Did the Elo scores of Arena Teams change after each fight?
Ignie Ferroque translates from latin to "with fire & with sword." It is a stock phrase used to describe the results of a destructive raid into an enemy's territory, whose sole purpose is to generate fear, terror, and destruction.
Where are people getting that the starting rating is 1600 from? I know someone threw it out there as how another ELO system handles things, but nothing from beta testers using the 70 arena.
The screenshot on the WoW website seems to imply you start off with 0 rating. If that is true, all the theorizing on how to prevent people from benefitting from dissolving their team and reforming is pointless.
Please point me to the post or even better, if someone doing the rated matches in beta can give us some hard numbers to work with.
I think the huge number of people running Arenas nonstop without any rewards is proof enough that there are enough people that PVP for PVPs sake
I'm not positive that it proves anything. For a lot of people, it's something to do that's better than sitting in the Military Ward of IF for 9 minutes waiting for AB to pop again (or 15-18min for AV)... my farming options are pretty limited at those kind of intervals, mostly consisting of Searing Gorge for Dark Iron or maybe Burning Steppes, depending on queues.
I've seen lots of people that chain sign up for Arena without going to a BG. When you get matched 3 times in a row with the same guy, you know he aint going to BGs
Plus people that don't need PVP rewards go to Arenas.
Our guild (rerolling on a new server come TBC) has as its focus playing to have fun in everything we do.
Thus we have a very relaxed approach to arenas: let folks form whichever teams they want, and duke it out if they do meet up in matches. May the better team win, but let both teams learn from the experience.
Collusion would be frowned upon, because it's unethical and detracts from the fun of the game. Do you want to win knowing your win is only because your guildmates gave up for you? Also, with the length of the season and the variety of teams, I doubt it'd be easy to tell one team to always lose to another... why should they? Let the teams fight it out and learn from each -- improving both teams will help the guild as whole, since there'll be hundreds of arena teams in a battlegroup, all competing against each.
We also don't plan to have a designated A-team and a B-team, because that leads to drama.
As for PvP or PvE specs, since we focus on having fun, we'll let folks spec whatever they want, so long as they pull their weight. I plan to be specced purely PvP (33/28) at 70 and am happy to eschew main tanking in favor of a guildmate who'll put some points in Prot. We're a guild, a team, a family... and the idea is to have fun throughout all aspects of the game, not to get into each other's hair over stuff that's ultimately irrelevant, like PvE progression or PvP arena rewards :)
That's just our take.
I know my guildmates are smart, nice, skilled people -- that's a precondition for being in our guild. And thus we trust each other enough to sort out teams and to feel the love across the guild, without getting caught up petty rivalries. We have no plans to min-max arena team composition by telling people who they'll play with, because the best arena teams will be the ones that have the most internal synergy -- at the top levels of play (where we aspire to be), spec and gear will become equivalent and the difference will be in how well a team works together. There, the teams that are comprised of friends who genuinely enjoy playing with each other will beat teams whose only bond is necessity, or a guildmaster's writ.
Thus I wouldn't worry about templating at all and I'd let people play who they want to play with, and how they want to play. I know if I was a guild member and was told what to spec and who to play with, I'd resent that because I'd much rather have a community of equals working together towards excellence... bottom-up management rather than top-down dictatorship, if you will :)
Where are people getting that the starting rating is 1600 from? I know someone threw it out there as how another ELO system handles things, but nothing from beta testers using the 70 arena.
The screenshot on the WoW website seems to imply you start off with 0 rating. If that is true, all the theorizing on how to prevent people from benefitting from dissolving their team and reforming is pointless.
Please point me to the post or even better, if someone doing the rated matches in beta can give us some hard numbers to work with.
In Beta, you start at 1500 rating (for 2v2, 3v3 and 5v5 teams).
Originally Posted by Igni
Did the Elo scores of Arena Teams change after each fight?
Yes.
Originally Posted by Kalman
Another option yet is to simply introduce a cooldown on how often you can form an arena team, one which is set for every member of the team at time of formation.
It does cost 50g to form an Arena team...
Originally Posted by Elerion
It is my understanding that you can't drop or add players to a team mid-season, to avoid rating boosting and FOTM abuse. Making rating player based on a split (and subsequent reform) will make that possible.
Incorrect, at least on Beta. We've been able to have people switch teams, etc. However, remember that a player doesn't have his or her own arena rating -- only the rating of the team he or she is on. But! You have to play on at least 30% of the teams' matches to even get a rating from that team. That's a pretty big limiting factor, as I see it.
Originally Posted by Copernicus
A. We don't know if collusion will be viable or not. It all depends on how the matchmaking works. If two highly-ranked teams can guarantee that they'll play each other 10 times in a row, because they're the only teams in their point brackets online, they can easily collude with each other to avoid playing any real teams that week. There are ways to prevent that, but I don't know if any of those are implemented.
Except... diminishing returns. Remember, ELO adjusts your rating based on your expected rating. Thus if you consistently lose a team, you lose fewer and fewer points with each loss (since your rating goes lower and lower) and they gain fewer and fewer points with each win (since their rating is relatively higher and higher to your rating). From what we've seen on Beta (not from throwing games, but just from losing them to Warlock/Mage combos as a Warrior/Paladin duo :P) two teams can pretty much only move each other +/-40 points or so.
Also, remember that the team benefitting from thrown matches will now be at an inflated rating, and unless they have the skill to preserve that rating, they'll lose it very quickly -- the system is self-correcting in that sense.
Incorrect, at least on Beta. We've been able to have people switch teams, etc. However, remember that a player doesn't have his or her own arena rating -- only the rating of the team he or she is on. But! You have to play on at least 30% of the teams' matches to even get a rating from that team. That's a pretty big limiting factor, as I see it.
30% of the week's games or 30% of the team's total games? That's quite a difference.
Originally Posted by Copernicus
Also, remember that the team benefitting from thrown matches will now be at an inflated rating, and unless they have the skill to preserve that rating, they'll lose it very quickly -- the system is self-correcting in that sense.
Unless the teams they face are also at inflated ratings - thus the term inflation ;p Still, it's good to know about the 50g formation fee. That's a decent barrier to the creation of dozens of throwaway teams.
Incorrect, at least on Beta. We've been able to have people switch teams, etc. However, remember that a player doesn't have his or her own arena rating -- only the rating of the team he or she is on. But! You have to play on at least 30% of the teams' matches to even get a rating from that team. That's a pretty big limiting factor, as I see it.
30% of the week's games or 30% of the team's total games? That's quite a difference.
I don't get all this talk about 10 games per week maximum. This thing with stopping at 10 games per week will only apply if you are happy with your current ranking. If you do not think that more games will get you a higher ranking, yes you could stop after the 10th game, or play more which shouldn't hurt you either.
As long as you think that you either currently are better than your current rating indicates or that you can become better if you play more and train more, you should definately continue to play games. The 10-game _minimum_ is only to prevent that teams that are no longer as good as their current rating keeps getting points off it.
I don't get all this talk about 10 games per week maximum. This thing with stopping at 10 games per week will only apply if you are happy with your current ranking. If you do not think that more games will get you a higher ranking, yes you could stop after the 10th game, or play more which shouldn't hurt you either.
As long as you think that you either currently are better than your current rating indicates or that you can become better if you play more and train more, you should definitely continue to play games. The 10-game _minimum_ is only to prevent that teams that are no longer as good as their current rating keeps getting points off it.
That's not exactly true.
You should keep playing more games past 10 if you feel that you will improve your w/l ratio (from those 10 games) by playing more.
IE:
You've played 10, won 8, lost two. The two losses were big mistakes which you know you wont repeat. So you play another 20 games and win 19 of them. Wining 27 and losing three is better than 8/2 (not always, but lets just generalise).
If you get lucky and win 10 games but don't think you can keep it up you should stop.
If you win 10 games but do think you can keep it up you *might* want to continue.
Basically: If you think your 10 games is better than your level don't play more, worse, do play more, or equal, *maybe* play more.
This last point is contingent of the exact specifics of the system.
I assume the system currently works like this:
1. The system averages your rating gain/loss to a ten game span. Ie: If you played 100 games it divides your total ranking gain or loss by 10 to work out how much you move up or down at the end of the week.
2. Your ranking is not updated on the fly but once a week. So you will not be playing higher ranked people until the next ranking week, and thus not able to gain a higher average from facing stiffer competition until that time.
Thus there is no reason to play more than 10 games if you win them all.
If the second point is incorrect there will be reason to play more than 10 games a week if you feel you are better than your current ranking, even if you've won all your games, but I don't think it is.
If the first point in incorrect then the system is not casual friendly and Blizzard are very stupid.
Interestingly this does allow for some collusion, if people are bothered enough:
Two teams.
Team A: Arena point farming team.
Team B: Skilled arena team.
Team B plays Team A 10 times and loses them all. Team A doesn't play any more games for the week.
Team B plays another 10 games and wins them all, and thus stays at the same rank. Or plays many more games and gets a winning average and continues to rank up also.
Thus only Team B needs to be good. But both teams will be able to maintain a high rank off the back of one team.
Of course for this to work at a high rating level Team B will have to be very good indeed.
Your rating then translates into arena points every week.
Thus once you're a few weeks into the season and have "found your rating," so to speak, you'll want to play the ten games and play them at the top of your game (winning them, ideally) so that you can move up in your rating, without risking going down.
It's only worth playing more games if your team hasn't found its rating/relative "rank" yet.
It was speculation at the time, but it turned out to be pretty accurate.
--
Team A: Arena point farming team.
Team B: Skilled arena team.
Team B plays Team A 10 times and loses them all. Team A doesn't play any more games for the week.
Team B plays another 10 games and wins them all, and thus stays at the same rank. Or plays many more games and gets a winning average and continues to rank up also.
Thus only Team B needs to be good. But both teams will be able to maintain a high rank off the back of one team.
Of course for this to work at a high rating level Team B will have to be very good indeed.
Your rating change is based on the rating of the opponent you beat; it's a statistical adjustment based on whether the probability of you winning turned out to be the actual truth of the matter -- if the system didn't expect you to win, and you did, you'll get a large rating boost. Conversely, if the system did expect you to win, and you didn't, your opponent gets the large rating boost and you get a large rating loss. Repeated wins against the same opponent give less and less rating, since the system has adjusted the two ratings to attempt to predict the outcome of the battle. If the system expects you to win and you win, your rating change will be minimal, if any.
You can't "farm" arena points, because you can't "farm" rating. Playing 500 games does you no good, unless you win 500 games. This is not the honor system of old.
For some uber godly team that never loses, there’d be no reason to ever stop playing, since they only win and thus can’t lose… which means their ranking goes up and up (though by less and less and less each time, since they never lose and their rating will be eventually higher than everyone’s and you don’t get very many points for beating someone the system expects you to beat). If ever they did lose, their rating would get hit hard, since they’d be so far above everyone else that the system would start to correct their rating downwards with any losses.
Collusion is ultimately a non-issue because:
- Good teams don't need collusion to succeed -- the rating system already allows them to succeed
- Thus only bad teams need to collude to win, and collude by asking good teams to let them win
- But this creates more "work" for the good teams, which means they have to lose to the bad teams, then work up their rating again each week... and not just to the level it was at before, but to a _higher_ level, so that the bad team can then basically flip places with the good team through the collusion, and then the good team has to repeat the ordeal by going up again ever higher...
To illustrate:
Week 1:
G: (1500)
B: (1500)
After Collusion:
G: (1450)
B: (1550)
End of Week 1 (after Good team "catches up again")
G: (1550)
B: (1550)
Week 2 After Collusion:
G: (1500)
B: (1600)
Now Good Team must fight its way up to (1600) again.
The catch is that the higher rating you get, the harder it is to move up rating... so Good team will end up doing a _lot_ of work for bad team. Also, a single loss by bad will undo a lot of collusion, which means Good team will have to work even harder... and since you can't perfectly pick your opponent (even if you queue at the same time, it's across an entire battlegroup and you might get unlucky), it turns into a very time-consuming effort with all the rewards for one team and just a lot of extra work for the other team.
I'm sure some folks will go to all the trouble of doing it, but I have no concerns about it because it won't be present at the top levels (mostly because it simply isn't sustainable).
One interesting thing that I thought of this morning. What you have in WoW that you don't have in either chess are classes. One class composition may prove to be strong against another class composition but be weak against a third class composition. So you might see a greater variability in wins and losses when teams of equal rating face off.
In the long run, it will only matter if you're a class composition that's weak against a type of team that has become common across your battlegroup.
Ignie Ferroque translates from latin to "with fire & with sword." It is a stock phrase used to describe the results of a destructive raid into an enemy's territory, whose sole purpose is to generate fear, terror, and destruction.
One interesting thing that I thought of this morning. What you have in WoW that you don't have in either chess are classes. One class composition may prove to be strong against another class composition but be weak against a third class composition. So you might see a greater variability in wins and losses when teams of equal rating face off.
In the long run, it will only matter if you're a class composition that's weak against a type of team that has become common across your battlegroup.
In Guild Wars, this concept was referred to as the "metagame" -- what most people are playing.
A mini-example using 2v2 WoW teams (2v2 is the least balanced)
Rock beats Paper, but Scissors beats Paper, but Rock beats Scissors.
Suppose that the current metagame is very Paper-heavy. If you're playing Rock, you're going to be losing a lot of games. On the other hand, if you play Scissors, you'll win a lot of games, just by virtue of having the counter to the current metagame. Generally, this will cause the metagame to shift towards Scissors, because all the Scissors players will come out of retirement, so to speak, since this is their hour.
In a few weeks time, the metagame might be Scissors-dominated, because Paper will have cycled out, since it was losing so horribly to Scissors. At this point, Rock will re-enter the metagame.
While this is a drastically simplified, I think we'll see some measure of "metagame trends" in competitive arenas.
Hold scrimmages NOW while Arena's dont mean anything to determine who will play on your teams for real points later. Operate it like a coach, not like a guild leader. If everyone accepts that they've earned their spot on Varsity, JV and "Freshman" teams, and people are given the opportunity to move up or down based on attendance and performance, then things will work themselves out naturally.
Your upbeat assessment as well as everyone elses helps assuage my concerns about intra-guild competition. In regards to your specific suggestion, I've been avoiding Arenas aggressively. The reason is because the Arenas played now will be different fundamentally from Arenas played at level 70 due to gear, new spells and new abilities.
Not only was I concerned about animosity between guild members right now, I'm also concerned about people getting prejudiced for or against certain team templates based on inaccurate testing.
I am not sure how you expect to compete by taking your best players and dividing them up into 2 "well-balanced" teams. It's fun to hope, but you really have to be realistic. Trying to balance things so people don't get jealous, or feel left off of the "good" team etc will lead to weaker teams overall. It sounds like you are very competitive and so are those in your guild.
Even intramural grade school basketball has starters and 2nd stringers. If your members can't handle it, it sounds like they will likely not be mature enough to progress through an arena season without infighting.
You may think to yourself, "Oh, they are all good enough to be on the top team so both teams will be huge domination !!" ...but it sounds like you are setting yourself up for dissappointment.
I recommend narrowing your players down. "Dividing up your skill" won't get you anywhere. You may not think theres much of a difference between your top ten players in your guild, and your 11-20, but you will be surprised.....and dissappointed.
My former guild considered itself a "PvP" guild, and the consistant wins and success did not come until we decided to stop treating our members like children(didn't want to hurt feelings), and define exactly who was considered a "starter", and who would give up their spot in a heartbeat, the moment any "starter" wanted into the group. The difference in a team with similar class composition who had 9/10 starters and one who had 5/10 starters with guildy fillers who we all considered "very strong PvPers" was mind-boggling. You can go on and call this an exception or whatnot, but we found this true with most opposing guilds on the server. We knew the names of the players on the 'A' team, and when their team would lack a few of them, even their "geared, skilled" replacements wouldn't make their team even as CLOSE to as deadly as when they have all of their starters. Some people are good at PvP, and some people are 100% consistantly on the top of their game, making the correct decisions at every turn. We didn't realize the difference until we ran into top teams and started having a team going for a majority of the hours in the day. We still don't consider the non-starters "bad" by any means. They are great players, but you simply must accept that "good" is not "great", and in WoW, just like many sports, there is a skill gap that can't always be easily seen.
I just don't see guilds who try to form "balanced teams" having any chance in hell against the top teams of their realm(unless its tiny and you are only guild with epicz or something).
Your rating then translates into arena points every week.
Thus once you're a few weeks into the season and have "found your rating," so to speak, you'll want to play the ten games and play them at the top of your game (winning them, ideally) so that you can move up in your rating, without risking going down.
It's only worth playing more games if your team hasn't found its rating/relative "rank" yet.
It was speculation at the time, but it turned out to be pretty accurate.
--
Team A: Arena point farming team.
Team B: Skilled arena team.
Team B plays Team A 10 times and loses them all. Team A doesn't play any more games for the week.
Team B plays another 10 games and wins them all, and thus stays at the same rank. Or plays many more games and gets a winning average and continues to rank up also.
Thus only Team B needs to be good. But both teams will be able to maintain a high rank off the back of one team.
Of course for this to work at a high rating level Team B will have to be very good indeed.
Your rating change is based on the rating of the opponent you beat; it's a statistical adjustment based on whether the probability of you winning turned out to be the actual truth of the matter -- if the system didn't expect you to win, and you did, you'll get a large rating boost. Conversely, if the system did expect you to win, and you didn't, your opponent gets the large rating boost and you get a large rating loss. Repeated wins against the same opponent give less and less rating, since the system has adjusted the two ratings to attempt to predict the outcome of the battle. If the system expects you to win and you win, your rating change will be minimal, if any.
You can't "farm" arena points, because you can't "farm" rating. Playing 500 games does you no good, unless you win 500 games. This is not the honor system of old.
For some uber godly team that never loses, there’d be no reason to ever stop playing, since they only win and thus can’t lose… which means their ranking goes up and up (though by less and less and less each time, since they never lose and their rating will be eventually higher than everyone’s and you don’t get very many points for beating someone the system expects you to beat). If ever they did lose, their rating would get hit hard, since they’d be so far above everyone else that the system would start to correct their rating downwards with any losses.
Collusion is ultimately a non-issue because:
- Good teams don't need collusion to succeed -- the rating system already allows them to succeed
- Thus only bad teams need to collude to win, and collude by asking good teams to let them win
- But this creates more "work" for the good teams, which means they have to lose to the bad teams, then work up their rating again each week... and not just to the level it was at before, but to a _higher_ level, so that the bad team can then basically flip places with the good team through the collusion, and then the good team has to repeat the ordeal by going up again ever higher...
To illustrate:
Week 1:
G: (1500)
B: (1500)
After Collusion:
G: (1450)
B: (1550)
End of Week 1 (after Good team "catches up again")
G: (1550)
B: (1550)
Week 2 After Collusion:
G: (1500)
B: (1600)
Now Good Team must fight its way up to (1600) again.
The catch is that the higher rating you get, the harder it is to move up rating... so Good team will end up doing a _lot_ of work for bad team. Also, a single loss by bad will undo a lot of collusion, which means Good team will have to work even harder... and since you can't perfectly pick your opponent (even if you queue at the same time, it's across an entire battlegroup and you might get unlucky), it turns into a very time-consuming effort with all the rewards for one team and just a lot of extra work for the other team.
I'm sure some folks will go to all the trouble of doing it, but I have no concerns about it because it won't be present at the top levels (mostly because it simply isn't sustainable).
Thanks for the details.
Isn't the situation you describe just the one I described (without the general guesstimates I put for numbers)? :P
And ranking updates on the fly? That's rather annoying. That means at some point you will have to play a lot of games if you want to improve your rating quickly. Blizzard needs to make this clear.
One interesting thing that I thought of this morning. What you have in WoW that you don't have in either chess are classes. One class composition may prove to be strong against another class composition but be weak against a third class composition. So you might see a greater variability in wins and losses when teams of equal rating face off.
In the long run, it will only matter if you're a class composition that's weak against a type of team that has become common across your battlegroup.
In Guild Wars, this concept was referred to as the "metagame" -- what most people are playing.
A mini-example using 2v2 WoW teams (2v2 is the least balanced)
Rock beats Paper, but Scissors beats Paper, but Rock beats Scissors.
Suppose that the current metagame is very Paper-heavy. If you're playing Rock, you're going to be losing a lot of games. On the other hand, if you play Scissors, you'll win a lot of games, just by virtue of having the counter to the current metagame. Generally, this will cause the metagame to shift towards Scissors, because all the Scissors players will come out of retirement, so to speak, since this is their hour.
In a few weeks time, the metagame might be Scissors-dominated, because Paper will have cycled out, since it was losing so horribly to Scissors. At this point, Rock will re-enter the metagame.
While this is a drastically simplified, I think we'll see some measure of "metagame trends" in competitive arenas.
There is no real metagame, not anywhere near like GW (which if you played at a high level didn't really play in the way your describe anyway - people would find something OP, win against everyone, it'd get nerfed then the search would be on for the next powerful formation).
At the high 5v5 level there might be some metagame, but at the mid and low levels it will come down to gear and teamwork. In the 3v3 and 2v2 arenas it will come down to whichever group is OP against pretty much all other groups.
I think when people are speaking about how long it will take to "find you're ranking", it's being a little pompous to presume that even the best players won't benefit from more practice - I think even once you've hit that 50/50 win/lose balance, it will be worthwhile to continue playing and strategising to work out your weak points and such. My experience in Arenas has tought me so far that there's no such thing as flawless execution.
Besides, it'd be so boring to PvP for a month then just do nothing until the end of the season :D
I think when people are speaking about how long it will take to "find you're ranking", it's being a little pompous to presume that even the best players won't benefit from more practice - I think even once you've hit that 50/50 win/lose balance, it will be worthwhile to continue playing and strategising to work out your weak points and such. My experience in Arenas has tought me so far that there's no such thing as flawless execution.
Besides, it'd be so boring to PvP for a month then just do nothing until the end of the season :D
You have to do 10 each week to gain ranking points for that week.
I think the huge number of people running Arenas nonstop without any rewards is proof enough that there are enough people that PVP for PVPs sake
I'm not positive that it proves anything. For a lot of people, it's something to do that's better than sitting in the Military Ward of IF for 9 minutes waiting for AB to pop again (or 15-18min for AV)... my farming options are pretty limited at those kind of intervals, mostly consisting of Searing Gorge for Dark Iron or maybe Burning Steppes, depending on queues.
Horde on my battle group have 1min maximum wait times and yet when I go to the arenas I meet a LOT of other horde players. So it is not just something they do as they wait for queues. It is simply fun to PvP against other people. Only problem is that we are all now balanced with level 70 in mind so its a bit biased to some classes.
There is light at the end of the tunnel.
The only problem is, it's often an incoming train.
As someone said earlier, if team A loses on purpose to team B, that just means that team A will work a lot more for the same rewards, while team B gets it for no work at all.
As I understood (and I may be wrong) win/loss points wont be the same (e.g. winer gets 20 points, loser loses 30 points) so this will mean that you cannot sustain your scheme without a LOT of work from the losing team.
12/24/06, 3:04 AM
in EJBSG 9
Similar Threads
