NewbSaibot: 18 TIMES THE SPEED OF LIGHT. Because FUCK YOU, Daut
1
ToT)MidiaN(   United Kingdom. Nov 03 2020 11:28. Posts 5070
I feel like edges in ZvZ are a bit smaller than the other matchups due to the nature of build order advantages which occur even before you get to scout and feel more severe than they are vs Terran or Protoss. However, I created a spreadsheet a couple of days ago with the winrates of the top 6 Zerg players based on stats taken from sponbbang filtered for 2020. I averaged out the winrates of the top 6 ELO rated Zergs against the top 4 ELO rated Terrans, top 4 Protoss players and the other top 6 Zergs. The highest average winrate against the other top 6 Zergs was Queen, which was an average 57.95%, even higher than it was vs the top 4 Protoss players (57.91%) and way higher than any zerg has managed vs the top 4 Terrans. Granted top 4 and top 6 isn't the same and the sample sizes in mirror matchups are a lot smaller, but it's at least some evidence that these pros actually do find edges. 193 is the most and 104 games is the least sample I have for any of the top 6 Zergs in ZvZ, whereas in ZvT 351 is the smallest sample I have and 233 games is the smallest sample I have vs Protoss. Both of those samples are SoulKey btw, it seems he simply plays less than the other guys, at least as far as spon games are concerned.
Using the stats from my spreadsheet I would say the Zerg rankings are as follows:
Daut, since you're a mathematician and probably have a better grasp of using stats than I do, what do you think is the better method for judging ability between adding all the wins and losses against a set of players and using that winrate to determine a ranking order, or take the winrates against each individual in the sample and averaging the winrates? I used the latter so that if a player had played substantially more games against Flash or Bisu their winrates aren't penalised as badly as they would be if I had just taken overall wins and losses vs the top 4 players. Hope you understand what I mean, just woke up and struggling to make sense!
One day good. One day bad. And some days, even hope
4
Daut   United States. Nov 03 2020 16:29. Posts 8955
Maybe the best way is to take the weighted average sum of the differences between their winrates against each specific player and the opponent's winrate in that matchup.
i.e., imagine soulkey wins in ZvT vs Flash 30% of the time, but Flash wins 78% of TvZ matches (so the zerg wins 22%). And soulkey played Flash in 30% of the matches you're looking at. You could give soulkey 8% * 30% (8% better than avg and 30% of his weighted better than average total goes to this stat) = 2.4% better than average just for Flash. Do this for every player in your group for every matchup, and I think you'll have a decent ranking for each one.
So just to kind of write a totally made up example out.
Let's say Soulkey plays against Flash, Light, Rush, and Sharp in your standings.
There's a slight problem in that Flash is such a GOAT, the win %s vs him have less range. Could also do % better, so like 30/24, 40/31, 52/40, 57/45 instead, and then average those in the same way.
NewbSaibot: 18 TIMES THE SPEED OF LIGHT. Because FUCK YOU, Daut
Last edit: 03/11/2020 16:55
1
ToT)MidiaN(   United Kingdom. Nov 03 2020 16:56. Posts 5070
That makes sense tbf, but I'm a bit lazy to do it haha. I sent you the spreadsheet in discord anyway if you'd like to have a go with the stats in there and compile a ranking using the method you mention?
Edit to make a quick note: I have sOrry as the 4th Terran rather than Sharp, he's currently got a higher ELO rating on sponbbang
One day good. One day bad. And some days, even hope
Last edit: 03/11/2020 17:18
1
blackjacki2   United States. Nov 04 2020 01:49. Posts 2581
since this never got posted
1
captainhook   Canada. Nov 07 2020 11:42. Posts 466
Ok, finally understood how to write spoilers. Fun series, too bad some games ended that quick. Also would have been an epic ending if Queens strategy in Game 6 would have worked
Last edit: 15/11/2020 13:17
1
lostaccount   Canada. Nov 15 2020 15:02. Posts 5811
I want to see a 4 player ffa tournament lol
my karma is done, now time to enjoy life, peace is the way karma is a way Jesus is a way
1
lostaccount   Canada. Nov 15 2020 15:04. Posts 5811
I haven’t watch it yet but I want queen to win cause I need a queen
my karma is done, now time to enjoy life, peace is the way karma is a way Jesus is a way
1
lostaccount   Canada. Nov 15 2020 15:51. Posts 5811
It be nice to see an all random tournament 2
my karma is done, now time to enjoy life, peace is the way karma is a way Jesus is a way
1
Spitfiree   Bulgaria. Nov 15 2020 19:25. Posts 9634
Jelle   Belgium. Nov 15 2020 21:21. Posts 3476
haven't watched asl in forever and I really got into it
GroT
4
Daut   United States. Nov 16 2020 03:02. Posts 8955
One of the least enjoyable seasons in recent memory, but still not too bad with some fun Flash random mixed in, but most of the games/series were just a little lopsided and the maps favored zerg a bit too much. Can't help but wonder how much different everything would have been if Light made it through. Light is absolutely wrecking people in the ultimate battle event, which is basically him playing a bo9 vs top protoss/terran players until he loses. I think he's won about 10 straight, and only person who came close was Soulkey.
edit:
Light has won 8 in a row:
7-2 hero
5-4 shuttle
7-2 larva
7-2 queen
5-4 best
5-4 soulkey
7-2 mini
6-3 snow
Really impressive stuff.
NewbSaibot: 18 TIMES THE SPEED OF LIGHT. Because FUCK YOU, Daut