Get rid of ELO?

[Hutzu]

How do you determine when the true Elo is reached?

[ILOCOS_NORTE]

How do you determine when the true Elo is reached?

There is no such thing like an exact true ELO at the moment. Like I said, you can become #1 in a week but with just one bad day you can lose 100+ ELO. You can win just to much ELO with a single game.

But:
Lets say a top player is winning 55% of his games right after the season started. From 100 games he will win 55 and lose 45. In average thats a plus of 150 ELO. His new Elo will be 1150. If the player keeps winning now like 53% if his games he will gain another 90 ELO from his next 100 games. => 1240 ELO. If the players keeps winning with now 51% he will reach 1270. From now on he will win just 50% of his games and wont move anymore. He reached his true ELO.

You see in theory its easy to say where the true ELO is. When a players w/l rate drops (or rises) to 50%.

Sadly one player can win 52% constantly and rise and rise while another player wins 60% in the early season and 50% after just few weeks. There is no omnipotent rule

But I think if we keep our 3 months seasons, all active players should have reached their true ELO.

[ZTX)Fiji]

"ones true elo" changes throughout the season.. it never stays the same omg... also immagin a bad player ending up with 3 good ones in a 4s... sure the other teams chances of winning are slim but the bad player can observe and learn from it.. and get into how the skilled ones plan strategies - then share it and come up with a counter strategy..... it would also elimiate the "oneguycall" type of games... or atleast strengthen the communication xD - theres no such thing as a fixed set of elo points per player - it changes constantly and with our sugestion it just changes more realistically as to how the individual players developes over the span of a season..... drop the idea of reaching a true elo point and then never improving xD it doesnt stick....

[dweiler]

For me personally the Elo-system is just boring. Those are my Elo's in 4/5 seasons: 1104, 1144, 1100, 1133. And now I am at 1108. I just know I will reach 1100 within a week and then play 50% the rest of the season.

Okay, I feel like I am gonna derail my own thread but anyway:

I did not know anything about the Elo-system, but because of yolo.style and tastay I decided to do a bit of research. I must say that it seems to me the usage of it here is highly flawed and very sketchy (and I found out it is Elo instead of ELO, named after the man who invented it, Arpad Elo). This is how I interpreted the flaws as someone who knows little about it, so please correct me if it's wrong:

1. Elo is used for single player games (or unchanging teams) and not for random teams. In random teams Elo has to be calculated for each person individually and it is a point of discussion if this is possible at all. For example, everyone gets the same amount of Elo, whether they carried or put the team down. So in some cases, Elo says nothing more than that one is often carried or "teamkilled". One might think that will straighten out across the league, but because of balancing high Elo player will be more likely to be "teamkilled" and low Elo players more likely to be carried, so it will not straighten out.

2. In team based games the Elo also has to take other factors into account than just wins/losses. For example, a game can be balanced qua wins/losses, but have 'types of players' in one team which do not match: having no caller in a team, or no one who is a good holder, makes your chance of winning much lower than if you do, regardless of your w/l. The Elo gain/loss should take that into account to be fair.

3. Elo is useful for skill-based games: for example in chess there are always the same pieces and same moves and the skill of the player makes the difference. Legion td is not entirely skill based: 3 shit rolls vs 3 pro rolls give a way different chance of winning a game than 6 balanced rolls. Losing 15 Elo when you are totally outrolled (or send luck, and so forth) does not reflect the chances of winning at all. In such games the Elo lost should be way lower - which of course is nearly impossible to implement.

4. Elo is a very refined system, and a small error in the measuring can disturb the fine line it is walking. It can be questioned if the Elo here is anywhere near that fine line. (See Ilocos' comments on Elo changes)

In defense of Elo, I can see Elo having some use as a heroic treatment in stacking/farming in public games. But after reading up, I can't help but feeling that it should not have been implemented in the LIHL in the first place? Maybe someone who knows more of this can elaborate on this?

PS I'm not sure if those findings are in favour of Elo or randomizing, because I do not believe there is any difference between balancing with Elo and randomizing anymore (except for the fact that there is a "rationale" behind the matchmaking and the negatives from that).

[Diablo_]

While ELO isn't perfect for (randomized) team games (and even less perfect for LTD in specific imo, because one player has a very limited impact on the outcome of the game, compared to mobas or other games), it's still the best skill rating system available and has shown to be kinda reliable in many different games.

[dweiler]

While ELO isn't perfect for (randomized) team games (and even less perfect for LTD in specific imo, because one player has a very limited impact on the outcome of the game, compared to mobas or other games), it's still the best skill rating system available and has shown to be kinda reliable in many different games.

Well yeah, perhaps in theory I disagree, but in practice it does work.

As far as I'm concerned, this can be archived. It was fun thinking about the mechanics behind the game, but I'm having fun playing LIHL with Elo for more than a year now, so there is no "emergency reason" to change it. Thanks for all of your opinions on this

PS I can tell Diablo_ did not read my post because he writes ELO instead of Elo.

[Diablo_]

It was fun thinking about the mechanics behind the game

Agreed!

PS I can tell Diablo_ did not read my post because he writes ELO instead of Elo.

I did read it
On the other hand, you should re-do your reasearch to find out that Elo is also wrong, as it's an Elo rating And even that isn't really correct as most/all the Elo-like ratings are at least slightly different to the one described by Elo himself ^.^

Alright, archiving this