It has been added, though I don’t see how you expect a loss to bring up the average?
I wasn’t sure if there were any others missing. Also, I think he has matches scheduled this week that might make a difference, so I’ll wait regardless.
You’re looking at the set win %. The 6.0 club has always been based on the game win %.
I think Fivec counts by game winrate, not set winrate. Still, grats on the achievement!
I do see how there could be some confusion, however, as I said in the OP that you have to have 100 matches played to be considered. What I meant was games. Even I don’t have 100 matches played!
Where’s game winrate then?
Bottom right corner of the MU chart itself.
I don’t suppose other people will be as excited as me about statistical rigour, but I found it pleasantly surprising that 6.0 and 100 games are nice round numbers that happen to be just sufficient to say that a 6.0 club member’s performance is statistically different from 50-50 odds of winning over the same number of games.
Obviously more games are better, but at least a sample size of 100 games tells us something slightly meaningful. If you want, it would also allow each step of 1.0 above to be a tier level, as 7.0 would also be statistically different from 6.0, etc.
To chip in about better performance measures, I also agree that the last 100 games would be a much more meaningful metric than lifetime performance, since the underlying win rate is more likely to be consistent once a player exits their initial learning period.
Assumption: Normal approximation to binomial distribution is applicable as we have more than 30 observations (N) and p=0.6 is not close to 1 or 0.
Estimated standard error = sqrt [ p * (1 - p) / N ] = sqrt [ 0.6 * (1 - 0.6) / 100 ] = 0.049
95% confidence interval = [0.6 - 1.96 * 0.049, 0.6 + 1.96 * 0.049] = [0.503 , 0.696]
–> excludes 0.5 (aka 50-50) and 0.7
Has anyone run an ongoing ELO-type calculation on the tournaments matches since 2014?
Nope, but if you want to create something like it, go for it. The data in the spreadsheet is as mineable as I’ve been able to make it.
It has been on my theoretical to do list for a while, as in eventually, when I’ve not got so much stuff to do I’ll get round to it. Obviously that hasn’t happened yet though.
I did do some reading up on different ways to calculate elo and I think it seems like an underlying logistic function would be sensible for yomi, since weaker players can simply win combat and beat a higher ranked opponent. This is what they tend to use in chess official rankings aswell, though they started with a normal distribution when elo was first introduced.
Welp, I went ahead and ran the Elo rankings over all games in the history. I couldn’t figure out an actually useful way to represent the changes over time, but here are the top ten players by average Elo:
deluks917 1647.715506 niijima-san 1646.212821 cpat 1639.649159 raziek 1635.015867 fivec 1629.652394 ntillerman 1618.968151 drnd 1616.286345 jengajam 1612.017653 mastrblastr 1606.954037 madking 1601.987083
EDIT: I believe in showing my work: https://github.com/cpennington/yomi-skill/blob/master/Skill%20Record.ipynb
But that can’t be right… Fivec is not and will never be good.
This is just a humble request. I am ectstatic about the work that you have already done. Is it possible to implement this in mysticjuicer’s spreadsheet, so that we could have a real time elo update? This would be a great tool. Thanks for the great work.
I think it’s not a simple thing to do within the confines of the spreadsheet, but of course it could be wrong.
If I understand it correctly, it would require feeding the results of each tournament into the elo calculator and updating all the rankings of those who participate. It’s the sort of thing that could be done if the JuicerChart™ became a website of its own.
Also super awesome work @vengefulpickle *applause* you are a man who gets things done! Though, the numbers look like they might be quite compressed. Does the algorithm use a grandmaster k-factor (16) or a more dynamic k-factor for regular play (32)? I’d just look, but it’s hard parsing a text representation of an ipython notebook on my phone.
@mysticjuicer hey I noticed I have two names on the results chart, is that easy to fix?
Ah, I was normalizing for case, but not underscores/spacing.
RIP most of that top ten. We’re keeping the game warm for ya, mates.
Yeah, it’s using a k-factor of 32, but the liftetime-average is squashing a lot of that variability.
I fixed a bit of the name normalization issues, and ran some calculations using the most recent Elo for each player. So, here’s are the top 20 players (of all time) (including their most recently played event).
Date Event Rating Event Date Player cpat 2016-09-25 FS.com Into Oblivion 1808.324200 2016-09-25 Fusxfaranto 2017-08-08 IYL Season 5 1702.045473 2017-08-09 Djister 2016-12-01 Rook's Rock n' Rumble 1694.527877 2016-12-01 Legion 2017-06-24 Fractured Factions 1685.733084 2017-06-24 ntillerman 2016-11-26 Topanda League 2 1684.714695 2017-02-17 Niijima-San 2017-07-30 NFTT Round 6 1673.792658 2017-07-30 SirHandsome 2017-08-01 Summer Smash IV 1670.884480 2017-08-07 Nemesis-Kanden 2015-11-08 Triple Threat Throwdown 1656.915434 2016-03-05 Caralad 2017-07-30 Solo Showdown II 1655.416689 2017-07-30 BD Corro 2017-08-05 IYL Season 5 1654.928403 2017-08-09 ratxt1 2016-11-20 Topanda League 2 1654.715575 2017-02-17 deluks917 2017-04-29 NFTT Round 3 1644.428898 2017-04-29 Leontes 2017-08-07 IYL Season 5 1640.243302 2017-08-09 flagrantangles 2017-08-02 Summer Smash IV 1638.267931 2017-08-07 FenixOfTheAshes 2016-11-07 Topanda League 2 1638.136553 2017-02-17 thehug0naut 2017-08-09 IYL Season 5 1631.049391 2017-08-09 Southpaw Hare 2017-08-09 IYL Season 5 1629.294845 2017-08-09 Mallorean_Thug 2016-11-12 The Yomi Olympic Carnival 1624.587128 2016-12-20 Fivec 2017-08-08 IYL Season 5 1624.266083 2017-08-09 Raziek 2016-09-25 FS.com Into Oblivion 1621.907318 2016-09-25
And here are the top 20 players who have played a tournament match in the last 6 months.
Date Event Rating Event Date Player Fusxfaranto 2017-08-08 IYL Season 5 1702.045473 2017-08-09 Legion 2017-06-24 Fractured Factions 1685.733084 2017-06-24 Niijima-San 2017-07-30 NFTT Round 6 1673.792658 2017-07-30 SirHandsome 2017-08-01 Summer Smash IV 1670.884480 2017-08-07 Caralad 2017-07-30 Solo Showdown II 1655.416689 2017-07-30 BD Corro 2017-08-05 IYL Season 5 1654.928403 2017-08-09 deluks917 2017-04-29 NFTT Round 3 1644.428898 2017-04-29 Leontes 2017-08-07 IYL Season 5 1640.243302 2017-08-09 flagrantangles 2017-08-02 Summer Smash IV 1638.267931 2017-08-07 thehug0naut 2017-08-09 IYL Season 5 1631.049391 2017-08-09 Southpaw Hare 2017-08-09 IYL Season 5 1629.294845 2017-08-09 Fivec 2017-08-08 IYL Season 5 1624.266083 2017-08-09 Zqxx 2017-08-01 IYL Season 5 1612.536755 2017-08-09 mysticjuicer 2017-08-07 Summer Smash IV 1611.708661 2017-08-07 lettucemode 2017-03-29 19XX: Waifu 2 1610.078548 2017-04-29 Fluffiness 2017-08-02 IYL Season 5 1602.606360 2017-08-09 Ivan 2017-08-05 IYL Season 5 1597.527763 2017-08-09 ClanNatioy 2017-07-19 IYL Season 5 1593.539530 2017-08-09 CloudCuckooCountry 2017-08-04 Summer Smash IV 1577.987334 2017-08-07 Shax 2017-07-18 IYL Season 5 1574.900974 2017-08-09
Apparently, I also have the dubious distinction of being the lowest ranked player to have played a tournament game in the last 6 months! So, nowhere to go but up!
Yeah, that happens sometimes. I try to minimize it as much as possible. I’ve harmonized your results under BD_Corro.