forum

Osu! Elo Project (CTB)

posted
Total Posts
24
Topic Starter
He Ang CTB
Introduction
Inspiration:
Good day everyone o/ Recently I've been watching the Catch World Cup, and I have observed that Solo ranking doesn't translate very well to Competitive performance. So I thought we can experiment with an age-old ranking system that has been mentioned/suggested but has yet to be implemented in Osu! competitively, namely the Elo Rating system.

Motivation:
The motivation of this project is to experiment with the viability of the Elo Rating system, as a credible and reliable system of ranking that is able to distinguish players through competitive results, regardless of Solo rankings/Performance Points.

The Idea: (long and may not need to know)
Elo Rating is rather mathy/statistical, it is complicated to understand and even more complicated to explain, so here's the general summary of its key features:
1. All new players start with a fixing rating (e.g. 2000 ERP, ERP stands for Elo Rating Point).
2. For every match there will be either Win, Lose, or Draw.
3. The amount of points gained/lost can be zero-sum (winner's gain = loser's loss), or non-zero-sum (winner's gain =/= loser's loss), depending on the K-Factor.
4. K-Factor works like a "rating gain/loss multiplier". If KF is a constant number for all players (e.g. KF = 30), then winner's ERP gain = loser's ERP loss.
5. If KF is affected by number of rounds a player has played, players can have different KFs. The winner's KF may be different from the loser's KF, causing the amount of ERP won be different than the amount of ERP lost, making the Elo System non-zero-sum (Elo inflation/deflation).
6. The sequence of match-ups affect the calculation of ERP. It is a path-function, not a state-function (depends on the chronological order of the results). E.g. you have played against the same player when he's at 2000 and when he's at 3000, you could have won 5-1 on both occasion, but clearly you would gain more ERP on the 2nd match-up, even though both results were 5-1 and both were against the same player.
7. You will lose ERP for a Draw against a player with lower ERP.
8. The more matches a player has played, the less volatile the ERP is for that player (gain less ERP and lose less ERP for the same results over time), due to KF getting smaller as the number of matches played gets larger.
9. Alternatively, the volatility could be null if KF is set to a constant number, rather than a function of accumulated matches.
10. All in all, you can expect a relatively fair system when it comes to raking players for their competitive results.
11. For more detailed mathematical explanations, you can refer to: https://en.wikipedia.org/wiki/Elo_rating_system

Comparisons:
1. A team of 4 non-players called "Panel of Judges (POFs), will be chosen from interested volunteers.
2. The job of the POJs is to give their personal opinions of the winners/losers of the upcoming matches of the week.
3. These opinions will be compared to the probability calculations derived from the players' ERPs.
4. These results may be posted in the forum, where the public can view the comparison between human predictions & statistical predictions.

Key Information
General info:
1. This project requires matches to be played.
2. The matches will be played in Catch The Beat (Osu!Catch) mode.
3. There will be exactly at least 8 players for this project.
4. There will be exactly at least 4 weeks of matches.
5. The matches will be played on Weekends (Friday evenings, Saturdays, Sundays).
6. There will be 1 new map pool for every week (4 map pools in total).
7. The match scores will be calculated in Score_V2.
8. The prize is 1-month Supporter for 3rd & 4th place, 2-month Supporter for 2nd place, and 4-month Supporter for 1st place. Read Updates.
9. No profile badges are to be expected.
10. No streams are to be expected.
11. The final eligibility of players to participate will be decided by the Project Managers.
12. The registration starts from now till 3rd August (Friday 4 weeks from now).
13. You can post your registration as a reply to this forum thread, or in any other ways that you can contact the Project Managers.
14. The matches will start as soon as the next weekend after the confirmation of 8 players.

Scheduling info:
1. Every player will get to player other players exactly twice each, throughout the entire project.
2. The 2 occasions will not be consecutive, as much as possible.
3. For 8 players, there will be a total of 56 matches. (7+6+5+4+3+2+1) x (2) = 56.
4. There will be 2 matches on each Friday evening, 6 matches on each of Saturdays & Sundays. (2+6+6) x (4) = 56.
5. Each week's schedule will be ideally posted on Sunday evenings.
6. Any request to reschedule the matches, please contact the Project Managers at least 2 days in advance.

Scoring info:
1. A K-Factor of 400/(total rounds a player has ever played in this project) 32, will be used in the calculation of ERPs.
2. As a repeat, all map scores are calculated in Score_V2.
3. All matches will be 5TW (5 To Win/Best of 9/BO9).
4. Draws count as 1-1, not 0-0.
5. You can Draw on TieBreaker, making the match 5-5.
6. If you fail at the end of the map, it is considered a loss.
7. If both players failed at the end, the player will high score will win.

Match info:
1. All rolls are determined on a Live Rolling (live-stream), for all matches of that week.
2. Each player will be given 2 keys.
3. A player can use 1 key to lock/ban 1 map.
4. A player can use 1 key to unlock/unban 1 map, previously locked/banned by the other player.
5. You can also unlock a map you yourself have locked, AKA wasting keys.
6. The winner of the roll will use his 1st key, the loser will then use 2 keys consecutively, and the winner will lastly expend his 2nd key.
7. The winner of the roll also get to pick the first map.
8. The usage of keys & picking of the 1st map will be done on the match itself.
9. All keys must be expended before the 1st map can be played.
10. Each player is given 1 warm-up pick of no longer than 5 minutes in net length.
11. The sequence of selecting warm-up maps does not matter.
13. The referee can deny warm-up maps if the match is running behind schedule for any reason.
14. Each map in the map pool can only be chosen once.
15. The player will be given 2 minutes to choose the map, the referee will choose for the player if it exceeds 2 minutes.
16. Once the map has been chosen, it cannot be changed.
17. The referee can invite the players 15 minutes before the match begins.
18. Refusing to acknowledge or comply with the referee's invite or any other instructions, will be considered as a misdemeanor.
19. If a player is late by 15 minutes, or any time later than 15 minutes that the on-site referee can tolerate, the player will Lose By Default (LOB) with a score of 0-5.
20. Similarly if the player misbehaves despite multiple warnings, for whatever reasonably deemed as disorderly conduct by the on-site referee or the Project Managers, the player can be given an LOB.

Map pool info:
1. As a repeat, there will be 4 map pools for each of the 4 weeks.
2. Each Map Pool contains:
- 3 NoMod (NM)
- 3 Hidden (HD)
- 3 HardRock (HR)
- 3 DoubleTime (DT)
- 3 CasualMod (CM)
- 3 FreeMod (FM)
- 1 Easy (EZ)
- 1 Flashlight (FL)
- 1 HalfTime (HT)
- 1 TieBreaker (TB)
- 22 playable maps in total.
3. NM & HD are to be played in their respective mods only.
4. You can equip HD for all other mod categories.
5. You are not allowed to equip any mod other than HD for all mod categories, except for FreeMod.
6. You can equip any mod combination you desire on FM, including DT, HT, FL, EZ, SuddenDeath (SD), NoFail (NF), SpinOut (SO).
7. In the case of DT and/or HT on a FM map, the non-DT/HT player will play first, followed by the DT/HT player.
8. CasualMods are by default NoMod maps, unless you want to equip HD.
9. Map pools can contain both Catch Specific maps and Standard Convert maps.
10. The difficulty of each of 4 Map Pools will be about the same, difficulties ranging from RO16(-) to Semi-Finals(+) levels within each Map Pool.
11. In case of mid-week map updates, the original versions of the maps will be played.
12. Maps will be found in the Map Pack, and only the exact maps in the Map Pack of that week can be played.
13. Map pools will ideally be announced on Sunday evenings.
14. Once the Map Pool is posted, it is considered finalised.


Results
ERP rankings after Week 1:
1.
2.
3.
4.
5.
6.
7.
8.
ERP rankings after Week 2:
1.
2.
3.
4.
5.
6.
7.
8.
ERP rankings after Week 3:
1.
2.
3.
4.
5.
6.
7.
8.
Final ERP rankings (after Week 4):
1.
2.
3.
4.
5.
6.
7.
8.
Spread Sheet link:

Updates
Map Pool link:

Player pool expansion:
Due to there being an overwhelming amount of interest, the player pool will be expanded beyond 8 players 40 players!

Prizes update:
(a) For 8 players or more:
- 2nd & 3rd:1-month supporter
- 1st: 4-month supporter.
(b) For 16 players or more:
- 3rd & 4th: 1-month supporter
- 2nd: 2-month supporter
- 1st: 4-month supporter
(c) For 32 players or more:
- 4th: 1-month supporter
- 2nd & 3rd: 2-month supporter
- 1st: 6-month supporter

(d) For 40 players:
- 4th & 5th: 1-month supporter
- 2nd & 3rd: 2-month supporter
- 1st: 6-month supporter

Calculations update:
- K-Factor has been changed from 400/(Total maps played by player), to KF = 32.

Advancement Format
The registration is officially closed! I'll invite the players to our Discord channel through mass PMs (don't want no weirdos joining in from nowhere o3o).
Since we have yet again expanded beyond our initial limits, we'll need to host pre-Elo rounds to shortlist the best 20 players!


For better viewing, download the PDF here (auto-download): https://puu.sh/B4w5x.pdf

"Why not 8 Groups of 5?"
- That'll be 8 x (4+3+2+1) = 80 matches, as compared to 10 x (3+2+1) = 60 matches in total. 20 more matches on top of the daunting 60, the referees are humans too :O
- Each player would have to play 4 opponents instead of 3, players have real life stuff too! :O

"Why not just Top 2 per Group advance?"
- It extends the players' possible play-time during this tournament.
- In the typical Round Robin format, the 3rd player from Group A may be much better than the 2nd player from Group B, but because of Group A being a much harder Group overall, the better player with a 3rd placing in his hell group will be disqualified, while the 2nd player in his trash group will advance. The new advancement format double-checks the fairness of Group Stage with a cross-group Knock Out Stage of the 2nd & 3rd place players.
- This is also a major incentive for the Top player of the Group to be rewarded with a straight entry to "Quarter-Finals" (in this case straight into Elo Phase).
- Conversely, it is a disincentive for players to mess around in a 1st Place VS 2nd Place match, as it will cost them their fast-track opportunity.
- On the flip side, the 3rd Place VS 4th Place match will be taken seriously as well, since it decides whether or not they can even advance.
- The 2nd Place VS 3rd Place match will be no less important, since it decides whether you get to play a better placing/worse placing opponent on the following week.

TL;DR benefits of this advancement structure:
- All Group Stage matches will matter significantly,
- More fair to Bad-Luck-Brians,
- More people can continue playing after the Group Stage.

Hope you guys have gotten a good understanding of this after a long read. If you have any questions or suggestions, you can post it down below or ask me personally. Any changes to the details of this project will also be updated below!
Yoomara
as much as i'd love to test this, my motivation to play is at an all time bottom. sorry!!
Topic Starter
He Ang CTB

Yoomara wrote:

as much as i'd love to test this, my motivation to play is at an all time bottom. sorry!!
It's kay :O Maybe next time when you're all fired up again I may have another tournament under way :D
LaviSorrow
Good luck :)
BPDelirious
I'd be happy to join either as a player or to help the staff but I won't be home for a while in the next 2 weeks :(
Topic Starter
He Ang CTB
Hello everyone! Do note that we have changed the 8-players limit to 32 players maximum. We have also turned up the top prize to 6-month supporter instead! Do check out the "UPDATES" section in the forum post to know more!

If the player count is 16, we can still proceed directly into the Elo phase of the tournament. That would still mean each player will play against 7 other players, just that now there are more players in the pool.

If the player count is 32, there will be a Knock-Out stage before the Elo phase of the tournament. After the Knock-Out stage, there will be 16 players remaining. So the tournament will be 5 weeks instead of 4 weeks (still starting in 1st week of August but end 1 week later).

Any other player count, the Panel of Judges will determine who can be part of the tournament.

I'll invite the finalised players into a Discord server, but that will be about a week from now.
Chatie
Very interested in playing, but not sure yet since I also signed up for the solo carry tournament.
Sounny
Yeah me too but i can play this tournament if i had the time :roll: :P ;)
Topic Starter
He Ang CTB

Sounny wrote:

ChatNYAN xD OwO wrote:

Very interested in playing, but not sure yet since I also signed up for the solo carry tournament.

ChatNYAN xD OwO wrote:

Very interested in playing, but not sure yet since I also signed up for the solo carry tournament.
You guys can still sign up when it's closer to the date :O The tournament probably won't start until 1st week of August!
Secre
i'll play if theres still space depending on when it starts
Topic Starter
He Ang CTB

chickenbible wrote:

i'll play if theres still space depending on when it starts
It'll probably start on the first week of August :O
Topic Starter
He Ang CTB
It seems like we already have 32 players, but 85 players still have not read my PM! >:O I guess I'll wait out for maybe 3 more days, then I'll start shortlisting the players for the actual event!
Rubytic
Ok seems I have a time to join the project. Can I join in as 'players' group?
Lay
I am interested in signing up as either staff or players, just let me know~
A Moon Gust
I'll participate in this.
Topic Starter
He Ang CTB

Rubytic wrote:

Ok seems I have a time to join the project. Can I join in as 'players' group?

A Moon Gust wrote:

I'll participate in this.
Okay added you two in the players group :O

Lay wrote:

I am interested in signing up as either staff or players, just let me know~
Aaaaahhh the players group is full D: I would love to have you as a Referee instead!
Lay

Lay wrote:

I am interested in signing up as either staff or players, just let me know~
Aaaaahhh the players group is full D: I would love to have you as a Referee instead![/quote]

That is perfect~
Any chance to help and work alongside so many familiar faces is a welcome one. Can't wait!
Topic Starter
He Ang CTB

Lay wrote:

Lay wrote:

I am interested in signing up as either staff or players, just let me know~
Aaaaahhh the players group is full D: I would love to have you as a Referee instead!
That is perfect~
Any chance to help and work alongside so many familiar faces is a welcome one. Can't wait!
[/quote]

Alrighty you're in as a Referee :D
MicroRocket
Open more places ! :v and gl
Topic Starter
He Ang CTB

MicroRocket wrote:

Open more places ! :v and gl
No more places for players sorry ;n; But I would love to have you as a Referee! :O
show more
Please sign in to reply.

New reply