Toribash
Original Post
Reinstate a Past Ranking System (An Argument for Change)
Please think clearly about what I'm going to say about the ranking system before you just say no.

This ranking system was a cool attempt at making the "better players" have a higher rank but it simply didn't work.

To quote [RAWR]Dargon:

The problem is ELO systems don't work in a game like toribash.

My main argument is that the current ELO system supports the absolute opposite of what Nabi Studios would want to have in an atmosphere to acquire new players. It also casts out enthused players who want to try lots of different things and learn from better more experienced people.


1) The current system rewards those who don't try a variety of mods.

Most high rank people (Read: 1-35) are there because they played the hell out of 1 mod and didn't venture off to try other mods. The point being that it's way easier to go 60 wins 2 losses in a game like jousting with one decent opener against new players and have a super high ELO (proof... well me) than it is to go 20-5 in a slower mod like Wushu.

Playing MORE mods should be what the game encourages; however, if you start playing aikido and only have an ELO of say 1610, when your jousting ELO is 1700, your rank plummets. Thus playing more mods currently is a bad thing.

2) The point system doesn't work and frustrates members.

A good example is what happened to me yesterday. I played brushu, and won a game against [fl0w]Deprav (a highly ranked brushu player), then a game against [Alpha]Logic (the #1 ranked brushu player in toribash), followed by a loss to the ranked 120th player or somewhere in that region. The net result? I actually lost 2 ELO points. I won 2 games, and lost 1, my 2 games were incredibly quality wins, and actually got a WORSE rank.

This just goes to show the math simply isn't working here. If you can beat the absolute best player, a fantastic player, and then lose to a great player, and be worse off when you started... well that's simply counter-intuitive isn't it?


3) The current system discourages playing people who aren't good (the Ambassador argument)

Better players of toribash are really the ambassadors for Nabi studios more than even GameMasters from time to time. They give new people a chance of what could come of them and the beauty of highly skill play. But the current system deterrs amazing people to play those who aren't good because one lost game can make the great player lose a LOAD of ranking. Getting a streak of 5, then losing once, only to have the same or worst ranking as you started is completely disheartening. You should want great players to want to play with not-so-great-ones to really show what your character can look like, what toris can do, and learn from them.

http://forum.toribash.com/tori_ranki...?username=Mojo This is the #1 Ranked player in Toribash. Check out the stats. Notice anything? Outside of JOUSTING and KATANA this player, the player who will represent all of the thousands of thousands of players who some started in 7 years ago as NUMBER ONE has a 1600 ranking in almost every skill mod out there. Has Wushu3, aikido, taekkyon, you name it. 1600 is the starting elo...

4) Our current top ranked leaders aren't role-models (a continuation of the Ambassador argument)

SEVEN of the top TEN players ranked globally in toribash haven't reached 3rd Dan Black Belt. Two of them haven't even reached black belt yet. These members aren't well-known good people in great clans that people aspire to. There are absolutely amazing players in clans like [RAWR] whos rank is near the 2000's meanwhile players you and I have never even heard of are sporting the highest of ranks. Is that really who you want to learn is the "best"? Who would you feel more comfortable being rank 5: TIcux, or "BigClitor" or "BigPrick" (I kid you not, those last 2 are both in the top 10)

5) Promotes rage quitting. It's simple: if you have a 1700 mmr in jousting and someone has a 1590 mmr rating, and they are about to lose to you, they'll lose a shit ton of ranking unless they ragequit. We should be encouraging people to stay the whole match.



TLR Summary

The current system makes unaccomplished players look like heroes to new players, tells people not to play a variety of mods, encourages the play of simple fast mods like judo and jousting over a variety of mods that might take more skill (aikido, aikidobigdojo, lenshu, wushu, taekkyon, judofrac) to name a few, mathematically seems to not work, and discourages the great players who should serve as ambassadors from playing with new players, and RAGE QUITTING.



The Solution? Bring back the QI system!!!

The highest rank people should be the people with the highest QI because that's what actually works in Toribash. Why?

1) Because rage quitting sucks and we should encourage people to stay through the whole match. Rank 15 whoever should want to accept defeat against rank 500,000 white belt honorably because he might be getting closer to Rank 14. Meanwhile, that white belt's DAY WAS MADE beating you.

2) It give quantifiable reason to PLAY TORIBASH. The more you play, the more QI you get, the higher your rank. The old people who have given years and years don't feel like idiots with thumbs up where the sun don't shine because they're rank 3000, and the newer players WANT black belt, 2nd dan, 5th dan, 10th dan, WHATEVER to improve their rank! Isn't a line-up like Tripstone, DJ, TIcux, Nuthug, Kamiko, Dargon a better top 6 than .... Mojo, Kindercat, zaeby, PREY3R, and pedart?

3) It doesn't discourage you from playing any mod you want. Sure, jousting is faster than aikido, and a person who only plays jousting will get QI faster and have a higher rank. But at least if you like jousting, and you decide to play some aikido, you won't completely f over your rank playing it.

4) It encourages coherency AND LEARNING throughout toribash. When you want to get good at a mod, you typically surround yourself with people much better than you (ergo, want to be an aikido player? Play team sambo members. Etc). But you're going to lose a lot doing that. I don't think you should hate yourself from trying to learn a mod because it dropped your rank from 200 to 1700.



And I ALREADY know the major criticism. THIS SYSTEM PROMOTES FARMING!!!

Yes. It does. Having QI determine your rank absolutely promotes farming. But what sounds better of a problem? Farmers farming, and more than likely getting caught because we have ways to catch farmers pretty easily (check out that community ban list!) or having hundreds of people rage quitting to save their rank? Losing rank by the hundreds from trying something new? I think the choice is obvious.


A last point

All of toribash has ever been is playing more toribash. Since you first joined, you probably wanted to wear a color that required you play more games, or get a belt you want, even a custom belt, etc. Make the system reflect what toribash is actually about-> stress-free fun, punishment free, playing. We already have our win-loss ratio publicly posted. If you want to prove your skilled try to increase that. There's no need for this system which 100% DOESN'T work.




To those of you who have read all of that, thank you. I appreciate it. To those of you who didn't, I recommend you do, because you're going to probably make posts that I've already addressed in this argument.
Last edited by Bodhisattva; May 23, 2013 at 03:21 AM.
Need help?
Creati0n says: still my favorite. <3
I sacrificed my firstborn for this great human being to join (M) ~R
Just Use Thunder!
Bump



So I've spent the last month or so trying to understand how exactly the ranking system works, proving that I know how to change my rank for the good or for the bad, and have compiled a few ideas to update our current ELO system since were content on not reverting back to QI.



1) MMR should be weighted.

That is to say that going from 1670 to 1680 MMR in a mod should be worth more towards your rank than going from 1600 to 1610. Logically, this makes since. It's very easy to go from noob to not completely terrible in the matter of 10 matches or so in a mod; however, to go from 1670 to 1680 in some mods is the difference between rank 12 and rank 7 in a publically hosted official mod like ninjutsu.

So what's the ramifications? Well basically, I could waste my time bringing my wushu3 rank from 1680 to 1690 and lose 3 mmr for every loss and gain .4 mmr for every win because my rank is high OOOOR I could play wushu2 wushu, wushu4, wushupro, and bring my 1600 ranks to 1630 in the same amount of alloted time since it'd be a lot easier to do so, and my rank would SKYROCKET a solid 4x more without the headache and without the chance of losing rank.


So as I said, the thing to do here is make your rank as you go up worth more. Going from 1690 to 1700 mmr should have a much more weighted effect in the calculatinos than 1600 to 1610.


Additionally, because of this error, the game favors you beating the living hell out of white belts over playing good players. Why would I play good players in jousting that know all the openers and I have at best a 3 win streak, then lose, and break even or WORSE at rank when I could do the same opener on 8 white belts and streak for 30 games and my rank goes off the charts high?

2) Polar opposites should not gain or lose MMR for playing each other.

This theory is adopting another happening in Starcraft 2's ladder system. If for god forsaken reason a Diamond League player is faced with a Bronze League Player, the game is worth absolutely no MMR for winning or losing to either player. The idea being it's simply unfair to put a good player in the position where if he loses he gets penalized severely and if he wins he gets essentially nothing.

So to adopt this theory into Toribash, maybe matches only count towards mod rank if you're within say 25-35 mmr of one another. (That is to say 1600 to 1635). This way the rank 1 wushu player doesnt have to streak on 7 people to make up for the accidental loss to a newer player (OR AN ALT, because alts run rampant in toribash).

This would also allow for newer players and older players to have incentive to play together, without the fear of losing large ranks over a misclick, a wrist break, or what have you.


3) Average out ranks amongst the most highly ranked mods played of that player(let's say 8 or so)

Half the reason some players have humongous global ranks (like 1750's) is because they only play ONE mod. Shouldn't the highest ranking player be more than just a jouster and katana player? I'm not saying we force everyone down the road of judofrac, wushu3, aikidobigdojo, yaddayaddayadda, all I'm saying is that the mmrs should be averaged amongst multiple mods of the individuals highest ranks.

This also would provide incentive for trying new mods. If you're not a 1670 mmr player at a mod, you won't plummet in ranking for trying something new and learning. I went from rank 17 to 85 in ONE DAY of winning 50% of my games because I was trying to learn aikidobigdojo from someone better than me. That sort of stuff wouldn't happen if we just were represented by an average of your best ranks.

I can't speak for everyone, but I'd rather have a rank 1 global player whos top 10 in wushu, top 25 in aikido, top 10 in judofrac, top 50 in jousting, top 50 in lolnade than just #1 ranked joustingfixed #1 katana.



By implementing these ideas we:

1) Alleviate some of the stress of playing games
2) Make a more accurate representation of who the better players are
3) Create incentive to a more equal opportunity toribash amongst pros and noobs
4) Get rid of all these no-name, cheating, mod rank farming alts who hold the top 15 or so positions.



Thoughts?

HB Moderated Message:
Your capacity to write is just amazing
Last edited by HolyButler; Jul 5, 2013 at 01:03 PM.
Need help?
Creati0n says: still my favorite. <3
I sacrificed my firstborn for this great human being to join (M) ~R
Just Use Thunder!
So, I'm not very good with logic and numbers, so I'm not gonna comment on that

Anyway, I saw some interesting ideas that could work well together, and I just felt like highlighting them to show you how all of this could work in harmony.
This is how we would handle it, if I were to decide;

All ranking is done in a few official servers. Or rather, doesn't have to be servers at all, just use the matchmake (which was actually removed, for some reason, I think nobody wanted to use it).

This time around, if you wanted to gain rank you would have to play in a matchmake server, it has to be that way. And it would work alongside rank. Playing outside of matchmake doesn't affect your rank at all, so you don't have to worr about losing it all the time.
The two players with the closest rank would play together. And I'm not saying that the winner would take the losers place or anything, the winner would get a few points and the loser would lose a few points, simple as that.
It has to work differently in different rank levels, though, because low-rank players would have to play forever to get to a decent rank, and high-rank players would lose too much rank for losing a single game.

The games would work according to something Snake originally wrote, I think.
There would be 4 games/round, once in Aikido, Tk, Wushu, and sw4 (or judo, though I'd rather have something with actual skill). Course, this or customisable, not for players, but for GM's and such. It could vary on a weekly basis.
Snake also mentioned a point system, question is if we should use that or just victory points.

Your mod rank is how much points you've gained (either by winning or by doing snake stuff) in each separate mod, and your global is how much points you've gotten in general. Not playing for, say 5 days (again, not set in stone) drops your rank from time to time.

Not sure how the ELO should work, because I'm stupid, but this is the basic idea. And there are problems with it, as you can probably see. For example;
"In high levels, what if no high ranks are online and I end up fighting a noob?"
Why then you win, and get just as much points as you would if you fought a high level player.
If the high levels aren't online and someone passes them, that's too bad for them.

Yeah I don't have more examples, but hey.
Last edited by Lazors; Sep 2, 2013 at 11:08 AM.
Brendan (he who passeth judgement on the frequent changing of signatures): I don't do hentai anymore
I have a different idea;

I used to play a game named "Gladiatus"(some of you may know it).That game's ranking system can actually work for Toribash.In Gladiatus your rank increases when you win against someone who has maximum 5 ranks better than you.(for example:If you are 123rd player,when you win against 118th player you become the 118th)
This system can be modified to Toribash and in my opinion it would really work.
For example you are #1500 when you win again #1000 you become number #1000 and he becomes #1001.
But if this system is applied to Toribash,we will need unactive players to be punished.For example if you are #1 and stop playing,system should throw you back to #2000 or if you are number #2000 and did not play against someone who has a challenger rank to your rank you get thrown back to #4000.
This would also provide a dynamic ranking system that is based on skills.
Hey DrHax I just want to say that you are making a really great point, I started playing a couple of days ago and have already achieved Blue Belt. I have made an effort to try every mod and where has that gotten me. I have an ELO rank that is decreasing even when I win. But you have proven to me that I should probably just ignore it
Sorry for the bump but I'd rather not make a new thread with the same topic, blablabla..

Anyways, ELO ranking is flawed, QI ranking is flawed; they promote farming and spamming random mods that are extremely easy to win against new players (lolnade, jousting, katana, etc).

Since no-one has given another form of rating that would be okay to use, why not just nerf the rating system altogether?

- Limit amount of ranked games you can play per day (eg. 10 ranked games per day)
Note: you still have 100 wins/games per day limit but only a maximum of 10 games will change your ranking
- Create ranked servers (public / black belt and above / ultimate), as in a server where players that want to increase the rank go to play (maximum 10 ranked games). The same ELO system would be there but this would slow down farmers by a huge amount.
Or alternatively players can just go into their private servers and type a /rankedmatch <on|off> command before a match starts.


This would nerf the ranking system because this stops farmers from having the opportunity of winning 100/100 games.
- If they are playing against extremely low ranks (white belts, etc) then 1 win would only be roughly +0.3 ELO; 10 ranked wins would only add up to a measly +3.0 ELO.
- If they win against high ranked players (around 1680+) then they would get around +1 ELO.
If that same guy gets 10 ranked wins on the same high ranked player (adding up to roughly +10 ELO) then good for him/her! At least then they won't be able to farm as quickly as they can with the current system that we have.
So here could be a way that could work. I know it's (perhaps) different to what most of you have been discussing (not really, but I'm not even sure because I stopped reading all the walls of text), but that's why I'm shotgunning it out here so perhaps it could spark some thought/ideas. Anyways, here's some of the points I considered for the system -

1) It is a ranking system that is similar to Qi, in that it only rewards "points" to winners and does not punish losers. This should alleviate the pressure of losing.

2) It is dynamic, in that it scales according to the relative ranks between the players, for example, were a lesser ranked player able to beat a higher ranked player, they would gain significantly more points than the other way around. This gives "newer" players an opportunity to climb the ladders. Much like the current ELO system.

3) The points you gain from defeating another player deteriorates over number of games played with that particular player (i.e. hyperbolic, becoming pretty much obsolete but not completely useless after x amounts of games), this will to some extent prevent farming against the same players and promotes playing with different people. This also prevents higher ranked players from "carrying" lower ranked due to the lack of punishment. If it demotivates gameplay with friends, then perhaps set a minimum, for example, assuming no other factors are influencing this, if this is the first time A encounters B, and A gets 1 point for winning B, A will get less and less points for continuing to win B until it reaches, say 0.1 points, and all each subsequent win after that gives 0.1 points. Point can reset daily/periodically.

4) Ranks reset after a period of time, much like seasons in many different games, promoting active play. Since everyone will be starting at the same point every reset, good skilled players will definitely have an advantage to take lead given they were active, as on average, they would be winning significantly more than your less skilled players (both gaining points at the same rate). These "seasons" should effectively differentiate the purpose behind ranks and Qi.

5) Perhaps categorizing people with different amount of points into "classes", or much like other games, "leagues", especially for newer players who would have little hope earning higher ranks as they start playing due to the steep learning curve, for example, progressing from rank 50k to rank 1 in the "newb" class seems much more significant than say from rank 250k to 200k overall. It gives newer players an incentive to keep playing and improving. Going up a league will also feel like an even larger achievement.

Anyways, think I've addressed a few of the problems brought up including the stress of playing for higher skilled players (no punishment), motivation and hope for newer players (incorporating leagues and seasons), promoting active play (seasons), decreasing farming (deteriorating points), the overall fun and usefulness factors (seeing change in rank/league more visibly and the fact that ranks reset), and overall a fair (as far as I see it) ish system for skilled players to take lead whilst allowing new people joining in a decent chance to climb ladders.
Last edited by OfUnknown; May 7, 2014 at 11:44 AM.
[Ele] Leader | [Bleu] Supervisor | Artists' Co-op
I love your 4th and 5th idea, OfUnknown. Mostly because it works so well in LoL and other games like that where it is very competitive.
If ranks were to be reset yearly, that would be great. It means that if you want to keep a high rank, you need to actually play the game instead of just attaining rank 1 and then quitting the game.
Perhaps if this was implemented there could be a large-scale event each year where players from different leagues/tiers compete against each other for a prize of some sort (eg. tier 1 players vs tier 1 players: big prize because it has bigger competition / is hard to win. tier 5 v tier 5: small prize because not as competitive /easier to win than tier 1 comp).

Some further ideas for this:
1) To rank up between each tier you need to do a series of ranked matches and if you fail to win then you have to wait a certain period before having the chance of ranking up again.

2) If you're inactive after a while and fail to complete ranked matches within a certain amount of time then you are demoted to a lower tier / if your rank drops too low you get demoted.

I feel this could work if ranked servers were put in place with a limit on how many ranked matches you can do per day.
I feel this could work if ranked servers were put in place with a limit on how many ranked matches you can do per day.

1) To rank up between each tier you need to do a series of ranked matches and if you fail to win then you have to wait a certain period before having the chance of ranking up again.

The problem with "official rooms" is that it would be demotivating waiting, as well as makes the game more restrictive, counteracting the openness of Toribash. Perhaps, allowing players to create official ranked rooms will solve this problem (i.e. it could simply just like current rooms, however server settings are set by the game and no OP is designated). Also limiting the ranked matches per day may reduce activity due to there being a discrete point where playing further just doesn't matter were the ranks to be taken seriously. The hyperbolic decrease in points however achieves a similar purpose while at the same time, make repeated gameplay not pointless yet obsolete enough people don't abuse the system, and additionally, it also promotes diverse gameplay (matching up with other players).

2) If you're inactive after a while and fail to complete ranked matches within a certain amount of time then you are demoted to a lower tier / if your rank drops too low you get demoted.

This may seem either demanding (i.e. game requiring you to play) or demotivating (knowing your rank is destined to become useless). I think this would also be unnecessary in the rank system I described, the points in the game will be ever inflating so inactivity would automatically be regulated if players do not keep up with the others (especially if the goal is to be at the top of the ladder).
Last edited by OfUnknown; May 7, 2014 at 11:42 AM.
[Ele] Leader | [Bleu] Supervisor | Artists' Co-op
I didn't mean just wait a few matches, I meant like couple of weeks. You can still do ranked matches and raise your ELO, but won't be able to get into the next tier until you complete the series of matches.

Limiting the number of ranked matches won't stop players from playing Toribash just because the next matches won't count towards your ELO. They'll continue to play for TC / fun. Majority of players do not even care about their ranks, this is just something for the minority of players that actually care about a number and the limitation is just there so they don't go overboard and farm. Surely waiting a day to do another 10 ranked matches won't hurt so much would it?

The resetting of ranks won't be useless, because as I said there could be an annual event put in place for this where players compete against other players in the same tiers for a prize of some sort.
Small prize for the lower tiers (not small as in low-priced items/small amount of TC but small as in a medium-priced item/s) and larger prize for the higher tiers. Since it would be a yearly thing the large prize could actually be a large prize like 1mil TC or something, just an idea.
The demotion thing wouldn't be demanding so that you have to play, but if you want to be rank 1 then you should at least play occasionally.. (note: you don't have to play if you don't want to)
Last edited by Holotor; May 7, 2014 at 12:02 PM.