Boards›Gameplay

What is ELO Hell?

Great Ozymandius··2 votes·712 views

This post is long, but illuminating, I think.

People point out that if you are playing below your 'true' elo then you will climb. And that's true.

ELO hell is when it takes way too long.

Let's write down a simple model.

Suppose the team with the most points wins, and a team gets a point by 'winning' a position. If your team wins top lane, loses jungle, wins adc, wins support, and loses mid, then your team gets 3 points, the other team gets 2, and your team wins.

Suppose you always win your lane. And the other lanes are determined by a 50/50 coin flip.

Then you lose if and only if three or four of the other lanes lose. That's a a 5/16 probability. So you have an 11/16 or 68% probability of winning. Pretty good. You'll climb elo quite fast that way.

But now let's say the lanes are not weighted equally.

Mid lane is worth 3 points and the other lanes are worth 1 point. So the only way to overcome a mid lane victory is if all the other lanes go the other way. That is, mid lane is three times as important as any other lane.

Now, suppose you are top lane and you win top lane every time. If your mid lane wins you are guaranteed to win (and there is a 50% chance your mid lane wins). If your mid lane loses (50% chance), then your only chance of winning is if jungler, support, and adc also win. There's a 1/8 probability of that occurring. So your probability of winning is 0.5 + 0.5 * (1/8), or 9/16. That is 56% chance of winning. While you'll still climb, it will take a lot longer. Moreover, it assumes you win your lane EVERY time. If instead you win your lane only 70% of the time, your odds of winning the game drop to about 53 percent. Not good. It will take a long time to climb elo at that rate.

Not all roles and not all champions are weighted equally. Assassins, especially back when they snowballed super hard in previous seasons, are like the mid lane example above. They have a huge amount of agency and they take it away from other players. If you were playing a tank top back in those seasons, you had little agency. Even if you won lane the vast majority of the game, it wouldn't be enough to make up for the other team's mid lane assassin who snowballed out of controlled and used it to destroy bot by roaming more often than your tp was up.

This is ELO hell -- when you play champions or roles that are not as impactful and consequently climb slowly even when you play your role well.

Now suppose you are the mid lane assassin but you aren't that much better than average. You win lane 60% of the time.

Now you win the game with probability 0.6*(1-1/8) + 0.4*1/8 = 57.5% probability. That means you win games more often by winning mid lane 60% of the time than if you played and won top lane every game. (!) It's better doing only a bit better than average in a role that is very important than it is to do extremely well in a role that is not as important.

Of course the model is simplified. Your contribution to the team is not just winning the lane; you can win the lane to different degrees; and which lanes or champions are most impactful shifts with the meta. But the lesson still applies:

(1) feast-or-famine champs are bad for the game because they take away agency for all the other players on the team.

(2) ELO hell is real and it is when your teammates play champions with more agency or lanes of more importance than your champions or lane. If you truly are better than your ELO then you want to play the high agency champions and play the most important positions in the given meta.

**Summary: ELO hell is when you climb elo slowly because you play champions or roles that have low agency. Feast-or-famine champions have high agency and are bad for the game because they substantially reduce the agency of all other champions. If you are better than your elo and you want to climb elo, then you should play feast-or-famine champions because they have more agency. **

13 Comments

Embertine3 votes

Why are you trying to make a mathematical equation for something as random and un-math related as a human being's actions and reactions according to various ever-changing elements in the game?

Your entire equation is thrown off simply from the person decided to either keep attacking when he's about to die or retreat.

Great Ozymandius1 votes

I'm guessing you don't like economics.

Great Ozymandius1 votes

Why are you trying to make a mathematical equation for something as random and un-math related as a human being's actions and reactions according to various ever-changing elements in the game?

MMR uses math to describe a human being's actions and reactions according to various ever-changing elements in the game. By your logic, we can't use math to match players in a game and, in fact, there is no good way to match players.

champion.gg uses math (winrates, pickrates, banrates, etc.) to summarize the strength and popularity of different chapmpions, items, and strategies. By your logic, champion.gg is useless.

GDP measures the effects human being's actions and reactions according to various ever-changing elements in the economy.

Look, I can go on, but the idea that math can't be used to describe people, their circumstances, or their decision-making is simply ignorant.

Ghostbear142 votes

You definitely put a lot of time and thought into that and I tend to agree I just like to simplify it a little. I think elo hell is the point in your climb when you notice you are better than most players in the game but not that much better that you can determine the outcome on your own. You have a much slower climb because you are now at the mercy of your team mates more than your own skill.

Great Ozymandius1 votes

That's a good thought.

Ghostbear141 votes

I can't take credit for it. Saw it in a video by a YouTuber named Foxdrop. I just agree with it.

shyie1 votes

I know numbers play a huge part, but relying on them and using them to create a system called elo hell is wrong. More than probability and team placement, if you're good enough and have a good tilt management system, you can carry a team physically and emotionally. I'm sure you're much better at the game than I am but this is just what I think. I'm close to silver from b5 and I think elo hell is bad, but it's manageable.

Great Ozymandius1 votes

My point is that, yes, you'll climb elo if you play well, but if you play less impactful roles or champions then you will climb elo more slowly, potentially very slowly.

If you were in low bronze you would have a tough time climbing out by playing support. In high elo, supports are quite impactful because the adc knows how to utilize what the support gives him in lane, can snowball harder off of the advantages the support gives him, and the team can better utilize the vision that the support offers. But in low ELO the support is not very impactful (unless he builds damage, but in that case he's still going to be less impactful than a lane champion because the lane champions get more farm to do more damage).

Playing support in low elo is ELO hell because you will climb very slowly even if you play support quite well.

That is the point.

Can i get an upvote please?

The Queen Viper1 votes

[{quoted}](name=Great Ozymandius,realm=NA,application-id=3ErqAdtq,discussion-id=ME4LgqkG,comment-id=,timestamp=2016-04-13T21:05:00.054+0000)

What is ELO Hell?

Fiction.

Great Ozymandius1 votes

Thoughts on what I wrote down? Or just a glib one-liner?

dim2a1 votes

I think you're off target in your analysis in trying to approximate team winning as the # of lanes that win. If you have an intentional feeder or a troll, you probably lose even if all other lanes win. If one lane loses very hard, similar story. So, in games with trolls/etc, your effort makes less difference than in games without those, which is not reflected in your analysis. And those are more likely in low ELOs

I remember a year or so ago someone posting about how he tried to climb from bronze 5 to gold-whatever (partial message was to "proove" that there is no ELO hell). To get to bronze 5, he on purpose tanked all promos by feeding as much as he could (he could only lose 9 actually), got placed in bronze 4, and lost some more games to get to bronze 5 (he said it's oK, since he'll carry teams later). This is prime example of creating ELO hell though -- other 9 players in his tanked games had almost no impact on the outcome of the game. And in later games he carries, it'd be a similar story, though to a lesser degree.

So, ELO hell is a place where climbing takes longer because the other players are too uneven. It is my belief that if you're 1.5-2 tiers below your "true" strength, you'll climb fast no matter where, so then ELO hell does not exist. But it can be frustrating for, say, mid-silver player to be stuck in bronze, and have very slow climb due to trolls/feeders/afks.

Great Ozymandius1 votes

That depends.

The presence of trolls/feeders/afks can potentially help you if you yourself are never a troll/feeder/afk because the other team has 5 potential trolls and your team has only 4 potential trolls. Let's say you always lose when you have a troll or feeder. The math is a bit complicated so let's say we are playing a game with two people on each team, to make it simple. Let's say there is a 50% chance that any player (other than yourself) is a troll. Then there is a only a 1/8 probability that your team has a troll and they have none. There is a 1/4 probability they have one troll and you have one troll. And there is a 1/8 probability they have two trolls and you have one. There is also a 1/8 probability you have no trolls and they have no trolls, a 1/4 probability they have 1 troll and you have none, and a 1/8 probability they have two trolls an you have none.

So there is a 1/8 probability your team has more trolls than their team, a 5/8 probability both teams have the same number of trolls, and a 3/8 probability they have more trolls.

Let's say that you are an average player (except that you never troll or disconnect), so you win 50% of your games when both teams have the same number of trolls. And let's say that the team with more trolls always loses. Then your probability of winning is: 01/8 + 0.55/8 + 1*3/8, or 11/16. That means you'll win 68% of your games even though you are already at the right elo!

If there are a lot of players in the game (way more than 5 per team) and there is a low but non-zero probability of there being a troll, then there is about a 50% chance they will have more trolls and a 50% chance you will have more trolls, so this would definitely hurt you if you are an above average player.

If there is an extremely high probability that other players are trolls (say, 100% probability) and the team with more trolls loses, then you will always win.

Xelnex1 votes

Thank you for making these mathematical stuffzz. I am new to the game and never played ranked yet, but i am practicing and trying out new champs before trying out ranked, because i dont wanna be in elo hell, (i am sure nobody does), so i wont be playing ranked unless i have enough knowledge and skill to do so.

I would like to say i main the adc role, i carry 50% of the time but sometimes my team doesnt build a tank,etc and everything horribly fails end game. So over all i should say i have a 33% win rate as adc. I am a good support, i play soraka and sona, and delay the death time of my allies, but cant make them immortal, so i have to fully depend on my team to carry because i cant (i can only protect), when i am a mid laner, my tower almost never falls, it stays at atleast for 20-30 mins 90% of the time. But the problem is that i cant gank, or i am not good enough.

Can you give me some pointers? i wish to learn, what champ should i pick and where? what are some examples of Feast-or-famine champs? when i play as adc and go against a lux or morgana, i use sivir to block their spells, when its full of skill shots like caitlyn, i go as kalista, as a mid laner against assassins like zed and yasuo, i go heim to out poke them, when against an ap midlaner i go veigar, etc. These are the things i do in game, win most of the time, but its not enough, i wanna be better so please help me with that.

Also who do i ban? as adc or support i ban blitz even when i counter him good most of the time, i just dont have enough courage to face him so i ban. Tell me everything.

[slayer-jinx-catface]

More from Gameplay

3 Refunds per season, upvote if you want 3 refunds per season.
Gameplay764 comments
7,167
votes
3 Refunds per season, upvote if you want 3 refunds per season.
Gameplay156 comments
3,510
votes
UPVOTE FOR URF
Gameplay606 comments
2,440
votes
Yes plz.
Gameplay456 comments
1,721
votes
3 Refunds per season, up vote if you want 3 refunds per season.
Gameplay114 comments
1,716
votes
When is ANYONE at riot going to address the awful new player experience?
Gameplay634 comments
1,383
votes

More by Great Ozymandius

Just because I am in bronze does not mean I cannot give good advice.
Gameplay166 comments
265
votes
Guardian's Angel is a tank item. Doesn't this seem strange?
Gameplay141 comments
79
votes
Buffing armor items will kill lethality building marksmen -- and that's a good thing.
Gameplay40 comments
30
votes
I got promoted to bronze IV! More advice, as promised.
Gameplay18 comments
26
votes
Shaco has never been FOTM since his post release nerfs.
Gameplay52 comments
17
votes
Thoughts on rebalancing tanks
Gameplay14 comments
10
votes