Ranked Elo Overhaul

Leave your suggestions about the game here!

Ranked Elo Overhaul

Postby Flake » Tue Jan 09, 2018 11:59 pm

──────────────────────────────────────────────────────────────────────────────────────────────────────────────────
Last edited on November 9th, 2018
──────────────────────────────────────────────────────────────────────────────────────────────────────────────────
This is being updated over time. Some information may be added or changed at a later date, but for the most part this will be the finished product.

Please note that this Elo Rework is not official and is therefore not guaranteed to be implemented into the game.

It should also be noted that the entire rework does not have to be used - this is mainly to highlight the main issues surrounding the current Elo system, and appropriate methods to fix some of these main issues.
──────────────────────────────────────────────────────────────────────────────────────────────────────────────────


Section 1: SUMMARY - WHY AN ELO REWORK IS NECESSARY & GENERAL IDEA


──────────────────────────────────────────────────────────────────────────────────────────────────────────────────
The current Elo system, put bluntly, isn't the greatest it could be. The main issue is that it mainly rewards grinding games heavily as opposed to a high win percentage, which is more indicative of skill. So, I believe that the player win percentage should have a larger impact than it currently does. The current Elo system also does not give bigger Elo differences for a lower amount of games played outside of Placement games, and the Elo change difference between Placement games and other games at current is arguably too large. The K-Factor for those in Diamond and above tiers is also too small when compared to K-Factors in other tiers.

As such, I propose a new Elo system fixing "the grind". This new system would use win percentage as it's primary variable, and uses a "K-factor". This K-factor gives slightly higher Elo changes with a lower amount of games played, and lower Elo changes with more games played. The K-factor also gives less prominent Elo changes if the player has a very high Elo (namely if the player is in Expert Tier, see Section 5 for more information).

The starting Elo will be set at 1200, and you will gain or lose an amount of Elo from 1200 based on how well you perform as compared to the rest of the population. In general, if you have an Elo below 1200 you are performing worse than average, and if you have an Elo above 1200 you are performing better than average.

You will also be rewarded for having more games with factions of a lower mean population win rate (for example, Neutral Killing) and punished for having more games with factions of a higher mean population win rate (for example, Town).

The difference between Team Elo (minus yourself) and Opposing Elo will also be considered if you are Town or Mafia, but not Neutral Evil or Neutral Killing (refer to Section 6, note 3. to see why this is the case).

Ranked Placement will require 20 games played rather than the current 10 to be given an Elo score, because it is a larger sample giving a more accurate Elo score than if the player were to only play 10 games.
──────────────────────────────────────────────────────────────────────────────────────────────────────────────────


Section 2: THE ELO GAIN/LOSS FORMULA


──────────────────────────────────────────────────────────────────────────────────────────────────────────────────
If Player Wins: Gain = K * ((1-A%) + (OE - TE)/(2500) + (A% - F%)/(3A%))

If Player Loses: Loss = K * ((-A%) + (OE - TE)/(2500) + (A% - F%)/(3A%))

NOTE: Elo gain/loss is capped at ±2, meaning it is IMPOSSIBLE to lose Elo if you win a game, and it is IMPOSSIBLE to gain Elo if you lose a game. See the first two points below for more information.
──────────────────────────────────────────────────────────────────────────────────────────────────────────────────
WHERE:
──────────────────────────────────────────────────────────────────────────────────────────────────────────────────
• If you win, Elo gain is restricted such that Gain ≥ +2. If a win were to result in an Elo change below +2 (e.g +1), the Elo change would be automatically changed to +2.

• If you lose, Elo loss is restricted such that Loss ≤ -2. If a loss were to result in an Elo change above -2 (e.g -1), the Elo change would be automatically changed to -2.

For all points below using "x", let "x" be the lifetime amount of games the player has played on Ranked from the last Hard Reset:

• "K" is the K-factor, a constant which changes depending on which of the four following circumstances is true (NOTE that 2400 Elo & Significance Level (7) is Expert Tier; see Section 5 for more information):

- If your Elo is 2400 & Significance Level (7) or higher, K = 18.

- If x > 200 and your Elo is less than 2400 and/or less than Significance Level (7), K = 25.

- If 50 ≥ x ≥ 200 and your Elo is less than 2400 and/or less than Significance Level (7), K = 30.

- If x < 50 and your Elo is less than 2400 and/or less than Significance Level (7), K = 40.

• "A%" (Average Tier Win Percentage) refers to the average win rate of players in your current Tier, give or take one tier, for all games they have played in those Tiers. For example, if a player is in Platinum and the average win rate of all players in Gold, Platinum and Diamond Tiers for all games all players have been in Gold, Platinum and Diamond for is 60%, A% = 0.6.

• "F%" (Average Faction Win Percentage) refers to the average win rate, of any given faction, of players in your current Tier, give or take one tier, for all games they have played in those Tiers. For example, if the player is Bronze and the average Mafia win rate of all players in Bronze and Silver Tiers for all games all players have been in Bronze and Silver for is 40%, F% = 0.4.

• "TE" (for Team) refers to the average Elo score of your faction, excluding you (refer to Section 6, point 3. to see why you are excluded).

• "OE" (for Opposition) refers to the average Elo score of the opposing factions.

- If you are Town, opposing factions include: Mafia, Neutral Killing, Neutral Evil.

- If you are Mafia, opposing factions include: Town, Neutral Killing.

• If you were Neutral Evil or Neutral Killing in the game, OE - TE = 0. If you were Town or Mafia, OE - TE is unchanged. Refer to Section 6, point 3. to see why this is the case.
──────────────────────────────────────────────────────────────────────────────────────────────────────────────────


Section 3: EXAMPLES OF ELO GAINS/LOSSES


──────────────────────────────────────────────────────────────────────────────────────────────────────────────────
Example 1 (Town Win)
──────────────────────────────────────────────────────────────────────────────────────────────────────────────────
A player is Silver, has 100 games played (K = 30) and wins a game as Town.

The Opposing Factions have an Elo of 1350, and Town (excluding you) has an Elo of 1200. In other words, OE = 1350 and TE = 1200.

Suppose A% = 0.5 and F% = 0.6 (See Section 2 for an explanation of A% and F%).

Using the Elo gain formula, K * ((1-A%) + (OE - TE)/(2500) + (A% - F%)/(3A%)), we have:

Gain = 30 * ((1-0.5) + (1350-1200)/(2500) + (0.5 - 0.6)/(3*0.5)) = +14.8 Elo change.
──────────────────────────────────────────────────────────────────────────────────────────────────────────────────
Example 2 (Town Loss)
──────────────────────────────────────────────────────────────────────────────────────────────────────────────────
A player is Platinum, has 250 games played (K = 25) and loses a game as Town.

The Opposing Factions have an Elo of 1800, and Town (excluding you) has an Elo of 2100. In other words, OE = 1800 and TE = 2100.

Suppose A% = 0.55 and F% = 0.65 (See Section 2 for an explanation of A% and F%).

Using the Elo loss formula, K * ((-A%) + (OE - TE)/(2500) + (A% - F%)/(3A%)), we have:

Loss = 25 * ((-0.55) + (1800-2100)/(2500) + (0.55 - 0.65)/(3*0.55)) = -18.3 Elo change.
──────────────────────────────────────────────────────────────────────────────────────────────────────────────────
Example 3 (Mafia Win)
──────────────────────────────────────────────────────────────────────────────────────────────────────────────────
A player is Expert, has 300 games played (K = 18) and wins a game as Mafia

The Opposing Factions have an Elo of 2400, and Mafia (excluding you) has an Elo of 2600. In other words, OE = 2400 and TE = 2600.

Suppose A% = 0.6 and F% = 0.3 (See Section 2 for an explanation of A% and F%).

Using the Elo gain formula, K * ((1-A%) + (OE - TE)/(2500) + (A% - F%)/(3A%)), we have:

Gain = 18 * ((1-0.6) + (2400 - 2600)/(2500) + (0.6 - 0.3)/(3*0.6)) = +8.8 Elo change.
──────────────────────────────────────────────────────────────────────────────────────────────────────────────────
Example 4 (Mafia Loss)
──────────────────────────────────────────────────────────────────────────────────────────────────────────────────
A player is currently Silver in their placement games, has 15 games played (K = 40) and loses a game as Mafia.

The Opposing Factions have an Elo of 1300, and Mafia (excluding you) has an Elo of 1250. In other words, OE = 1300 and TE = 1250.

Suppose A% = 0.5 and F% = 0.4 (See Section 2 for an explanation of A% and F%).

Using the Elo loss formula, K * ((-A%) + (OE - TE)/(2500) + (A% - F%)/(3A%)), we have:

Loss = 40 * ((-0.5) + (1300 - 1250)/(2500) + (0.5 - 0.4)/(3*0.5)) = -16.5 Elo change.
──────────────────────────────────────────────────────────────────────────────────────────────────────────────────
Example 5 (Neutral Killing loss)
──────────────────────────────────────────────────────────────────────────────────────────────────────────────────
A player is Silver, has 20 games played (K = 40) and loses a game as Neutral Killing

OE - TE = 0 since the player is NK (See point 3. in Section 6 for why this is the case)

Suppose A% = 0.5 and F% = 0.1 (See Section 2 for an explanation of A% and F%).

Using the Elo loss formula, K * ((-A%) + (OE - TE)/(2500) + (A% - F%)/(3A%)), we have:

Loss = 40 * ((-0.5) + (0)/(2500) + (0.5 - 0.1)/(3*0.5)) = -9.3 Elo change.
──────────────────────────────────────────────────────────────────────────────────────────────────────────────────
Example 6 (Neutral Evil win)
──────────────────────────────────────────────────────────────────────────────────────────────────────────────────
A player is Gold, has 40 games played (K = 40) and wins a game as Neutral Evil

OE - TE = 0 since the player is NE (See point 3. in Section 6 for why this is the case)

Suppose A% = 0.5 and F% = 0.3 (See Section 2 for an explanation of A% and F%).

Using the Elo gain formula, K * ((1-A%) + (OE - TE)/(2500) + (A% - F%)/(3A%)), we have:

Gain = 40 * ((1-0.5) + (0)/(2500) + (0.5 - 0.3)/(3*0.5)) = +25.3 Elo change.
──────────────────────────────────────────────────────────────────────────────────────────────────────────────────


Section 4: SIGNIFICANCE LEVELS


──────────────────────────────────────────────────────────────────────────────────────────────────────────────────
One of the main issues with this Overhaul is that a player can simply gain a good win rate (and therefore a good Elo) in a (relatively) small amount of games by chance, which may discourage them from playing any further games so that they can maintain this good Elo. The introduction of Significance Levels entirely negates this weakness, as seen below.

Significance Levels are given based on the amount of games played - in other words, it is a measure of experience. Significance Levels are grouped as such:

Significance Level       Games Required      Amount of Significance

Level (1)                      19 or less                  Negligible
Level (2)                      20 - 49                     Very weak
Level (3)                      50 - 99                     Weak
Level (4)                      100 - 149                 Moderate
Level (5)                      150 - 199                 Good
Level (6)                      200 - 249                 Good
Level (7)                      250 - 349                 Good
Level (8)                      350 - 449                 Strong
Level (9)                      450 - 599                 Strong
Level (10)                    600 - 799                 Very strong
Level (11)                    800 - 999                 Very strong
Level (12)                    1000 or more            Extremely strong

Elo will now take the format of: Elo (Signifiance Level).

For example, a player with an Elo of 2600 and 500 games played will have an Elo of 2600 (9), since 500 games played correlates to a Significance Level of (9).

It should be noted that Significance Levels would be RESET (back to Level 1) with a Hard Elo Reset (all player Elo back to starting Elo). However, with a Soft Elo Reset (back to minimum of player's tier) or a Season change with no reset whatsoever, the Significance Level is unchanged.

It should also be noted that the amount of games is measured by the amount of Ranked games played FROM the last Hard Elo Reset.

With this change, you can easily distinguish between things affecting Elo (mainly win rate) and Experience, and at the same time you can easily compare things affecting Elo (mainly win rate) and Experience to see if the player's apparent "skill level (Elo)" is actually significant or not (based on the Significance Level). For example, a player with Elo 2700 (5) may not be as skilled as a player that is 2500 (11), since (11) is a much higher significance level and so the latter player has proven they can keep a consistently high Elo score. However, the former player's significance level, while relatively good, is still telling us that their very high Elo could have occurred by chance and so they may fall in Elo (below 2500) once they reach higher levels of significance.

With this system, removing grind now has next to no weaknesses. However, this system does have a weakness that comes with it. It could be argued that, since the Significance Level seemingly has no impact on your Elo Tier, the Significance Level doesn't really matter. Well, the fix to that is simple - it now DOES have an impact on your Elo Tier (see Section 5 below for more information on this).
──────────────────────────────────────────────────────────────────────────────────────────────────────────────────


Section 5: ELO TIERS


──────────────────────────────────────────────────────────────────────────────────────────────────────────────────
NOTE: See Section 4 (above) for information on Significance Levels

Bronze: Less than 900 Elo & Significance Level (2+)

Silver: 900 Elo to 1499 Elo & Significance Level (2+)

Gold: 1500 Elo to 1799 Elo & Significance Level (2+)

Platinum: 1800 Elo to 2099 Elo & Significance Level (5+)

Diamond: 2100 Elo to 2399 Elo & Significance Level (6+)

Expert: 2400 Elo or more & Significance Level (7+)

Master: Top 20% of Expert Players (Updates Weekly)
──────────────────────────────────────────────────────────────────────────────────────────────────────────────────


Section 6: IMPORTANT NOTES


──────────────────────────────────────────────────────────────────────────────────────────────────────────────────
1. The average Elo score of any given player will be approximately equal to the starting Elo, which is 1200 Elo.

2. The player is required to have played at least 20 Ranked games (placement games) to be given an Elo.

3. Differences in Opposing Team's Elo and Your Team's Elo (for Town/Mafia) does not take into account your own Elo. This is because you are excluded from your own Team's average Elo score, as if you are included and you have a high Elo, after many games you will be punished heavily in terms of Elo, and if you are included and have a low Elo, after many games you will be rewarded heavily in terms of Elo. As such, you must be excluded from the average of your own team's Elo score. This is why differences between opposing team's Elo and your team's Elo cannot be considered if you are Neutral Evil or Neutral Killing, because you must be excluded, but you are the ONLY member of that faction.

4. Things such as A% (Average Tier win %) and F% (Average Tier Faction win %) would probably be most accurately gathered over time (so that the percentages are constantly changing depending on various factors), but the data could also be gathered in an Off Season.

5. The Elo Tier you are in will automatically place you into games with those of similar ability. For any given Tier, in the general case, you will be able to play anyone in your Tier, a Tier above your Tier or a Tier below your Tier.

The only exception is the Master Tier, where you will be able to play against Diamond Tier. This is to increase the diversity of players those in the Master Tier have, and also to prevent those in Expert from easily overtaking those currently in Master by being in the Top 20% of Expert players.

As such, the following hold:

• If you are Bronze: You can play with Bronze and Silver players

• If you are Silver: You can play with Bronze, Silver and Gold players

• If you are Gold: You can play with Silver, Gold and Platinum players

• If you are Platinum: You can play with Gold, Platinum and Diamond players

• If you are Diamond: You can play with Platinum, Diamond and Expert players

• If you are Expert: You can play with Diamond, Expert and Master players

• If you are Master: You can play with Diamond, Expert and Master players

6. A person will be demoted from their Tier if they are below the Elo requirement for their Tier for 5 consecutive games, with the exception of demotion from Master Tier to Expert Tier (see point 8. below).

7. As stated in Section 5, the Top 20% of players within the Expert Tier will be placed in the Master Tier, and this Top 20% will update every week. This means that if someone is in the Top 20% of Expert Tier (Master Tier) one week but is not in the next, they will then be demoted to whatever Elo Tier they are now in immediately.
──────────────────────────────────────────────────────────────────────────────────────────────────────────────────


Section 7: THE PROS & CONS


──────────────────────────────────────────────────────────────────────────────────────────────────────────────────
PROS

• Less based on grinding Elo. The main changing factor is the win rate of the player. As such, in this Elo rework those with a high amount of games played and a mediocre win rate will not be rewarded as much.

• 20 Placement Games means that there is a larger sample of games than the current 10 Placement Games, meaning that when the player gets their Elo score it is more likely to be indicative of their true skill.

• Allows for a ranking system where those of similar ability (plus or minus one tier, refer to Section 6, point 5. for more info) are placed in games together.

• Considers almost all factors, including things such as Your Faction's and Opposing Faction's Elo when Town/Mafia (refer to Section 6, point 3. for more info on why this is only Town/Mafia), increases in Elo due to large amount of NK/NE/Mafia games (due to low faction win %), loss in Elo due to larger amounts of Town games (due to high faction win %) and the K-factor which considers the amount of overall games played.

• Self-sufficient, since it is based on win percentages. This formula would not have to be updated if balance changes were to be made affecting the faction win percentages. However, a reset would be necessary if balance changes were to occur so that the previous inaccurate data is not taken into account. This would likely be done in the form of a new Season.

• Elo gains/losses from game to game are generally higher and therefore more exciting than the current Elo gains/losses. This may be a deterrent for trolls/gamethrowers/leavers, since losing a game means a heavier punishment than in the current system.

• Higher amount of complexity allows for a more accurate Elo rating and overall skill indication.

CONS

• Relatively complex. Perhaps quite time consuming and/or difficult to implement into the game as a result.

• The amount of Elo increase/decrease due to your faction and Elo increase/decrease due to higher/lower Elo opponents may be weighted too heavily or not heavily enough, since it is impossible to know the optimal weighting of either of these two factors.
──────────────────────────────────────────────────────────────────────────────────────────────────────────────────
Last edited by Flake on Fri Nov 09, 2018 10:32 am, edited 377 times in total.
User avatar
Flake
Benefactor
Benefactor
 
Posts: 985
Joined: Tue Oct 28, 2014 8:34 am
Location: England, UK

Re: New ELO System Rework Idea

Postby BasicFourLife » Wed Jan 10, 2018 7:32 am

Hmm I don’t understand at all what are you talking about but I would rather have larger elo drops / gains, because getting + 1 - 5 elo per game is pretty annoying and not rewarding at all.
But I am not sure if your win rate is the thing that shoulf affect your elo, because Town of Salem is luck based game and only 10 games is a small sample.
Which I disagree on is that elo should be based on your X % amount of win rate.
1) Elo should be calculated and more affected by your Faction’s winrate (a very little bit about your role also, but if your role would also affect your win rate it would mean BMG is confirming that their roles are unbalanced and broken since all roles in the same faction should have around the same elo gain / loss).
2) Then it should take affect into your opponents elo.
3) Then your teammates, since you could have very high elo teammates and low elo opponents.

This is not the type of elo change I agree on.
BasicFourLife
[Forum Mafia XVI] Winner
[Forum Mafia XVI] Winner
 
Posts: 1532
Joined: Sat Dec 17, 2016 6:47 am
Location: UTC +2/+3 - Somewhere hidden

Re: New ELO System Rework Idea

Postby Flake » Wed Jan 10, 2018 8:16 am

BasicFourLife wrote:Hmm I don’t understand at all what are you talking about but I would rather have larger elo drops / gains, because getting + 1 - 5 elo per game is pretty annoying and not rewarding at all.
But I am not sure if your win rate is the thing that shoulf affect your elo, because Town of Salem is luck based game and only 10 games is a small sample.
Which I disagree on is that elo should be based on your X % amount of win rate.
1) Elo should be calculated and more affected by your Faction’s winrate (a very little bit about your role also, but if your role would also affect your win rate it would mean BMG is confirming that their roles are unbalanced and broken since all roles in the same faction should have around the same elo gain / loss).
2) Then it should take affect into your opponents elo.
3) Then your teammates, since you could have very high elo teammates and low elo opponents.

This is not the type of elo change I agree on.


Hi, thanks for responding! ELO drops/gains have been rectified and are much higher than the current ELO system.

As for the sample size, I have updated it to 20 as I am aware 10 was too small. I also believe firmly that Town of Salem is not entirely luck and requires skill, which I believe can be shown through a players win % after a significant amount of games. Saying Town of Salem is entirely luck implies no ELO system should be implemented at all, since ELO is supposed to measure skill.

Point 1) I have rectified this issue in OP

Points 2) & 3) I have rectified these issues in OP

While I partially understand how you feel about using win rate as a means of determining ELO, at the same time I disagree with you, because the current system takes the amount of games played into account FAR too much. It's so much grinding because there is no exponential rise in ELO with a low amount games (which it should be), so the current ELO system rewards a lot of games. There should be an exponential rise in ELO for a smaller amount of games played, to reward actual skill as opposed to grinding.

While there are some flaws with this, which is inevitable for any ELO system given the nature of Town of Salem, I believe the flaw of the current system (hugely grind based) will probably be far worse than the flaws of this proposed system when it is polished.

Thanks :)
Last edited by Flake on Fri Jan 12, 2018 2:52 pm, edited 4 times in total.
User avatar
Flake
Benefactor
Benefactor
 
Posts: 985
Joined: Tue Oct 28, 2014 8:34 am
Location: England, UK

Re: New ELO System Rework Idea

Postby BasicFourLife » Wed Jan 10, 2018 8:27 am

Yeah I know skill is non excistant in current ToS since the only thing you have to do is play a lot of games and you can have a 33% and still get to 1500+
BasicFourLife
[Forum Mafia XVI] Winner
[Forum Mafia XVI] Winner
 
Posts: 1532
Joined: Sat Dec 17, 2016 6:47 am
Location: UTC +2/+3 - Somewhere hidden

Re: New ELO System Rework Idea

Postby Flake » Wed Jan 10, 2018 8:38 am

BasicFourLife wrote:Yeah I know skill is non excistant in current ToS since the only thing you have to do is play a lot of games and you can have a 33% and still get to 1500+


Mhm, and that's the exact fatal flaw with the current ELO system. It rewards gains over losses and has no exponential growth with a lower amount of games, which it should do.
Last edited by Flake on Wed Jan 10, 2018 1:10 pm, edited 1 time in total.
User avatar
Flake
Benefactor
Benefactor
 
Posts: 985
Joined: Tue Oct 28, 2014 8:34 am
Location: England, UK

Re: New ELO System Rework Idea

Postby Joacgroso » Wed Jan 10, 2018 8:56 am

So are you saying a win as NK should worth the same as a town win? Or did I missunderstood something?
Sorry if I have spelling mistakes. English is not my main language.
I added a petition to my achievement rework!
Plese come and vote to support it!

If you have any suggestions related to achievements, please post them on my thread so we can suggest them together!
Also, support this suggestion: Total elo rework, by Flake
User avatar
Joacgroso
Arsonist
Arsonist
 
Posts: 2623
Joined: Mon May 02, 2016 6:21 pm
Location: Argentina

Re: New ELO System Rework Idea

Postby Flake » Wed Jan 10, 2018 9:28 am

Joacgroso wrote:So are you saying a win as NK should worth the same as a town win? Or did I missunderstood something?

Hi! I now realised my reasoning from before was flawed because even a sample of 200 games is too small to not take difficulty of NK/NE into account, so I have now updated the ELO system with new special ELO gains/losses for NE/NK. Thank you for the reply :)
Last edited by Flake on Fri Jan 12, 2018 3:01 pm, edited 1 time in total.
User avatar
Flake
Benefactor
Benefactor
 
Posts: 985
Joined: Tue Oct 28, 2014 8:34 am
Location: England, UK

Re: ELO Rework

Postby theyellowgreninja » Wed Jan 10, 2018 4:23 pm

I really like this. I feel like it gives people a fighting chance if they get paired with some players who aren't the best at the role they're playing a few times, which, right now, spirals a little our of control. Also, bigger differences in ELO gains/losses means you'll feel rewarded for playing well, and more punished for playing poorly (even though you really wouldn't be, as the differences in ELO ranks [Gold, Siver, ect.] would also be made greater, I assume). It also takes into account how well you play, not as much how MUCH you play, which is super important, I think, as it makes players who are really good but just started the game rise up and be where their skill lever says they'd be
However, would you increase the Placement Game size from 10 to 30? Or would your first 20 "ranked" games also be used to place you? I feel like 30, while a good, big number, is a lot of plcaement games, and 15-20 could be better in the case of people just wanting to see their ELO.


Forum Mafia:
3 - 6 Record Spoiler: NFM40 - Citizen died N3 - Mafia/Witch Win
NFM41 - Consigliere lived - Mafia Win
SFM28 - Citizen lived - Mafia/Neutral Win
NFM44 - Serial Killer died N1 - Mafia win
VFM22 - Citizen died N3 - Mafia/Serial Killer win
SFM39 - Braum lived - Demacia (Town) win
GFM2.43 - Common Myeloid Progenitor Colony (Ret) lived - Neoplasm (NK) win
SFM56 - Ninja Killed by Veteran N3 - Mafia win
VFM48 - Citizen lynched D6 - Mafia Win

You may also know me as HighQualityRips or Henry Hatsworth in Ranked or All Any
User avatar
theyellowgreninja
Witch
Witch
 
Posts: 56
Joined: Sat Oct 01, 2016 11:49 am
Location: Your Subconcience

Re: ELO Rework

Postby Purple998 » Wed Jan 10, 2018 4:29 pm

I like this rework, it makes it more fair.

/support
Who needs a "GF", when you have your own hands? -Mafioso




If you read this yu gai xd
User avatar
Purple998
Veteran
Veteran
 
Posts: 444
Joined: Wed Apr 05, 2017 12:00 pm
Location: Behind you

Re: ELO Rework

Postby Flake » Thu Jan 11, 2018 4:16 pm

theyellowgreninja wrote:I really like this. I feel like it gives people a fighting chance if they get paired with some players who aren't the best at the role they're playing a few times, which, right now, spirals a little our of control. Also, bigger differences in ELO gains/losses means you'll feel rewarded for playing well, and more punished for playing poorly (even though you really wouldn't be, as the differences in ELO ranks [Gold, Siver, ect.] would also be made greater, I assume). It also takes into account how well you play, not as much how MUCH you play, which is super important, I think, as it makes players who are really good but just started the game rise up and be where their skill lever says they'd be
However, would you increase the Placement Game size from 10 to 30? Or would your first 20 "ranked" games also be used to place you? I feel like 30, while a good, big number, is a lot of plcaement games, and 15-20 could be better in the case of people just wanting to see their ELO.


I agree, and I have decreased the amount of required games to 20 now. Thanks for the feedback!

Purple998 wrote:I like this rework, it makes it more fair.

/support


Thank you for the feedback!
User avatar
Flake
Benefactor
Benefactor
 
Posts: 985
Joined: Tue Oct 28, 2014 8:34 am
Location: England, UK

Re: (PLEASE GIVE FEEDBACK AND/OR REVIEW) Total ELO Rework

Postby fable979 » Thu Jan 11, 2018 8:33 pm

Too much maths. I kennot.
----------------Subscribe to Vsefotonz on Youtube---Copy and paste this rignt now!!!
User avatar
fable979
Lookout
Lookout
 
Posts: 90
Joined: Sun Jun 26, 2016 12:28 pm

Re: (PLEASE GIVE FEEDBACK AND/OR REVIEW) Total ELO Rework

Postby Sting » Thu Jan 11, 2018 9:51 pm

I fully agree. The "K factor" in this Elo system is a massive mistake. I'm top of the board right now with a 71% winrate, 312 games and 2004 Elo. Around Diamond my Elo gains were around +2 on average, and I gave those who I lost to +8. Now that I'm master, I get +1 for every single win which is not a good system. It means that somebody at a 50-55% winrate can just play 2x my games and be on the same Elo as me (which as I type, is happening. Won't name the person tho) Yes, I should be expected to win the majority of my games at a higher elo (as my winrate implies I do) but the elo/loss gains are simply not worth it once you reach Diamond/Master, I would have to win 130 games to go up 100 Elo (around 30-40 hours of gameplay; assuming 15-20 minutes per game) while every loss would give those in tiers below me a big boost. Yes, it's sort of defined and makes sense as it's an Elo system, but this is a team game where one player misplaying can lose you the game. Because of this, I think there should be a larger emphasis on the winrates of the roles you win as, right now there is too much of an weight on your rank aka K Factor.

It's more efficient for me to not play and wait for players to get Master since every win is +1 and I give opposition +9. Which isn't a very good system, it punishes people who can rise faster as they win more.
Sting
Medium
Medium
 
Posts: 152
Joined: Tue Aug 05, 2014 2:38 am
Location: Eire

Re: (PLEASE GIVE FEEDBACK AND/OR REVIEW) Total ELO Rework

Postby Flake » Fri Jan 12, 2018 7:23 am

Sting wrote:I fully agree. The "K factor" in this Elo system is a massive mistake. I'm top of the board right now with a 71% winrate, 312 games and 2004 Elo. Around Diamond my Elo gains were around +2 on average, and I gave those who I lost to +8. Now that I'm master, I get +1 for every single win which is not a good system. It means that somebody at a 50-55% winrate can just play 2x my games and be on the same Elo as me (which as I type, is happening. Won't name the person tho) Yes, I should be expected to win the majority of my games at a higher elo (as my winrate implies I do) but the elo/loss gains are simply not worth it once you reach Diamond/Master, I would have to win 130 games to go up 100 Elo (around 30-40 hours of gameplay; assuming 15-20 minutes per game) while every loss would give those in tiers below me a big boost. Yes, it's sort of defined and makes sense as it's an Elo system, but this is a team game where one player misplaying can lose you the game. Because of this, I think there should be a larger emphasis on the winrates of the roles you win as, right now there is too much of an weight on your rank aka K Factor.

It's more efficient for me to not play and wait for players to get Master since every win is +1 and I give opposition +9. Which isn't a very good system, it punishes people who can rise faster as they win more.


Agreed! Thank you so much for the feedback.

fable979 wrote:Too much maths. I kennot.

hahaha
Last edited by Flake on Fri Jan 12, 2018 12:45 pm, edited 1 time in total.
User avatar
Flake
Benefactor
Benefactor
 
Posts: 985
Joined: Tue Oct 28, 2014 8:34 am
Location: England, UK

Re: Total Elo Rework (PLEASE GIVE FEEDBACK AND/OR REVIEW!)

Postby Flake » Fri Jan 12, 2018 11:50 am

NOTICE: I've now changed the rework a large amount. It is now dependent on faction win % and so is self-sufficient as a result, meaning the ELO Gain/Loss equation would never have to be altered with the introduction of balance changes. This also increases the accuracy of the rework. Many other changes have been added also.
User avatar
Flake
Benefactor
Benefactor
 
Posts: 985
Joined: Tue Oct 28, 2014 8:34 am
Location: England, UK

Re: Total Elo Rework (PLEASE GIVE FEEDBACK AND/OR REVIEW!)

Postby theyellowgreninja » Fri Jan 12, 2018 2:47 pm

Oh dang I didn't even realize it wasn't self-sufficient but now it looks super nice, 20 is a decent Placement Games number, I really like this
/Fullsupport


Forum Mafia:
3 - 6 Record Spoiler: NFM40 - Citizen died N3 - Mafia/Witch Win
NFM41 - Consigliere lived - Mafia Win
SFM28 - Citizen lived - Mafia/Neutral Win
NFM44 - Serial Killer died N1 - Mafia win
VFM22 - Citizen died N3 - Mafia/Serial Killer win
SFM39 - Braum lived - Demacia (Town) win
GFM2.43 - Common Myeloid Progenitor Colony (Ret) lived - Neoplasm (NK) win
SFM56 - Ninja Killed by Veteran N3 - Mafia win
VFM48 - Citizen lynched D6 - Mafia Win

You may also know me as HighQualityRips or Henry Hatsworth in Ranked or All Any
User avatar
theyellowgreninja
Witch
Witch
 
Posts: 56
Joined: Sat Oct 01, 2016 11:49 am
Location: Your Subconcience

Re: Total Elo Rework (PLEASE GIVE FEEDBACK AND/OR REVIEW!)

Postby Joacgroso » Fri Jan 12, 2018 3:18 pm

I'm starting to read the OP, but I admit I don't know much about balancing elo, so I can't judge if this is good or not (although it is certainly better than the current one, so I support it). There are 2 things I want to say:
1. Why would you create a second thread for the same subject? You can add the poll here.
2. Why is the average elo 1500? Wouldn't it be easier if it were 0 and people could have negative elo?

edit: Maybe you explained this in later sections (I'm reading the 2nd), but if the suggested equation can only work if you have +20 games played, how will it work during the placement matches?
edit2: I can't say I understand all that maths, but I like this suggestion. I will link it in my signature.
Sorry if I have spelling mistakes. English is not my main language.
I added a petition to my achievement rework!
Plese come and vote to support it!

If you have any suggestions related to achievements, please post them on my thread so we can suggest them together!
Also, support this suggestion: Total elo rework, by Flake
User avatar
Joacgroso
Arsonist
Arsonist
 
Posts: 2623
Joined: Mon May 02, 2016 6:21 pm
Location: Argentina

Re: Total Elo Rework (PLEASE GIVE FEEDBACK AND/OR REVIEW!)

Postby Flake » Fri Jan 12, 2018 4:44 pm

Joacgroso wrote:I'm starting to read the OP, but I admit I don't know much about balancing elo, so I can't judge if this is good or not (although it is certainly better than the current one, so I support it). There are 2 things I want to say:
1. Why would you create a second thread for the same subject? You can add the poll here.
2. Why is the average elo 1500? Wouldn't it be easier if it were 0 and people could have negative elo?

edit: Maybe you explained this in later sections (I'm reading the 2nd), but if the suggested equation can only work if you have +20 games played, how will it work during the placement matches?
edit2: I can't say I understand all that maths, but I like this suggestion. I will link it in my signature.

Hey! Thanks for the feedback.

1. Good point, I didn't know you could edit in a poll to be honest, I thought you had to make a new thread to add the poll haha
2. That is actually VERY interesting. I guess I used 1500 as an average so that the Elo system could be compared to Elo systems like Chess, but this would actually make more sense!

3. I actually forgot to explain this! I'll explain it now and add it ASAP:

Basically, the Approximate Elo score formula, 1500 + 80ln(X-17.5) * (z) + 20*((T)*((AVG% - T%)/AVG%) + (M)*((AVG% - M%)/AVG%) + (NK)*((AVG% - NK%)/AVG%) + (NE)*((AVG% - NE%)/AVG%)), would be used for the 20th game. In other words, you don't actually gain or lose anything from game to game in the first 19 games, but on the 20th game you get given an ELO using the formula I stated above.

EDIT: I've updated the Original Post explaining this.

4. Thank you very much! Much appreciated :) I will also add your suggestion to my sig.

theyellowgreninja wrote:Oh dang I didn't even realize it wasn't self-sufficient but now it looks super nice, 20 is a decent Placement Games number, I really like this
/Fullsupport

Thanks! Much appreciated :)
User avatar
Flake
Benefactor
Benefactor
 
Posts: 985
Joined: Tue Oct 28, 2014 8:34 am
Location: England, UK

Re: Total Elo Rework (PLEASE GIVE FEEDBACK AND/OR REVIEW!)

Postby Joacgroso » Fri Jan 12, 2018 11:16 pm

Something else: Would people's elo be affected if they die n1? I think it would be pretty unfair losing/gaining elo despite you didn't influence the game at all.
Sorry if I have spelling mistakes. English is not my main language.
I added a petition to my achievement rework!
Plese come and vote to support it!

If you have any suggestions related to achievements, please post them on my thread so we can suggest them together!
Also, support this suggestion: Total elo rework, by Flake
User avatar
Joacgroso
Arsonist
Arsonist
 
Posts: 2623
Joined: Mon May 02, 2016 6:21 pm
Location: Argentina

Re: Total Elo Rework (PLEASE GIVE FEEDBACK AND/OR REVIEW!)

Postby Jackparrot » Sat Jan 13, 2018 7:37 am

Flake wrote:NOTICE: I've now changed the rework a large amount. It is now dependent on faction win % and so is self-sufficient as a result, meaning the ELO Gain/Loss equation would never have to be altered with the introduction of balance changes. This also increases the accuracy of the rework. Many other changes have been added also.


I dont think it should be dependant on faction win rate
I am not as active as I used to be, although I should be somewhat active.
Jackparrot
Sheriff
Sheriff
 
Posts: 580
Joined: Thu Mar 30, 2017 10:16 am

Re: Total Elo Rework (PLEASE GIVE FEEDBACK AND/OR REVIEW!)

Postby Flake » Sat Jan 13, 2018 7:37 am

Joacgroso wrote:Something else: Would people's elo be affected if they die n1? I think it would be pretty unfair losing/gaining elo despite you didn't influence the game at all.

Fair point. I could introduce a k factor in the equation to factor in N1 deaths and N2 deaths. For example, k = 0.1 for a N1 death, k = 0.75 for a N2 death (but NOT D2 death), k = 0.75 if revived and k = 1 if you die any point after N2, or die D2.

Jackparrot wrote:I dont think it should be dependant on faction win rate

Care to explain why?
User avatar
Flake
Benefactor
Benefactor
 
Posts: 985
Joined: Tue Oct 28, 2014 8:34 am
Location: England, UK

Re: Elo Rework (PLEASE GIVE FEEDBACK AND/OR REVIEW!)

Postby ProfessorArceus » Sat Jan 13, 2018 8:48 am

Faction elo gain/loss should not be different, as we have no recent statistics to say which faction is harder or easier to win as.
I'm your average British fuckup.
Spoiler:
2:27, 'Fuck up Jeremy Corbyn' ~ The average YouTube comments section.

STRONG AND STABLE GOVERNMENT Y'ALL


FM Stuff
Spoiler: I'm not shameless.
User avatar
ProfessorArceus
Investigator
Investigator
 
Posts: 252
Joined: Sun Jul 24, 2016 10:28 am
Location: At night, in your bed, at morning, in your closet

Re: Elo Rework (PLEASE GIVE FEEDBACK AND/OR REVIEW!)

Postby Flake » Sat Jan 13, 2018 8:54 am

ProfessorArceus wrote:Faction elo gain/loss should not be different, as we have no recent statistics to say which faction is harder or easier to win as.

But faction win percentage measures EXACTLY how hard a faction is to win as... which can be measured VERY easily by the devs.

Unless you mean that it should not be different to the formula used in this rework and you're replying to Jackparrot? I can't really tell
User avatar
Flake
Benefactor
Benefactor
 
Posts: 985
Joined: Tue Oct 28, 2014 8:34 am
Location: England, UK

Re: Elo Rework (PLEASE GIVE FEEDBACK AND/OR REVIEW!)

Postby Joacgroso » Sat Jan 13, 2018 11:14 am

Also, there should be exceptions to the night 1 k factor. For example people who died to vet/bg, bgs who died guarding, leavers, people who were revived, (vets who didn't alert?), etc.
Personally I think elo gains should depend of faction winrate. Otherwise, a NK win would worth the same as a town win, and that would be very unfair.
Sorry if I have spelling mistakes. English is not my main language.
I added a petition to my achievement rework!
Plese come and vote to support it!

If you have any suggestions related to achievements, please post them on my thread so we can suggest them together!
Also, support this suggestion: Total elo rework, by Flake
User avatar
Joacgroso
Arsonist
Arsonist
 
Posts: 2623
Joined: Mon May 02, 2016 6:21 pm
Location: Argentina

Re: Elo Rework (PLEASE GIVE FEEDBACK AND/OR REVIEW!)

Postby UzayAltay » Sat Jan 13, 2018 1:20 pm

First , /Support .
Second , Add it a k factor . My suggestion is 1.5 or 2 for 0-1000 , 1 for 1000- 2400 and 0.5 over 2400 . The formula will like k * ( Your Formula ) It's Result

1) Easier to Gain ELO in Bronze Elo , so Players will easily Get out of Elo hell , If they are good .
2) Without this, ELO TIERS will be meaningless at one point , because of people can Gain Elo over 2900 or 3000 or more . At chess , there is something like that which starts at 2200 If I am not wrong , and there isn't a 2900 Player yet ( World Champion is between 2800-2900 Stage . )

Also , If I didn't miss , you didn't explain the status of Gaining ELO in replacement matches , If It Just keeped as a statistic , it can be better to lose All matches in Replacement Matches , so your Winrate became %0 , than With normal matches began , your winrate goes up very Fast . You should explain how much we gain/lose ELO in replacement Matches .
28-31 Spoiler: Town Games(18-23)

NFM 50 , NFM 51 ,NFM52 , 14D , 14 E , NFM 54 ,14H ( AF ) ,14G , NFM 55 , NFM 56 , 15C , NFM 57,NFM 58,15F,SFM45,16B , VFM36 , 16D , SFM 47 , VFM38 , NFM62 , 16G ,VFM 39,EpisodeXVII , 17B , 17C , VFM44 , 17D , 17F,18C,18D,18E,VFM55,VFM57,SFM64, 19C,VFM58, VFM59,19D,VFM60,SFM66

Scum Games (10-8)

NFM 48 , NFM 49 , TFM 65 ,TFM 66 , Episode XV ,TFM68 ,VFM 34, NFM61 , VFM42 , SFM53 , VFM43,17E,VFM49,SFM60,CFM19,VFM54,EpisodeXIX,SFM63

Ongoing :
UzayAltay
[Forum Mafia XV] Winner
[Forum Mafia XV] Winner
 
Posts: 567
Joined: Tue May 02, 2017 10:17 am

Re: Elo Rework (PLEASE GIVE FEEDBACK AND/OR REVIEW!)

Postby Joacgroso » Sat Jan 13, 2018 2:12 pm

He already explained that. In the first 19 matches, there would not be any elo gain/loss. After the 20th match, the formula would start to work using the stats from the previous games.
Sorry if I have spelling mistakes. English is not my main language.
I added a petition to my achievement rework!
Plese come and vote to support it!

If you have any suggestions related to achievements, please post them on my thread so we can suggest them together!
Also, support this suggestion: Total elo rework, by Flake
User avatar
Joacgroso
Arsonist
Arsonist
 
Posts: 2623
Joined: Mon May 02, 2016 6:21 pm
Location: Argentina

Next

Return to Suggestions

Who is online

Users browsing this forum: No registered users and 3 guests