Back to Announcements

ELO Rating System

powerofdeath
almost 7 years

Link to the Rankings and Game History:

goo.gl/FEMBy9

What is this?

Starting Friday September 15, I will be keeping track of every Survivor and Big Brother Games.

Everyone will start with an ELO rating of 1000.

When you place in the top 50%, your ELO will increase, and when you place in the bottom 50%, your ELO will decrease.

After 30 Days, the person with the highest ELO will win $25 from me and all the bragging rights!

The game must be either Survivor, or Big Brother. It cannot be TWL, TAR, or any other game variant. The size of the games must have 10 players or more or it will not be ranked for this system.

This system is just for fun, and I'm excited to see how this turns out. Ci'ere did this back in 2015 when cosmoiswatching dominated many games. (I dont know if he actually won or just did very good)

This system will last from Morning games on Friday September 15, and ends whatever game is hosted before Sunday October 15 turns to Monday.

Further explanation of how ELO rating works:

I take account of few things:

1) Your current ELO against the Game's Average ELO.
2) The Size of the Game you played in.
3) You actual placement in those games.

Let say you played well in the past, and increased your ELO to 1200. Then you are placed in a game filled with players who didnt do so quite good. The game's Average ELO is around 1000. If you win, you wont gain as much points as someone who's current ELO is around 900. However, if you perform badly, you will lose much more points than those who would lose with 900 ELO. This is also true for opposite. A player with low ELO in higher Average ELO games will win a lot more points if they won the game compared to others, and lose less points compared to others if they get really early boot.

Game Size MATTERS! You will gain a lot more points if you win larger games, like 20 players compared to a smaller games like 12 players. You also loses a lot more points if you perform badly in larger games compared to smaller games.

Maximum amount of points you can get for 10 players game is 25 points! For each additional players, you add 5 points to the maximum points. So for a 14 players game, it would be 45 points, and for 19, it would be 70 points. This is before scaling your current ELO against the Average ELO. So you CAN get more than the maximum value if you pull an upset victory or flops in low ELO games.

1st place would gain the maximum value ELO while the first boot would lose the maximum value ELO. Most of the times, if you place in the top 50% but didnt win, you usually get a fraction of the maximum value. Same goes to the bottom 50%. They will lose a fraction of the maximum value.

Winners of each games will instead get 1.5x bonus while everyone get 1x. Hopefully this encourage players a bit more to actually try to win instead of just getting as far as you can.

If you hosted a game AT LEAST once this season, you will automatically increase your ELO by 25. You do not get any more bonuses if you host more than 1 game. Hopefully this will encourage more different players to host this month. More variety of hosts is healthier for this lobby. Hopefully this also prevent from discouraging players to not host because they are afraid of not being able to farm ELO points.

Schikgil
over 6 years

Knife says


powerofdeath says

megaraptor123



megaraptor789's evil twin


Up there with his other french evil twin Megaraptore
Knife
over 6 years

powerofdeath says

megaraptor123



megaraptor789's evil twin
powerofdeath
over 6 years

lichi says

Just noticed this, but why do I have 1000 points if I've played one game?


probably being right in the middle, like 8th place in 15 players game

fun fact: jpriced was tied for first place with 1138 with tillie and megaraptor123

then his elo dipped a little
lichi
over 6 years
Just noticed this, but why do I have 1000 points if I've played one game?
deletedover 6 years
sweet i didnt know this was a thing or i wouldve tried hard
powerofdeath
over 6 years
updating
powerofdeath
almost 7 years
am now updating this
jack
almost 7 years

qrstuv says

what does ELO stand for


electric light orchestra
Voice
almost 7 years

Snow says

Elo is a very flawed system for something like Survivor. I get that you're trying to rank people, but it was designed for chess, a system with only two players and two outcomes. Even if you ignore the two players thing, you're also dealing with the struggle of Survivor having more than two outcomes.

A player who consistently gets just above halfway in games will climb Elo faster and higher than someone like Tucker, who typically either places top 3 or gets booted early for being too big of a threat. Tucker's the better player.

Also, I'm really looking forward to all the dinguses who overvalue this thread and use it as the ultimate judgement of the strength of a player.


great example but ive never won a game so
powerofdeath
almost 7 years
Obviously its a bit flawed. We all knew why we took down the Trophy System before the Trophy Games.

Like I said, this is purely for fun, and I want to see how people would do, how meta change a bit, just for one month. This is just a temporarily thing.

One of the pros about this is census. By the end of the month, we actually get to see how many different players we have over one month period, which is kinda cool.

"Also, players would need to understand that ELO does not measure skill. Rather, it's a measure of how consistently you can get far in games." -Highspace

I knew this already.

Also, this isnt for trophies, so I honestly doubt people would be as bad like they were back then when people actively target players who perform well.

If I do find this to start affecting players. I could probably keep the last 2 weeks hidden, then reveal the final result.
Otherscott
almost 7 years
I kind of think this would be better as a thing to go back through and calculate without warning after the fact rather than pick a starting point if you want prizes to be awarded for like a month.

Because HighSpace is right, you aren't getting independent results, the ELO is affecting people's strategies.
HighSpace
almost 7 years
Well that's annoying, all my mini survivors have always been 9 players. Oh well!
FFSierraDamnThomas
almost 7 years

HighSpace says

Hey where's the 7 player game I hosted >:(

Anyways, how I feel about ELO in survivor... I think it would work as a thing that runs year-round, with no monthly prizes or whatever. Because right now, it incentivizes the players in the top spots to target the other players in top spots, not because of the game, but in order to climb to #1.

Also, players would need to understand that ELO does not measure skill. Rather, it's a measure of how consistently you can get far in games.

I would also skew the ELO calculating system so that the points lost between 15th-24th is relatively similar, while points gained between 1st-10th has a greater difference.


only 10 player games and up will be put into the system
HighSpace
almost 7 years
Hey where's the 7 player game I hosted >:(

Anyways, how I feel about ELO in survivor... I think it would work as a thing that runs year-round, with no monthly prizes or whatever. Because right now, it incentivizes the players in the top spots to target the other players in top spots, not because of the game, but in order to climb to #1.

Also, players would need to understand that ELO does not measure skill. Rather, it's a measure of how consistently you can get far in games.

I would also skew the ELO calculating system so that the points lost between 15th-24th is relatively similar, while points gained between 1st-10th has a greater difference.
powerofdeath
almost 7 years
updated with bboy29/ BB game
MegaRaptor789
almost 7 years
https://en.wikipedia.org/wiki/Elo_rating_system
deletedalmost 7 years

qrstuv says

what does ELO stand for


It's a guy's last name I think?
Tartosis
almost 7 years
Glicko is neglected
hedger
almost 7 years
yeah, ELO is a terribly flawed system with multiple variables but let the man have his fun.
Snow
almost 7 years
Elo is a very flawed system for something like Survivor. I get that you're trying to rank people, but it was designed for chess, a system with only two players and two outcomes. Even if you ignore the two players thing, you're also dealing with the struggle of Survivor having more than two outcomes.

A player who consistently gets just above halfway in games will climb Elo faster and higher than someone like Tucker, who typically either places top 3 or gets booted early for being too big of a threat. Tucker's the better player.

Also, I'm really looking forward to all the dinguses who overvalue this thread and use it as the ultimate judgement of the strength of a player.
Gocubsgo1908
almost 7 years
Seem like good ideas, prevents people from intentionally be a goat to go deep in the game.
powerofdeath
almost 7 years
Here is the updated list of everyone's ELO!

goo.gl/FEMBy9

I decided to tweak a few things. Winners of each games now get 1.5x bonus while everyone get 1x. Hopefully this encourage players a bit more to actually try to win instead of just getting as far as you can.

I also added that if you hosted a game AT LEAST once this season, you will automatically increase your ELO by 25. You do not get any more bonuses if you host more than 1 game. Hopefully this will encourage more different players to host this month. More variety of hosts is healthier for this lobby. Hopefully this also prevent from discouraging players to not host because they are afraid of not being able to farm ELO points.
powerofdeath
almost 7 years

UltraAug says

i'll play if you will allow me to use alts


I'm not sure whether or not to allow alts share ELO with main accounts, or have them have separate ELO
powerofdeath
almost 7 years

NYGiants says

Will average lobby ELO be taken into account when awarding ELO for games, will "harder" games give greater rewards?

Example:

Person A has 1500 ELO, they win a 15 person vivor where the lobby average ELO is 1200.

Person B has 1500 ELO, they win a 15 person vivor where the lobby average ELO is 1800.

Both players are of the same skill level, but player B had better players in their game.


Most likely.
NYGiants
almost 7 years
Will average lobby ELO be taken into account when awarding ELO for games, will "harder" games give greater rewards?

Example:

Person A has 1500 ELO, they win a 15 person vivor where the lobby average ELO is 1200.

Person B has 1500 ELO, they win a 15 person vivor where the lobby average ELO is 1800.

Both players are of the same skill level, but player B had better players in their game.