How is ranking in 4v4 determined?
Posts: 318
I recently had a match where my team consisted of ladder ranks 2697 ost, 152 okw, 819 okw and 2163 ost against a team of players ladder ranked 12 USF, 46 Sov, 300 Sov, and 30 Sov. My team lost, as a result I lost about 200 places in the ladder, this seems counter to how I had thought the ranking system was determined, it would make sense to me that I would lose rank if the opponents were ranked less, but they were all ranked much higher than anyone on my team. (I don't care about ladder position I'm just curious how this ranking system is figured out).
Posts: 1276
Posts: 17914 | Subs: 8
Its fully possible to have say 1000 ELO ost player ranked at 100 go against 1000 ELO sov player ranked at 10, because ELO is distributed internally within the faction.
Additionally, team games, random ones specifically, work on average ELO rating.
That means a team with 70, 10, 10 and 10 ELO is considered equally strong as team with 25, 25, 25 and 25 ELO players.
Posts: 318
I don't care about rank, but the quality of game play with my high ranking factions makes them actually enjoyable, but the game play in low ranking factions feels like I'm smashing my teeth with a hammer. In the factions where I have high rank, the stats are almost always very even across all players in the match, games typically last 20-30 minutes 50/50 about win/los., In my low rank factions, I'll usually have 3-5x the damage, kills, caps, Kill/Death ratio of everyone else on my team, games are either over in about 5-10 minutes or drag on over an hour with about a 25/75 win/loss.
Posts: 440
Posts: 17914 | Subs: 8
How is Elo determined in COH2? Does it interact with ranking at all in organizing the automatch? Reason I ask is if I have a faction where I'm ranked in the top 100, I usually get other very high rank people on my team, if I have a very low rank on a faction, it usually gets placed with other very low rank players, this seems to cause a complete imbalance, where you are likely to win matches with your high ranking faction with high Elo but get stuck into a black hole which basically only goes down and down with a low ranking faction if you have high Elo where you are paired with low Elo players in attempt to "even out" the combined Elo of the team. Seems tailor made to cause complete imbalance where one team has more evenly skilled players where the other team has maybe two good players and two bad players or one very good player and 3 bad players.
I don't care about rank, but the quality of game play with my high ranking factions makes them actually enjoyable, but the game play in low ranking factions feels like I'm smashing my teeth with a hammer. In the factions where I have high rank, the stats are almost always very even across all players in the match, games typically last 20-30 minutes 50/50 about win/los., In my low rank factions, I'll usually have 3-5x the damage, kills, caps, Kill/Death ratio of everyone else on my team, games are either over in about 5-10 minutes or drag on over an hour with about a 25/75 win/loss.
Its completely separate from ranking and its invisible, as I have said on the example, top 10th soviet player can have different elo then top 10th okw player.
You gain and lose elo based on opponents elo. For example, beating much weaker opponent will give you say +2 elo(which does -NOT- translate to rank directly, but is connected to it), but losing to much weaker opponent can result in loss of 20 elo for you and much higher gain for him.
The best explanation possible would be to play DoW2 multiplayer if it wasn't completely dead, because on top of rank you can see your own elo, gains and loses after every game.
However all the above aside, elo tries to make sure to average out and put you on the level of skill that is best suited for you.
Posts: 2885
The source of confusion is the fact that you can't check what your elo is. You can only check your ladder position, while the connection between the two is one directional. Knowing all the elos, you can easily create a ranking. But knowing only a ranking you can say close to nothing about the elo.
By intuition you may expect the ranking to represent your elo, but it really doesn't do that well. You might think that there is similar sum of elo in all ladders and that people at similar spots will have close elo values. But it is not true. A person at position 100 in one ladder can possibly have a few times bigger elo than a person in the same position 100 in different ladder. This creates situations when judging by only the ranking positions matches feel unfair.
What is more, there is no guarantee of density in the ladder. It is possible for example that person on ladder position 500 will have twice as more elo as the person on position 600 in the same ladder, and it is possible that person at position 700 has only a few elo points less than the position 600.
In effect, it is possible that you will go down few hundread of places after losing a match. On the bright side, you shouldn't really care about your ranking, but your elo. If you drop 200 places after a loss it is very probable that you will advance similar amount after a win, just becouse you found yourself in a very dense part of the ladder. Of course, a lot depends on elo of your opponents and allies as well.
Posts: 318
Posts: 8154 | Subs: 2
Rank: for people who had been active on the last month, based on the ELO of the player in that specific team/individual, faction and mode.
Decay: "supposedly" if you are inactive, your ELO slowly starts to go down. Couldn't find too much with the exception of this quote from 2013:
I've asked for information and have a bit to share with you.
We're going to be fine tuning this over the upcoming weeks until we settle on something that we're fully happy with. This was something that you guys have been asking for for awhile and the balance team and myself were pushing for, so I'm glad to see it get into today's patch.
Basically if you haven't played for X days (currently set to a week) then we decrement the top 200 rated players per leaderboard by (rating - 1000)*Y where the current value for Y is set at 4%.
The mathless version is that if you don't play for a week you'll start to drop in rank bit by bit until you either play again, or you drop out of the top 200.
Again these values may be adjusted, so your feedback on how it's working is important.
Now for real ladder experience:
1-decay is imperceptible. Last year, i was inactive with a partner for like 6-12 months, lost my first returning game and i think i gained ranks. Which means that either decay is ineffective or that most top ELO players were inactive.
2- You can be top 200 easily with some few wins in your first placements.
3- You can also climb by sheer volume of games played. If you can maintain a 55/65% WR despite who you are facing, you'll eventually climb and surpass "more skillful" players.
4- Anything bigger than 2v2 random, is meaningless. There's such a huge variance and you are mostly gonna be facing teams, that just take it with a grain of salt and accept the YOLO.
Posts: 493
Can you farm ELO by playing many games in different unranked arranged teams? Usually 1st game in new team matches you vs complete noobs, so e.g. player with 1k rank who also own noobs in such way will have higher ELO than another guy with equal rank who plays only random and his total W/L ratio is lower? So as a result a farming guy matches more often with more skilled players of equal ELO despite his low rank?
About decay from self experience: after 14 days of inactivity your rank drops at one moment equally to 1-2 loses, after a month your rank becomes hidden. After that you play first game as "unranked", and if you win it returns you to the pretty same state you had.
Posts: 17914 | Subs: 8
Posts: 2885
How is Elo determined? I seem to grasp how it is somehow related to rank, but also not related.
Disclaimer: Relic doesn't make their elo values for a player public, so it is not possible to accurately reverse engineer the way it works. That is why this post is based on the way chess elo system works and some assumptions coming from in game experience.
What is certain is that there are two phases of determining your elo and thus rank. First phase is placement matches when the system knows nothing about you and thus tries to find an initial place for you, so you don't have to climb all the way from the bottom. What it does looks similar to binary search withing the ladder. So it gives you average opponent first then, if you win a good player and so on. That is how you can find yourself almost at the top of the ladder if you win 8 to 10 of your first matches. Now the problem with this is that the algorithm still has to use matchmaker to find opponent for you. That means the points used to search the ladder can be far from ideal spots, causing the accurancy of search to drop. In ideal world, where all the right players were searching at the same time as you, such algorithm could possibly determine a placement with accurancy of 2-4 places, depending on the ladder. In reality it can go much further than that.
The other possibility here is that it starts with some middle elo number, like a median and gives you huge K for first 10 matches (as explained further). That also explains how you can end up anywhere in the ladder and how it is so inaccurate.
After your initial placement is determined, you are given an elo number found by the first phase and you will climb or fall in much slower fashion. After you find a match, the average elo of both teams is used to determine probabilty of win for both parties. Then, after you finish a match this probability is used to adjust their ranks. If you win the result is 1, if you lose the result is 0.
Then your new_rank = old_rank + K(result - probability).
For example if your probability of win was only 0.125 becouse your opponent was much better, your old elo was 1000 and K is 32, then your new elo is 1000 + 32(1 - 0.125) = 1028 if you win and 1000 + 32(0 - 0.125) = 996 if you lose. Mind that we don't know what is the actual K that relic uses. In chess it varies from 16 to 32. But in fact it doesn't really matter that much as we don't know the scores either.
Posts: 318
Converse this with my OKW, I somehow have top 100 ranking, but every team I am on is very balanced, and the matches are what I would consider evenly made/fought.
There is also almost never any AFK or drops in the OKW matches, virtually every match with my OST has AFK, droppers.
Posts: 2885
As for your after game stats, mind that although they say something about your performance, they say nothing about how well you helped your team to win. For example in team games, as okw has no access to caches, ostheer players who choose to build fuel caches for the whole team often happen to have low damage at the end but will win a lot of matches. On the other hand a player that does a lot of damage but doesn't support the team at all might have very good stats at the end but he and his team will lose either way. That a possible source of your problems.
In the end, your elo will always represent your true skill after enough matches and the matchmaker will always try to find fair matches. Even if you think it is your opponents who did bad, remember that if you improve yourself, you will statistically win more matches and go up in rank either way.
Remember that the whole idea of elo system is to determine your ranking using only your wins and losses. That means, you shouldn't use in game performance to estimate other people's ranks, as it is not where the source of that ranking is. If you want to know the ranks of your allies and opponents, it is better to use a piece of software called CELO.
Posts: 318
Just trying to figure out why factions with poor rank equates to horribly balanced/organized matches, while high rank generally receives competitive/well balanced matches IE the game is enjoyable with high rank but very irritating with low rank.
Posts: 186
I'm often the only person in the entire team that builds caches in OST games, something I have noticed is that the worst performing member of the team is always the person next to me (in terms of splitting a 4v4 map in half between two players each vying for a fuel/muni/VP on the sides. I know stats don't tell the whole story, but when you have multiple times the amount of the other team members across the board, that does speak to having to carry the entire match, over and over, especially the point capture stats.
Just trying to figure out why factions with poor rank equates to horribly balanced/organized matches, while high rank generally receives competitive/well balanced matches IE the game is enjoyable with high rank but very irritating with low rank.
It might be that you are letting your teammate down at important times so he does take losses and gets a hard time recovering while you are building caches. In my personal experience while playing as axis or allies it is always the easiest way to win (especially when aiming for faster tech then the enemy) to not improve your teams, but to deny the enemy teams ressources: because a single sector more then a fair split will give you +6 fuel and + 10 ammo compared to a single cache that yields +3 fuel or +5 ammo more then the enemies income.
If you have problems achieving breakthroughs in the mid or early game i would recommend using smoke + flamers or light vehicles depending on the map and faction.
Posts: 2307 | Subs: 4
I'm often the only person in the entire team that builds caches in OST games, something I have noticed is that the worst performing member of the team is always the person next to me (in terms of splitting a 4v4 map in half between two players each vying for a fuel/muni/VP on the sides. I know stats don't tell the whole story, but when you have multiple times the amount of the other team members across the board, that does speak to having to carry the entire match, over and over, especially the point capture stats.
Just trying to figure out why factions with poor rank equates to horribly balanced/organized matches, while high rank generally receives competitive/well balanced matches IE the game is enjoyable with high rank but very irritating with low rank.
Because the game tries to match you with whomever is searching for a game at the same time. It would work better if there were more players playing.
If your rank with random 4v4s is low you're kinda screwed. It's pretty much entirely random who is going to have the worse teammates.
Back when I played 4v4 randoms a bit for fun, I was playing soviets and doing retarded spam strats to mess around. Tanked my elo there. It's pretty much unplayable now, everyone it matches me with is god-awful. It would require insane amounts of effort to raise my elo to get out of those ranks. It's entirely dependent on which team gets lucky and has the least bad players.
Moral of the story is start out good
Posts: 318
Because the game tries to match you with whomever is searching for a game at the same time. It would work better if there were more players playing.
If your rank with random 4v4s is low you're kinda screwed. It's pretty much entirely random who is going to have the worse teammates.
Back when I played 4v4 randoms a bit for fun, I was playing soviets and doing retarded spam strats to mess around. Tanked my elo there. It's pretty much unplayable now, everyone it matches me with is god-awful. It would require insane amounts of effort to raise my elo to get out of those ranks. It's entirely dependent on which team gets lucky and has the least bad players.
Moral of the story is start out good
This is exactly what happened, in the first 10 games (these are with a smurf account) in Soviet and OKW, no one dropped, no one was AFK, everyone in the match was a good player, the result of which was high rank after the first 10 matches (Soviet had 100% win rate, OKW had 70% win rate). When I did this with OST, the first 4 games had AFK players, probably 6-7 of them had people on my team that dropped after 1-2 minutes, the result of which was a 10% win rate and rank 2500 or so....now if I play Soviet or OKW in 4v4, they are mostly enjoyable, but every single game with OST is anal cancer, and I mostly lose the matches, further sinking rank, making them progressively more and more cancerous.
Posts: 2885
In effect you have to play quite a lot of matches to get out of there. You can do that in as little as 20 too, but you need to make sure that you are actually much better than your current position in the elo ranking. So I would recommend playing all the other factions and game modes till you are certain you are good enough and then play exclusively the mode and faction you are in hell with till you get out of that mud.
Posts: 2307 | Subs: 4
This is what players call "elo hell". When you find yourself low enough in the ladder, you will find so many leavers, afk players and trolls that your win depends first on luck and only second of your skills. That upsets the bases of elo system as it assumes players play to win.
In effect you have to play quite a lot of matches to get out of there. You can do that in as little as 20 too, but you need to make sure that you are actually much better than your current position in the elo ranking. So I would recommend playing all the other factions and game modes till you are certain you are good enough and then play exclusively the mode and faction you are in hell with till you get out of that mud.
Or don't play randoms and get an arranged team, which is a separate elo rank. Playing random team games is just pulling teeth with the randomness.
Livestreams
111 | |||||
29 | |||||
10 | |||||
64 | |||||
53 | |||||
13 | |||||
8 | |||||
1 | |||||
1 | |||||
1 |
Ladders Top 10
-
#Steam AliasWL%Streak
- 1.829222.789+35
- 2.34857.859+13
- 3.587233.716+3
- 4.1095612.641+19
- 5.882398.689+4
- 6.280162.633+8
- 7.996646.607-1
- 8.379114.769+1
- 9.300113.726-1
- 10.717439.620+1
Replay highlight
- cblanco ★
- 보드카 중대
- VonManteuffel
- Heartless Jäger