Loading...
  OR  Zero-K Name:    Password:   
Title: [A] Teams All Welcome #1
Host: Nobody
Game version: Zero-K v1.14.6.1
Engine version: 2025.06.21
Battle ID: 2457741
Started: 21 days ago
Duration: 8 minutes
Players: 22
Bots: False
Mission: False
Rating: Casual
Watch Replay Now
Manual download


Show winners



Preview
Filter:    Player:  
sort

21 days ago
why is this game so favoured for one team? Surely with this many players there was a better player allocation than this?
+1 / -0
The chance of victory on the site uses current ratings, iirc. You can't trust it. Currently it reads 41.2% for the left team.
+0 / -0

21 days ago
Essentially because the bottom skilled players of the left team are not done yet falling even futher below start rating.
+0 / -0
According to the WHR hindcast, Eridug took a two year break and then came back the day before this, improving from a true grey to a secret gold while absent. Ladder rating hasn't caught up yet, but Eridug's current WHR as of this comment is about 1850.

Apparently Eridug's massive improvement came from discovering the strategy of building a Caretaker with a guard order on a better player's main factory.
+2 / -0

21 days ago
Team-rating never worked and probably won't work unless we have a crazy neural network to determine:
- the individual inpact a specific player has on the outcome of a game with multiple players per team.
- what synergies/anti-synergies are likely to occur.
+2 / -0
I have suggested changes that I think would be better than what we have, but in the end, there is no fixing a pool of the same 32 players. There are only so many ways you can arrange 32 players that don't end up horribly unbalanced.

32 out of millions you can start by putting all the players in similar skill brackets together, but 32 out of 32?

Also "to the degree that it can" the balancer pairs clans and parties together, further reducing the number of ways the teams can be arranged.

No algorithm can fix one player being spectacularly weak and the opposite team rushing that side. The only fix to that is to increase the ZK player base so players can be isolated into their respective skill brackets.
+0 / -0
I'm happy that the WHR as the ultimate best predictor of a match is coming into question. When I said that before, I got harshly attacked by FIrankFine that I wouldn't know what I was talking about.
+0 / -0

20 days ago
quote:
No algorithm can fix one player being spectacularly weak and the opposite team rushing that side. The only fix to that is to increase the ZK player base so players can be isolated into their respective skill brackets.

Better pregame decisions can fix that problem. Lane rotations can fix that problem. If a team fails to do either, that is not the fault of the balancing algorithm, except to the extent that spotting that kind of weakness in your team and buttressing it is a measurable skill.

In a small teams game, or on a map inappropriately spread for the player ccount, there's only so much that can be done about a weak isolated player. That's not what happened here.
+1 / -0

20 days ago
quote:
Better pregame decisions can fix that problem.


If they made better decisions, they wouldn't be so low ranked to begin with.
+2 / -0
19 days ago
A predictor will remain a predictor. We can't expect everybody to play exactly as their rating all the time. I knew people that played generally as one game as (reasonable, non reportable) trolls and next game as blue level. Not much you can do about those...

Probably a simple and easy thing to do would be to rename "Chance of victory" to something more correct like "Chances of victory based on current ratings". Additionally, maybe it should be clarified (by a text) if the ranks that we see listed are of today or at from the moment the battle took place. Such a thing can also make the balancer look "bad" if not clearly explained.
+0 / -0