Elements PvP League - Suggestions and Feedback

[mrpaper]

*snip*

Actually that is incorrect and your example highlights the issues with that system.  A player who went 70-130 with a 1420 rating like you WOULD get prizes.  In Season 2 of BL that player would have gotten 6th place, which is a prize spot, despite having a rating that was better than only 4 of the 92 players.  In Season 1 his/her adjusted 1620 would have resulted in a 4th place finish despite a rating that is better than only 6 of 132 players.  Same situation with CL where the player would have finished 5th in Season 2 despite a rating better than only 4 of 35 players and 5th in Season 1 despite a rating better than only 5 of 38 players.
All you've done is create a new system that rewards spamming games more than winning, the exact problem from before.  Simply put, your method places are too much value on a game versus a win.

No it doesnt't because others score are also inflated which you forgot to calculate, so a 1620 score might mean a 20th place, but not a top 6 for sure.  It would help but wouldn't mean that much, xenocicius is in 6th place now with only 22 games played, with the 22 games he is at 1623 and will be at minimum 1631 after 30 games which is the minimum.  So a losing record won't be enough.   Then again, 200 games would be an incredibly great amount of games played and I wouldn't mind a player who is involved that much getting a 6th place  prize.

[TStar]

If you look back at my examples you'll see that I did factor in a games played bonus for all players in the recalculated standings of my example, so again it stands.

The real issue here is what Leagues should reward.  Do we want to reward skill and success, or do we simply want to reward activity no matter the results?  A player can be at the bottom of the standings all season and in the final week win 10 games against the top 5-10 players and that could be enough to boost them up to a prize spot.  I can play 150 games at around a 1500 rating and just go 5-0 against the top 2-3 players in the final week to end up with a score over 1800 which probably gets me a top 3 spot, if not a win.  At some point you have to accept that rewarding activity isn't what Leagues are about.  After having a player with 6 matches win a prize we put in the 30 game rule to make sure that players were active.  The problem is there will never be a minimum that satisfies everyone.  There are some League players who have trouble hitting 30 games, but they enjoy Leagues so they still play.  The more you raise the minimum, the more people you prevent from being able to join until you reach a point where only hardcore players with the time to play 70/90/120/etc. games can even compete.  Is that really what you are trying to advocate?  A system where only those people with excessive time to play an average of 1 or more games per day can even hope of competing?  I've played over 100 games in a single month in Season 1/2011 BL and let me tell you it's a grueling chore that takes incredible amounts of time and energy to accomplish.  I'm still burned out a bit from that over 9 months later which is a reason I have scaled way back to 60-70 games per season instead of trying to maintain a 300-game pace again.  A system that has players having to put that level of time and dedication into Leagues just to compete for top spots is something we are trying to get away from, not go back towards.

[bucky1andonly]

The whole point of ANY league ANYWHERE is to have active members participating regularly.  The more they miss, the more likely they will not be able to rejoin again, or possibly kicked out of that league.  Leagues are not there for people to pick and choose when they want to play, although with this league you actually can, but most leagues have schedules, and you are expected as a member of that league to participate.  Participation is the focus of any league.  If you don't have time, don't join.  But for this league, a minimum of 30 games is nothing for anyone that plays elements. 

There are a lot that join and quit because they are doing poorly, a lot of which are not even active on the forum.  Don't cater to these people, you should focus on those that want to take league seriously.  I personally stopped because of the stupid rule about shards causing me to auto lose a game, which I feel is a legitimate reason too.  But I will finish up my 30 just in case the league rules and ranking system is changed to accommodate those that want to play a serious league and I will join the next league season.  Save the 'fun' style crap to other pvp events and make this an official league by actually discriminating against those that have time for this.

73 bans from playing this season because they didn't finish up the required amount of games.  They obviously were not serious enough. 

1) Add minimum weekly games, not reaching it incurs a penalty on their points, no going below 0 points.
2) Add in a maximum amount of games, since the game is luck based, and prevents spamming
3) Do a simplified point system, 3 for a win, 1 for a loss, no losing of points except not meeting the weekly min

100 max, 50 min, 3pt win, 1pt loss, -10pt weekly min penalty, everyone starts at 0 points
weekly min of 5 games (over 3 months, 12 weeks is 60 games), even joining at the end of month 1, you will get 40 games min no matter what, and getting another 10 should be no problem for serious players.

It's simple, the math is basic, even losing you still have a chance because you get 1 point and any player can easily calculate how many games they need to win/lose combined to catch the leader.  With a maximum amount of games, players cannot keep spamming games.  It encourages serious players to join and participate regularly.

5 games a week is not a lot to ask of serious players either. 

[mrpaper]

3) Do a simplified point system, 3 for a win, 1 for a loss, no losing of points except not meeting the weekly min

... this was discussed as a suggestion and rejected because you cant win if you don't play a ton... but I wouldn't mind that.  Since it won't happen ... I would like if you guys consider my system.  Yes it will help people who plays more, but I believe there should be an advantage of playing more.  But tons of games played won't be enough to take you to the top.

By the way, there is a major fail in the ELO system.  It gives or takes points depending of the ELO of you're opponent... Now say I fight dracomageat who is just below .500 but as a bad ELO, If I lose against him, I lose TONS of points but he is not a bad player.  Now instead of playing him, I decide to play a noob who has less then 3 games played, his ELO is at roughly 1500 and I have an almost 100% chance to win ... does that really make sense?   Which also leads to other weird stuff like some people not figthing me because of my winning record and going mostly against those people.  ANd last but not least, do you guys really think someone who has trouble getting to reach 30 matchs should win a league when some peolpe plays over 100 games?  I also thought of something else that seems to bother TStar.... we could put an * in the rules saying you can't get a top 6 prize if you don't at least have as many wins as you have losses (play .500) so you can't only go and spam league.

[CCCombobreaker]

By the way, there is a major fail in the ELO system.  It gives or takes points depending of the ELO of you're opponent... Now say I fight dracomageat who is just below .500 but as a bad ELO, If I lose against him, I lose TONS of points but he is not a bad player.  Now instead of playing him, I decide to play a noob who has less then 3 games played, his ELO is at roughly 1500 and I have an almost 100% chance to win ... does that really make sense?   

This is perhaps the only reasonable objection to ELO.  It changes your score based on your opponents score, which assumes their score accurately represents their skill (so that your score can represent yours).  The problem then is that some players do not play at their top skill all the time (Matrim and YCH don't really belong in the basement).

 That new players start in the middle does put a pretty big target on their back, but the idea is that they are unknown and at some point TStar was 1500 this season and if I'd have beaten him then I did not get as many points as he'd be worth now.  But since the league generally has many more new players that are bad than are good (the league mean is based off a few people with a lot of wins and a lot of people with some losses - many more people <1500 than >1500) it does encourage people to pick on new players, but it is hard to inflate your score off them to the 1600+ required to get a prize because there is a limited supply of new people, and they only have a handful of games (if they are bad) before their point value tapers off and they become more of a risk.

ANd last but not least, do you guys really think someone who has trouble getting to reach 30 matchs should win a league when some peolpe plays over 100 games?e.

If someone with 30 games is in the top 6 I have no problem with them getting a prize.  Would I rather them play more games... yes!  But adding more arbitrary rules is just lame.  1 game every 3 days on average is a reasonable commitment.  And playing the games is part of the reward... so playing lots of games means you got to play lots of games, and does not mean you should have some privilege that gives you a bonus towards prizes.

[TStar]

You touched upon one of the biggest downsides to the current ELO system 3C, and it's something we've discussed at length and tried to come up with a solution.  The best option was suggested originally but RootRanger and QuantumT where we would use a KRACH system where rankings are dynamic and change as your previous opponents scores changed.  So if you beat me when I was at 1500 and I later gain in the ranks, your score would rise as well to reflect that.  Basically a players current ranking would be the determination in how much they are worth, not their ranking at the time you played them.  The problem is the formulation for this is too intensive for Googledocs or any other currently known free online application to process making the only way it would be possible to use is if all rankings were done offline by a LO on an Excel spreadsheet.  That's a system that's not realistic.  Until online database technology catches up to the processing needs of a more complex system like KRACH we will stick with ELO.

[Jenkar]

You touched upon one of the biggest downsides to the current ELO system 3C, and it's something we've discussed at length and tried to come up with a solution.  The best option was suggested originally but RootRanger and QuantumT where we would use a KRACH system where rankings are dynamic and change as your previous opponents scores changed.  So if you beat me when I was at 1500 and I later gain in the ranks, your score would rise as well to reflect that.  Basically a players current ranking would be the determination in how much they are worth, not their ranking at the time you played them.  The problem is the formulation for this is too intensive for Googledocs or any other currently known free online application to process making the only way it would be possible to use is if all rankings were done offline by a LO on an Excel spreadsheet.  That's a system that's not realistic.  Until online database technology catches up to the processing needs of a more complex system like KRACH we will stick with ELO.

Any link to the way such a system works? (wikipedia, for example?)

[Jen-i]

I cant right now jenky but tHe issue is that KRACH needs you to do interative calculations. not something googledocs handles very well at all.

[RootRanger]

This (http://www.collegehockeynews.com/info/?d=krach) explains things well.

KRACH is calculated as W:L ratio times a weighted average of your opponent's ratings. But whenever someone's rating changes, you need to recalculate the entire ratings, and then recalculate them again and again until the new calculations are about the same as the last ones.

Four main things in a rating system are important:
1.) Wins always improve the rating. Otherwise, players would always avoid battles that couldn't help them.
2.) Losses always hurt the rating. Without this, the leagues would mostly be about games played and by skill.
3.) Strength of schedule counts. If this didn't matter, people would avoid strong opponents and only want to play weaker players.
4.) Every game counts equally. This is probably the least important quality, but it is still improves a rating system because it is fair to every player.

KRACH accomplishes all four points while being able to predict odds for any game. But ELO has three of these qualities, and I guess that's good enough.

[bucky1andonly]

elements relies too much on luck for any sort of skill based ranking system, if i have to keep repeating this i will.  for luck based anything that is ranked, you use a simple rank system.  the amount of skill needed for elements is so little that it comes down to choosing what deck based on what cards you own.  drop this nonsense and go simple, put a max limit on number of games allowed to prevent spamming, possibly a weekly max too, and a minimum still with a weekly min.  give a set amount of points for a win, give 1 for a loss, add in a penalty for not meeting weekly min.  this league is about playing games, and not about skill level, because skill is relative to what cards you draw.  a bad deck can still beat a good deck, because of the luck factor.

[agentflare]

elements relies too much on luck for any sort of skill based ranking system, if i have to keep repeating this i will.  for luck based anything that is ranked, you use a simple rank system.

Counterpoint: TStar. His w/l is definitely not within normal distribution.

[TStar]

elements relies too much on luck for any sort of skill based ranking system, if i have to keep repeating this i will.  for luck based anything that is ranked, you use a simple rank system.  the amount of skill needed for elements is so little that it comes down to choosing what deck based on what cards you own.  drop this nonsense and go simple, put a max limit on number of games allowed to prevent spamming, possibly a weekly max too, and a minimum still with a weekly min.  give a set amount of points for a win, give 1 for a loss, add in a penalty for not meeting weekly min.  this league is about playing games, and not about skill level, because skill is relative to what cards you draw.  a bad deck can still beat a good deck, because of the luck factor.

If Elements relies mostly on luck, explain why some players seem to excel season after season in Leagues and others do poorly season after season.  Shouldn't the luck even out and have them do poorly?  It seems that maybe there is more of a skill element involved that you might realize and it might be possible you just haven't figured out that dynamic yet.  Also you are a CL-only player and from what others say CL is a big RPS type of meta so I can see how luck might be a larger factor in that situation.  The BL meta is nowhere close to an RPS one so perhaps there is an additional skill-based element in that meta that the CL one currently lacks.