos"""
    f=open(filename,'w')
    players.sort(key = lambda players:players[1])
    players.reverse()
    rank=1
    s=""
    for P in players:
        s+=str(rank)+'\t'+P[0]+'\t'+str(round(P[1]))
        if P[0] in Provisionnal:
            s+='\t'+'*'
        s+='\n'
        rank+=1
    f.write(s)
    f.close()
    
def main(playersfile,gamesfile,gf2):
    """main program ; run with 'euro.txt' and 'eurores1.txt' for tests"""
    players,Prov=parse_former_Telo(playersfile)
    Prov=[]
    results=parse_results(gamesfile)
    for game in results:
        updates_ELO(players,Prov,game)
    results=parse_results_2(gf2)
    for game in results:
        updates_ELO(players,Prov,game)
    EloFile('neweloEU.txt',players,Prov)

def main2(playersfile,gf2,gf3):
    players,Prov=parse_former_Telo(playersfile)
    results=parse_results_2(gf2)
    for game in results:
        updates_ELO(players,Prov,game)
    results=parse_results_2(gf3)
    for game in results:
        updates_ELO(players,Prov,game)
    EloFile('neweloAsia.txt',players,Prov)










[Updated on: Tue, 16 May 2017 01:28]

      
Qorlas
Senior Member
Faster Than Mongolian Horse

User Pages
Posts: 2058
Registered:
January 2008
Re:Tournament ELO: Update Dec 28th- 2p USA after NC 2016 Tue, 16 May 2017 14:14
We (me, Dea, Truck, Sysyphus and Drake) had a long discussion when we created the TELO years ago. I remind a lot of simulations to find the best k-value.... too high and last results mean too much too low and the opposite happens.

I have somewhere te excel file of Telo with all the matches.. probably the old pc.

We had the archive of all the results from 2006 or so. Regarding players disappearing.... we used a 2 years validity of ranking. If a player doesn't play for 2 years then your score is freezed out.
Provisional was made with a penalty system according too the series played...

The penalty system was used to avoid clear mistakes of the system like the one of Michelecannone/Belelule.

Everything was on tickipedia, is it gone? I could search it around.


P.S.:
Do not start from scratch. I am reading again the same brainstorming we had years ago

P.P.S:
Include all the results do not use a start date. In that way you aare not considering Chris results in the past that are quite better than his proposed ranking... In any case that is not crucial.. Chris will have long tournament in any case Wink
      
VojFromMars
Senior Member

Posts: 164
Registered:
January 2016
Re:Tournament ELO: Update Dec 28th- 2p USA after NC 2016 Tue, 16 May 2017 14:31
Of course I would be interested to get the results of the discussions you had.

There is a deep unsolvable problem with EU Elo : when you have a several subsets of players that never (or rarely) compete together, the ratings you'll get will only rank the players inside their subset and the ratings will have no relative meaning between the different pools.

This problem is well known in chess or go ratings for national (or in some cases regional) ratings for isolated populations.

EU Map competitions are EU League (~10 matches), EMC (1 match for half of the players, 2 matches for a quarter of the players so very little comparision points say for an average League player)
and some Fusion matches (where most of players by now are really dedicated players so it boils down again to the same pool of Div A players).

I guess at one point you were using k=24 for League and Fusion and k=30 for EMC for this very reason - better adjust ratings of different subsets between them.. This still doesn't seem quite enough so some manual adjusting is necessary. Which usually means something is bad.

Of course, after a while, ratings should converge to their true value. But it might take several years, and players could also improve their skill, quit the game etc... in the meantime.

Best solution would be to run an open tournament "mixing it up" (using a swiss system pairing) once a year.
      
Qorlas
Senior Member
Faster Than Mongolian Horse

User Pages
Posts: 2058
Registered:
January 2008
Re:Tournament ELO: Update Dec 28th- 2p USA after NC 2016 Tue, 16 May 2017 16:27
No,

we gave the different tournaments different value of K.

Not all the tournaments are at the same level of importance.

NC, SPWC and EMC are tier 1 for example.
      
Qorlas
Senior Member
Faster Than Mongolian Horse

User Pages
Posts: 2058
Registered:
January 2008
Re:Tournament ELO: Update Dec 28th- 2p USA after NC 2016 Tue, 16 May 2017 16:39
Sending you the file. Keep in mind that you really need all the history of matches here.
      
VojFromMars
Senior Member

Posts: 164
Registered:
January 2016
Re:Tournament ELO: Update Dec 28th- 2p USA after NC 2016 Tue, 16 May 2017 17:19
Qorlas écrit le Tue, 16 May 2017 16:39

Sending you the file. Keep in mind that you really need all the history of matches here.


Thanks, got the file.

Obviously it's much easier with the original file. Of course I need all the previous data, without it I'm pretty much doing what I can, and collecting back all original data and guessing the factors formerly used would be a tremendous work.

On a side note, I'd like you guys not to be too harsh about it. I already spent some time on it, without getting much help in the start. I've done the first calculations with the data I had and the factors I could guess as well as possible. Those calculations appeared off the usual charts, and people started to complain. Obvisouly, I understand that. But please keep in mind I'm doing this on my free time, instead of say ... playing TTR.

I'll do again the calculations using previous formulas this night. And I'll also compute al alternate formula for unrated players - for fun and comparison.
      
AWT allan
Senior Member

Posts: 343
Registered:
January 2014
Re:Tournament ELO: Update Dec 28th- 2p USA after NC 2016 Tue, 16 May 2017 20:03
How am I this high in EU rankings?!
      
Truckerteller
Senior Member
Conductor Level 1

User Pages
Posts: 947
Registered:
October 2007
Re:Tournament ELO: Update Dec 28th- 2p USA after NC 2016 Tue, 16 May 2017 23:02
You obviously don't need all the previous results, but it helps to get a sense of the calculation.

I don't read code/python, but my suggestion would be to test with just two matches and three players, say 150 points rank difference each match. With a 30 k, rank changes should be 9, 21 or 6, since expected score should be roughly 0.7 vs 0.3.

Can't really do excel here, but I had calculated with match telo, my euro rank would be 1890-1895 or so. EDIT : I calculated it to be 1790-1795. So something is really off here. Maybe you're multiplying the result with nr. of games in the match?

Just trying to help (while on holidays).

Ps. Limited amounts of results will always be an issue, but I don't think split populations are an issue here.

[Updated on: Wed, 17 May 2017 10:48]

      
DrakeStorm
Senior Member
T2R 2014 World Champion

Posts: 1053
Registered:
March 2006
Re:Tournament ELO: Update Dec 28th- 2p USA after NC 2016 Wed, 17 May 2017 08:33
I haven't looked at my original excel spreadsheet in a long time. (I have very little programming experience, so the VBA code is probably a complete mess).

I know we decided to use match TELO for formats that we thought we had enough data points for, which at the time was USA (and maybe Europe I forget). We used game TELO for the others (Asia, Swiss, AAT, etc.)

We used all the previous tournament data (except 1 year of NC because it was formatted like crap on tikipedia as I recall)

In my excel spreadsheet you could pick the k-factor for the various tournaments and run the data. We ran it with all sorts of different k-values, and I think I decided (don't remember what other people thought), that for Match TELO the premier tournaments would be k-30 (NC, SPWC, ??), and then lesser ones would be less. And for Game TELO the base k-value was 8 (10??) I think.

For Fusion, I can't remember what I did about the split match (the one where it was 3 EU games and 3 USA games). For Game Telo it is fine, but for Match TELO, I'm not sure if I made it 2 matches.

I was never a TD, but I recall there was a penalty to your TELO (or DoW ELO, I forget) for every data point less than 20 you had for seeding the tournaments.

I know Sysyphus kept up the excel spreadsheet with data after I stopped working on it. I assume most data is there except maybe the last few tournaments. So just ignore my coding and use the data.
      
RUS Sivorro
Senior Member
T2R Multi-Player Winter Championship 2014 Winner

Posts: 1383
Registered:
April 2014
Re:Tournament ELO: Update Dec 28th- 2p USA after NC 2016 Mon, 22 May 2017 11:16
I saw only now that my friend and coach SuperPello have now 5th euro TELO.
I really like him but I really don't like your calculations Smile

Are you sure current TELO is correct on all maps?

[Updated on: Mon, 22 May 2017 11:17]

      
VojFromMars
Senior Member

Posts: 164
Registered:
January 2016
Re:Tournament ELO: Update Dec 28th- 2p USA after NC 2016 Mon, 22 May 2017 13:01
I'm pretty sure it's not correct. But I'm also pretty sure the differences with the right calculation won't be far off...

To be runned, soon.
      
VojFromMars
Senior Member

Posts: 164
Registered:
January 2016
Re:Tournament ELO: Update Dec 28th- 2p USA after NC 2016 Mon, 22 May 2017 13:03
[Reserved]

[Updated on: Mon, 22 May 2017 13:03]

      
Qorlas
Senior Member
Faster Than Mongolian Horse

User Pages
Posts: 2058
Registered:
January 2008
Re:Tournament ELO: Update Dec 28th- 2p USA after NC 2016 Mon, 22 May 2017 15:46
What is the database you used for it?

It is definitely weird that Superpello is so high in the proposed ranking considering the results in Euro last year.
      
VojFromMars
Senior Member

Posts: 164
Registered:
January 2016
Re:Tournament ELO: Update Dec 28th- 2p USA after NC 2016 Mon, 22 May 2017 19:47
I created the database of Eu League 2016 and Fusion Cup.

All results are in the related threads.

I've used k=30 instead of k=24 though..

I'll recalculate those results later, I've still have a rush of work. I'll also suggest some alternative calculus for new players, which you'll like or not...

More on that in one week.

SuperPello won 8 matches and drew 1 - ok in League C but if unrated players are computed at 1600 his rating is in no way surprising... What results are you referring to Qorlas ?
      
Truckerteller
Senior Member
Conductor Level 1

User Pages
Posts: 947
Registered:
October 2007
Re:Tournament ELO: Update Dec 28th- 2p USA after NC 2016 Mon, 22 May 2017 20:17
As an example, Superpello was rated 1549 previously, now 1755.
If we assume an expected result around 4,5 and his realized score is 8 (8 wins 1 loss) he should rise ca. 3,5 x 30 = 105 in rating. 200+ is way too much.

My rating went from 1681 to 1931. With 9,5 out of 11 and an expected score of 5,8 - 6,0 I figured to rise to 1790-1795.

Like I wrote before, pls have a look at the results of one or two matches first, to check if the calculationsame are ok.
      
SuperPello
Senior Member

User Pages
Posts: 892
Registered:
January 2011
Re:Tournament ELO: Update Dec 28th- 2p USA after NC 2016 Mon, 22 May 2017 20:46
Wow, I'm the man of the day Laughing Laughing Laughing

      
VojFromMars
Senior Member

Posts: 164
Registered:
January 2016
Re:Tournament ELO: Update Dec 28th- 2p USA after NC 2016 Mon, 22 May 2017 21:44
Ok. I finally found the mistake. Thanks for the advice of running unitary tests TT but ... of course I do that every time. Mistake was two arguments (k and score) weren't appearing in the same ordre in the function definition and the function call... Stupid and not so easy to track.

I've recalculated - quickly - Europe TElo with k=24 and previous data. Here's the list - which will be used for computation of EMC matches (not for the seeding this one is too late) :

1 nicolas21 1780
2 Truckerteller 1763
3 Marschall Vorw 1746
4 Morientes ZL 1737
5 Sivorro 1729
6 335d 1715
7 Sysyphus 1710
8 Val 1679
9 Lexx 1678
10 yaelka 1660
11 Patterson 1660
12 Hecki 1657
13 CMEPTb 1657
14 dandee 1653
15 GSV3 1646
16 cannonade 1644
17 IL2T 1644
18 RMarkes 1643
19 xbomanx 1639
20 benny 1639
21 belelule 1638
22 Stormwarrior 1633
23 SuperPello 1626
24 vigulisam 1622
25 VojFromMars 1619
26 Knockando 1617
27 dea1 1617
28 toddcalu 1617
29 Enix - Poulsard 1609
30 Alautre 1605
31 Will 1603
32 allan 1601
33 DrakeStorm 1599
34 Alexkey 1599
35 Megasiv 1599
36 thesteamwistle 1598
37 OllieB 1594 *
38 The Tinman 1593
39 Lucullupus 1589
40 michelecannone 1587
41 PopeRoma 1587 *
42 mikola 1586
43 Marviktori 1585
44 LeGoupil 1585
45 TheBendite 1584
46 chekist13 1584
47 Cromze 1583
48 Loca 1577
49 Diplodokus 1574
50 tiploup 1572
51 dizz 1571
52 The Man Without 1568
53 lirex 1568
54 Pammes 1567
55 Ironhorse 1567
56 DN 1565
57 Elric 1559
58 Ukky 1556
59 Julichka 1551
60 Doc 1542
61 Qorlas 1535
62 Dgèdge34 1523
63 onyx puffin 1510
64 MaximouR 1510
65 Photios 1507
66 Aurelie in wonderland 1504 *
67 JenAck 1491
68 Tempest0 1487
69 Cat 1487
70 Tyrana 1463
71 Beefy 1458
      
VojFromMars
Senior Member

Posts: 164
Registered:
January 2016
Re:Tournament ELO: Update Dec 28th- 2p USA after NC 2016 Mon, 22 May 2017 21:47
And here's the Eu list with k=30.. With so few dataI don't really see a reason not to use k=30.. well open to discussion obv.

1 Truckerteller 1781
2 nicolas21 1767
3 Marschall Vorw 1746
4 Morientes ZL 1744
5 Sivorro 1734
6 335d 1715
7 Sysyphus 1710
8 Val 1679
9 Lexx 1678
10 yaelka 1670
11 CMEPTb 1664
12 Hecki 1662
13 Patterson 1657
14 dandee 1654
15 GSV3 1646
16 cannonade 1646
17 RMarkes 1643
18 IL2T 1642
19 SuperPello 1642
20 xbomanx 1639
21 benny 1638
22 belelule 1635
23 Stormwarrior 1633
24 VojFromMars 1626
25 toddcalu 1617
26 Knockando 1614
27 vigulisam 1612
28 Enix - Poulsard 1609
29 dea1 1608
30 Megasiv 1607
31 Alautre 1605
32 Will 1603
33 allan 1601
34 Alexkey 1599
35 thesteamwistle 1598
36 DrakeStorm 1598
37 The Tinman 1593
38 OllieB 1593 *
39 Lucullupus 1589
40 LeGoupil 1587
41 michelecannone 1587
42 mikola 1586
43 Marviktori 1585
44 TheBendite 1584
45 chekist13 1584
46 PopeRoma 1583 *
47 Cromze 1583
48 Loca 1582
49 Diplodokus 1574
50 lirex 1572
51 dizz 1571
52 The Man Without 1568
53 Pammes 1567
54 DN 1565
55 Ironhorse 1565
56 tiploup 1561
57 Elric 1559
58 Julichka 1551
59 Ukky 1549
60 Doc 1542
61 Qorlas 1535
62 Dgèdge34 1523
63 MaximouR 1510
64 onyx puffin 1509
65 Photios 1505
66 JenAck 1494
67 Cat 1487
68 Tempest0 1485
69 Aurelie in wonderland 1483 *
70 Tyrana 1463
71 Beefy 1453
      
RUS Docent CMEPTb
Senior Member
Romanov Train

User Pages
Posts: 413
Registered:
March 2011
Re:Tournament ELO: Update Dec 28th- 2p USA after NC 2016 Wed, 14 June 2017 13:32
VojFromMars wrote on Mon, 22 May 2017 22:47

And here's the Eu list with k=30.. With so few dataI don't really see a reason not to use k=30.. well open to discussion obv.

Let's keep:
k=30 - NC, SPWC, EMC
k=24 - Fusion Cup

What about League? Keep the same for all: A, B, C? Or to make differences? How it was before?
      
Qorlas
Senior Member
Faster Than Mongolian Horse

User Pages
Posts: 2058
Registered:
January 2008
Re:Tournament ELO: Update Dec 28th- 2p USA after NC 2016 Wed, 14 June 2017 16:00
Same for all k=24
      
Truckerteller
Senior Member
Conductor Level 1

User Pages
Posts: 947
Registered:
October 2007
Re:Tournament ELO: Update Dec 28th- 2p USA after NC 2016 Wed, 14 June 2017 16:58
I think currently, all those who play fusion play it seriously. I don't really see a reason to assign it a different k factor.

Previously, a higher k factor was assigned to best-of-9 series, which makes sense, but might also be a high penalty if you lose an SPWC final 5-4. On the other hand, NC matches used to be best-of-5 in the early days.

k=30 for everything makes most sense to me, with a minimum of 3 wins necessary to win or tie.

If you're going with k=24 for some tournaments and k=30 for others, you should rerun the whole set of results and not start with the list Sysyphus published, since that has k=30 for all matches.

      
RUS Docent CMEPTb
Senior Member
Romanov Train

User Pages
Posts: 413
Registered:
March 2011
Re:Tournament ELO: Update Dec 28th- 2p USA after NC 2016 Wed, 14 June 2017 18:46
Sounds good. it's good to have simple rules.

[Updated on: Wed, 14 June 2017 18:47]

      
VojFromMars
Senior Member

Posts: 164
Registered:
January 2016
Re:Tournament ELO: Update Dec 28th- 2p USA after NC 2016 Mon, 19 June 2017 02:46
Europe TElo 2p

After EMC 17

1 Truckerteller 1781
2 nicolas21 1776
3 Marschall Vorw 1746
4 Sivorro 1736
5 Morientes ZL 1721
6 chrismmm 1720
7 335d 1715
8 Sysyphus 1710
9 Val 1679
10 Lexx 1678
11 yaelka 1670
12 Hecki 1663
13 cannonade 1651
14 CMEPTb 1648
15 GSV3 1646
16 VojFromMars 1643
17 RMarkes 1643
18 Patterson 1642
19 IL2T 1642
20 SuperPello 1641
21 xbomanx 1639
22 benny 1638
23 dandee 1634
24 Stormwarrior 1633
25 belelule 1617
26 toddcalu 1617
27 Knockando 1613
28 vigulisam 1612
29 Enix - Poulsard 1609
30 dea1 1608
31 Alautre 1605
32 Will 1603
33 Olimpionik 1602 *
34 allan 1601
35 Alexkey 1599
36 DrakeStorm 1598
37 OllieB 1593 *
38 Megasiv 1592
39 Lucullupus 1589
40 michelecannone 1587
41 LeGoupil 1587
42 mikola 1586
43 Marviktori 1585
44 chekist13 1584
45 TheBendite 1584
46 Cromze 1583
47 PopeRoma 1583 *
48 lirex 1582
49 thesteamwistle 1579
50 The Tinman 1574
51 Diplodokus 1574
52 dizz 1571
53 Loca 1568
54 The Man Without 1568
55 Pammes 1567
56 Ironhorse 1565
57 DN 1565
58 tiploup 1561
59 Elric 1559
60 Julichka 1551
61 Ukky 1549
62 Qorlas 1547
63 Doc 1542
64 onyx puffin 1541
65 Dgèdge34 1523
66 Photios 1513
67 MaximouR 1510
68 JenAck 1494
69 Cat 1487
70 Aurelie in wonderland 1483 *
71 Tempest0 1477
72 Tyrana 1463
73 Beefy 1453
      
RUS Docent CMEPTb
Senior Member
Romanov Train

User Pages
Posts: 413
Registered:
March 2011
Re:Tournament ELO: Update Dec 28th- 2p USA after NC 2016 Mon, 19 June 2017 10:07
Is it k=30 or k=24?
      
VojFromMars
Senior Member

Posts: 164
Registered:
January 2016
Re:Tournament ELO: Update Dec 28th- 2p USA after NC 2016 Mon, 19 June 2017 14:05
RUS Docent CMEPTb écrit le Mon, 19 June 2017 10:07

Is it k=30 or k=24?


k=30 as agreed (I'll double check back home)
      
RUS Docent CMEPTb
Senior Member
Romanov Train

User Pages
Posts: 413
Registered:
March 2011
Re:Tournament ELO: Update Dec 28th- 2p USA after NC 2016 Mon, 19 June 2017 14:49
VojFromMars wrote on Mon, 19 June 2017 15:05

RUS Docent CMEPTb écrit le Mon, 19 June 2017 10:07

Is it k=30 or k=24?


k=30 as agreed (I'll double check back home)

Last year Euroleague and this year FusionCup also k=30?
      
VojFromMars
Senior Member

Posts: 164
Registered:
January 2016
Re:Tournament ELO: Update Dec 28th- 2p USA after NC 2016 Mon, 19 June 2017 20:24
RUS Docent CMEPTb écrit le Mon, 19 June 2017 14:49

VojFromMars wrote on Mon, 19 June 2017 15:05

RUS Docent CMEPTb écrit le Mon, 19 June 2017 10:07

Is it k=30 or k=24?


k=30 as agreed (I'll double check back home)

Last year Euroleague and this year FusionCup also k=30?


Yes and yes.
      
RUS Docent CMEPTb
Senior Member
Romanov Train

User Pages
Posts: 413
Registered:
March 2011
Re:Tournament ELO: Update Dec 28th- 2p USA after NC 2016 Mon, 19 June 2017 20:31
Perfect!

Thank you very much, Voj!
      
Pages (3): [ «  <  1  2  3 ]     
Previous Topic:New Eurolegue Rules Evolution
Next Topic:EMC 2017
Goto Forum: