People often say that Seraphim should be nerfed and Aeon should be buffed. The data supports that assessment.
Pro Faction Usage
I thought UEF was about as common as Cybran but seems like it’s significantly less so
put the xbox units in the game pls u_u
UEF is as common as Cybran, there are about 18,000 more players a month than these guys and new players have always been biased to UEF and Cybran.
Problems with the data:
-
Random is not accurate
Case in point is Farm who has done 0 ladder and 0 tournaments and so has played basically only teamgames and tmm. He has played pure random in both of this situations. So I have no idea how his graph shows him picking anything else. -
Data includes all types of game modes (ladder, tmm, global, dual gap, sentons, etc)
-
Data includes all types of players (ladder, tmm, global, dual gap, sentons, etc)
This is dangerous because of the way that TrueSkill operates. You can see this in certain players (Rogue_Toaster) who mostly plays Cybran in his dominant sphere (dual gap) and you can see his win rate reaches around 50% in that situation. This is what you would expect over the long trend of games in the area where he specializes. He also began testing out stuff like tmm once it got released and I noticed him playing other factions. With TMM giving you a soft reset to your rating to ensure you get placed in an area that you are comfortable in even if you play nothing but astro in custom games, this means that if he played a faction (like sera) a lot during those initial games he has a ridiculous win rate with sera which manipulates the data. -
All win rates are averaged out
Nexus has basically only played tournament games and occasional 2v2s this year. Weighing him as equal to dudes that have played as often as he has across the year every month creates inherent noise. -
Tournament formats are not accounted for
This applies especially to Nexus, where his win rates are going to be heavily influenced by what faction he was actually allowed to play on maps in the LotS faction ban format. If he was going to beat say Swkoll (sorry Swkoll) regardless of the faction situation, then attributing the win to his faction choice is not coherent. It also doesn't say anything about a faction being the best on an average of maps because that faction is the one that likely got banned in these tournaments.
Looking at the people with above 60% win rate, which is a sign of RIDICULOUS imbalance as 55% is considered to be something meta oppressive in a variety of games, 9 of the 12 (only Tagada and Grimplex UEF feel like they have a coherent spread of games) have suspiciously low game counts. I wouldn't trust this to say much about the state of balance and would expunge data points like Nexus (85% is beyond meta defining and essentially means he should be closing his way on 3000 rating if he decided to play again which I think everyone realizes is hilariously stupid and not reality).
I dont trust these statistics. Basically everyone at high level plays random only. Also, if someone played tmm/ladder
with all but one factions selected, would that add a point to random or to the faction they just happened to get?
profile picture credits to petric
Iirc the client chooses a faction for you when you play matchmaker. The only way to be playing as random in the sense of these statistics is to select random in a custom lobby
@blackyps said in Pro Faction Usage:
Iirc the client chooses a faction for you when you play matchmaker. The only way to be playing as random in the sense of these statistics is to select random in a custom lobby
sneed is like a 90% senton player and has 0 random games here over the year which is played with everyone going random.
@blackyps
That also doesn't work. The stats regarding my profile clearly prove it. I play basically only random and yet according to stats I only played like 5% of games as random in this year.
As the game is launching all players are assigned a faction:
I'm not entirely sure when the data is send to the client, but I suspect it happens at either of these lines:
- https://github.com/FAForever/fa/blob/deploy/fafdevelop/lua/ui/lobby/lobby.lua#L2174
- https://github.com/FAForever/fa/blob/deploy/fafdevelop/lua/ui/lobby/lobby.lua#L2181
- https://github.com/FAForever/fa/blob/deploy/fafdevelop/lua/ui/lobby/lobby.lua#L2191
Where the latter two would be intercepted using the ICE adapter. Anyhow, all of that happens after the random faction has been assigned by the lobby. I think choosing a random faction is only properly tracked when you join match making.
A work of art is never finished, merely abandoned
Where the latter two would be intercepted using the ICE adapter. Anyhow, all of that happens after the random faction has been assigned by the lobby. I think choosing a random faction is only properly tracked when you join match making.
There is no random faction in matchmaking. The server just picks a faction from your choices and then tells the game to use that faction for you.
