0K 1v1

Page of 2 (37 records)

sort
[SrOhkay]FlorisXIV 13 years ago So why do or don't you play those. what maps do you like in 1v1. Would you like your ELO rank for 1v1s displayed separately. I like 1v1s a lot much more then big team games. The 1v1 are fast hectic and you can use different kinds of tactics because you have more space available to move around. A lot of maps. The ones that come to mind now. Archers' Vally Red Comet Badlands is cool, though a bit small Comet Catcher :-) (...there are so many more these are a couple of old spring maps XD) Yes, different ranking for 1v1 would be good (but a lot of work prob). Big team-games are a different sort of game the 1v1. They emphasize the players mechanics and multitasking skills a lot more. (maybe that's why ppl dont play them as often) I can safely say that i suck at team games and do reasonably well in 1v1 About rushing. There are so few 1v1s that all the different rushes with their counters haven't even been tried yet. Ever rushed a reaper or jack on Altair Crossing or a commander assault on Brazilian Battlefield? +0 / -0
Licho 13 years ago If all people were using zero-k lobby i could implement quickmatching and make 1v1 happen for people who like it. +0 / -0
[SrOhkay]FlorisXIV 13 years ago that sounds awesome Licho, but, the community cant support a 1v1 automatch system. It would be a wasted effort atm. Just a dedicated 1v1 host with it's own ranking may be better at this time. +0 / -0
Licho 13 years ago We have dedicated 1v1 host .. And quickmatching wont be for 1v1.. it would be everythingmatching.. Each player sets what types of games he likes and it then matches people according to their preferences. +0 / -0
sprang 13 years ago A separate 1v1 elo/weight does sound good, and a separate 1v1 ladder. This would be an avenue to compete and climb up the ladder, without affecting teamgame balance. Perhaps free-for-all could also use this same elo, when it's played with 1-man teams. If this was the case, it could be called something like "personal elo" instead of 1v1 elo. +0 / -0
Licho 13 years ago No plans to implement such a thing. Elo is elo. +0 / -0
CarRepairer 13 years ago Elo is Elo. But it should still be per-game. ZK Elo should be isolated from springtanks Elo. +0 / -0
luckywaldo7 13 years ago Elo is elo, and ZK 1v1 and ZK team games are about as vastly different from each other as Spring Tanks is from either, about as different as Starcraft is from anyway. Skill in one is not equivalent to skill in the others. There would be some correlation to be sure but the margin of error is huge to try to rank everything with the same number. +0 / -0
stimUpgradedMarine 13 years ago How is elo stored at the moment? Does it get re-calculated from all the battles you've played everytime a lobby asks for the elo, or is it a set value that updates after each game? (just asking) +0 / -0
Licho 13 years ago Its stored in central db and cannot be recalculated because rules changed and initial elo wasnt same for everyone.People started with elo from slringie legacy db of 3 year history. +0 / -0
NeonSturm 13 years ago If every one is using ZK-lobby? Make a linux-lobby, remove all bugs which are worse than the SL-bugs ( and SL has many bugs ), and I will use it. --- About elo: I would support a more variable elo system: you have so many different categories: * space * mexes * surprises * income space means how large the map is, and your skill to guard a large area in contrast of your skill to protect a small area. * on small maps, you can use porc more efficient ( especially chainsaws and screamer ) * on large maps, you can benefit more from surpise attacks and having the right attackers / counters where you need them income is assoziated with the map size, and determinates how much metal can move how fast to your bases on how many different battle locations and if have more than one thread in your head. * are you attacking always the same spot and wasting metal there? * or only trying to let your enemy porc there too much to fit the attackers so that your enemy waste so much metal there that you can overrun him otherwhere? * how fast can you move your constructors to reclaim metal? * do you spend more metal in cons than they get out of the wreckages? * cloaking and stealth is more efficient on large maps, but more expensive - can you decide about the best map related strategies better then your enemy? the number of mex-spots is also dependend how much surprises are possible. * you can't defend a large front with defenders. * big maps and spaced porc will be killed with a single hammer and one or two warrior against raiders, but not too many groups that they can be killed with a single strategie easily. * more units and less porc means you can counter it by destroying mexes with fleas * can you defend your radar spam, or do you make less adv. radars? .... ELO IS NOT A NUMBER * we need at least to sort the maps and implement different elos. * the different elos should get multiplied with a percentage how much they fit the map. * get the sum of all elos and divide it through the sum of all percentages. * finally you got a map / playercount / whatever specific elo rating for each battle. --- A single elo is for chess - chess is a game which have * always the SAME map size * always the SAME 'income' * always the SAME 'terrain' * always the SAME player count * »ONE« limited »NUMBER« of tactical combinations * always the same elo factor - factor 1 ;) +0 / -0
NeonSturm 13 years ago One Example Implementation: Player count 1 to 10 PER TEAM ( think about the time we could have up to 16 players per team ) PlayerCountElo { lower 1400, higher = 2000 } MapSizeElo { smaller 2150, bigger = 1250 } percent = 0 current = 0 --- Battle has 3 players per team. factor_higher ( playerCount % of maxPlayers 3 / 10 * 100 = 30% ) factor_lower = 100% - factor_higher current += ( lowerElo * factor_lower 1400 * 70% 980 ) current += ( higherElo * factor_higher 2000 * 30% 600 ) percent += 100% --- Map has 4v2 ( Spooler Buffer Beta - cool down - it is only an extreme example ;) ) factor_bigger smaller / 64x64 * 100 6%x3% > lol? 2 numbers? wtf! ;) > figure out if you start at the larger edge or at the smaller one! > divide the larger/smaller number through the player count. > multiply both numbers to get one percentage factor_bigger 6% * 3% / 3ppls 9% factor_smaller = 100% - factor_bigger current += ( smallerElo * factor_smallerMaps 2150 * 91% 1962 ) current += ( biggerElo * factor_biggerMaps 1250 * 9% 113 ) percent += 100% --- now you have a current of 1580 for PlayerCount and 2075 for MapSize your percentage is 200% at all. if you now compare the current/percentage = avgElo from my calculations with the average of the original elo values you can notice a difference. current/percentage ( 980 + 600 + 1962 + 113 )/200% 3655/2 = ~1828 avg(elos) ( 1400 + 2000 + 2150 + 1250 )/4 1900 --- the negative difference of 72 is low, but what is if this player plays a game with 8 players on a small map? it would be: > current/percentage ( 280 + 1600 + 1962 + 113 )/200% 3955/2 = ~1978 > avg(elos) ( 1400 + 2000 + 2150 + 1250 )/4 1900 the difference is higher, because we calculate with ( 140020% + 200080% ) instead of ( 140070% + 200030% ) for the playerElo now we have a skill difference of 150 elo by a situation with is about 25% different from the first one. I think I have explored a possibility to get an elo difference of up to 500 - only by adding the individual skills in the bad and good situation for this player. In very extreme cases your enemy can be skilled completely in contrast and get effective 500 in the other direction - a possible error of 1000 elo of our current system ( in really extreme situations which are maybe impossible to simulate with zk, a player can get the average of the elo-differences in all categories more or less - 1200 difference ). Because of clan balances we have not really random teams. if we take into account that players differs only about 300 instead of 600 per skill ( average ) we still have up to 600 elo mathematical difference. --- I suggest a balance which eliminates an 150 elo error and randomize clan balanced games more than we can do it with the current algoritm +0 / -0
luckywaldo7 13 years ago Individual elo for every combination of game, game type, map, and everything feels like serious overkill. Nobody expects elo to be a super-accurate measurement of skill or anything, it is just to help games be balanced enough that everyone wins and loses about 50/50, so everyone has a chance to have fun. +0 / -0
NeonSturm 13 years ago did you read all my text? it is because of some games which are completely fail balanced. This can decide about a lost homeworld, lost even up to 20k credits in buildings or 900 IP per bomb run because you lost your front / didn't get credits for the defense. Also I have written about alternating teams, which means, that the server can try to balance it for another map and play there if it can't balance otherwise. +0 / -0
luckywaldo7 13 years ago I didn't read through all your math because math is completely meaningless if the numbers that go into it are meaningless. You start with the assumption that a player in a smaller team is -100 and in a larger team is +500, and in a smaller map is +650 and -250. Differences of 100 are pretty noticeable Ion, and you are throwing out differences in level of 600 and 900 of performance from the same player. There is a difference of maybe 900 from the very best and the very worse players recorded in Zero-K. That is utterly absurd. Sure, some people might be a bit weaker in certain scenarios, but you are going to see a pretty strong correlation. Not to mention, a lot of your math doesn't even make sense. Why are you making a calculation based on hypothetical max number of players? The number could be changed to 128 max with still 3 players in the game, so nothing actually physically changed, but your calculations would change. Your formulas just don't have any physical meaning. +0 / -0
luckywaldo7 13 years ago Which isn't to say I'm set on disagreeing with you Ionstorm. I think you might have a point but you need math that makes more sense, less values and formulas pulled out of nowhere. +0 / -0
NeonSturm 13 years ago and now you are going to throw out a extreme number : 128. With 10 I would avoid, that "1" get's completely ignored - maybe we should change the minimum to -1 to compensate it in the other direction. I spoke about a theoretical possible difference of ( personalAverageSkillDifference 2 ) between teams at a 1v1 game compared to 8v8 games on a map sized completely in contrast to this map. Also I have got a maximum of 1200 with 600 elo difference. If most players have an average of 300 skill difference between the worst and best situation, then they get at average an error worth 150 elo in the current balance. If you want to play 2v2 on PW, you could get a 1200 elo error ( 300players ) in the worst case - avg error would be 300. In >=5v5 games it should be more compensated through alternating games and only be 150 ( half of a player ). We play >=5v5 in PW games many times. Also I sayed, that the current system works with a single number, and can produce only a single way to balance. * Each game changes the elo difference of teams by about ø8-12 elo / player * That will be compensated with an exchange of 1 newbe ( newbe have low weight ) * That lead to never changing teams A new skill based calculation can not only make it possible to play at another map if it is unbalanced at one map, it also allow to shuffle the teams for each new map / team size. +0 / -0
luckywaldo7 13 years ago It sounds like you are assuming flat probability of performance, which would give a player with 1500 elo and +/- 150 elo difference an equal probability of performing at 1350, 1500, or 1650, but zero probability of playing at 1349 or 1651. That is silly. In reality, the distribution is probably normal. In which case, error actually decreases as team size increase. So the larger the team game you play, actually the more balanced it is. +0 / -0
[DE]Rick 13 years ago Iam pretty good in teamgames but i really suck in 1v1's. They are totaly diferent games. Its like playing a diferent spring mod. +0 / -0
NeonSturm 13 years ago what's about the team shuffle? I think we could need that ^_^ What's about the ability to play on a different map if it is needed to balance better? * 3 ships here, 3 ships there - tale this planet where you can balance better. * >5 ships - allow +-1 ship. > 9 dhips - allow +-2 ships for the map script +0 / -0