Is an Idea about rankings at FAF with stats from Tourneys/tbh
-
This might merit it's own thread, but I ran the ELO algorithm on some major tourney matches since LoTS 2019 and got these ELOs:
Nexus 1099.272909
Tagada 1070.33372
Espi 1060.08074
Blackheart 1047.381412
Petric 1044.282831
TURBO2 1024.775141
Blodir 1021.094128
Swkoll 1010.023158
Tex 1004.427425
Shen 1003.998818
Paralon 1003.335188
Yudi 1002.167972
Box 995.9772686
Bennis 995.6637473
WhiteOwl 995.61104
NarNerdPower 995.3611353
WoundedElk 995.0967664
Bloodoath 995.0031833
Raider 995
keyser 994.8622503
Thomas 991.235657
Farms 990.694092
Lucied 990.6756125
Zlo 990.6425223
Ryfun 990.6351441
Death_Squad 990.0719798
Isen 990.0698799
Adjux 989.8622984
Bully 985.9777318
Aurico 985.5310263
boom 985.5117396
Smith 985.2848209
Hybrid 980.8372869
Jagged 977.9022702
BlinChlk 976.6583887
Blast 973.5038839
Kot 970.030741
Arch 964.6365672
Turin 964.1520327
Eternal 962.337489The dataset only has 196 series right now so it's very rough but it gives a rough idea of who has had good tourney performance in the last 2 years.
-
Does it get resetted every year or will it just be a permanent ladder?
-
@swkoll Did you use the same sigma as in chess? The distribution appears to be very tight. Or is it just the lack of games so far?
-
@harzer99 I used https://pypi.org/project/elo/ with 1000 default rating. Everyone just has very few games because there haven't been that many tourneys + I didn't bother going back past 2019 LotS.
-
Not sure if this elos really told who is better, somehow it need to compare with true ladder winnings, maybe league seasons stats.
-
Ladder league results are misleading in terms of who is the best cause very few top 5 players participate in them.
-
and that is true too <(
-
I think that in terms of determining top 10 players it's best to only look at the big premiere events like Swkoll's invitationals and LoTS (and upcomming Mini Lots tourneys)
-
That's basically what Swkoll's Leadearboard does and for the top 8 players I think it's pretty accurate but it gets less accurate with the lower seeds.
-
@tagada so is my stats there post I does is shows real statistics.Good
-
Sadly those elo stats are completely useless, the distribution is way too tight - this list says things like bennis vs nexus is only like 65-70% winrate for nexus.
Your best clue to see who is good is asking the top players, we have nothing else. It also varies a ton depending on form/rust/maps/etc.
-
No shit. But its a cool thing to look at the relative hierarchy for tournament matches.
