An Interesting Little Experiment:

Started at 900 ELO, played 40 games, alternating between using two different champions, playing a total of 20 games with one, and 20 with the other. Final result - 19 wins and 21 losses finishing at 892 ELO rating.

Next, I redid the test, starting at 1500 ELO on another account and playing 40 more games, alternating between using the same two champions as before. Final result - 22 wins, 18 losses finishing at 1519 ELO.

A few Details:

In all games, I played to the best of my ability and never voted to surrender. I've played ~1000 games of SR lifetime. Additionally, since 90% of LoL is trolls, I communicated solely with pings in all games. Finally, if the champion I was supposed to be playing was taken/banned I would dodge that game to keep things consistent.

What this all means:

If we assume their ranking/ELO system is indicative of, well, *anything* then, if I'm a *bad* player, after playing 40 games at 1500 ELO I should have dropped some. Instead, I gained a small amount of ELO.

If I'm a *good* player, after playing 40 games at 900 ELO I should have raised some. Instead, I lost a small amount of ELO.

But neither of these things happened. Instead, the results seem roughly 50% wins, 50% losses, no matter what the ELO, and no matter if I'm a good player or a bad one. This indicates their ranking system is deeply flawed.

What else:

While playing a extremely low or extremely high ELO might give different results (or if the player is extremely terrible or extremely good), the vast majority of LoL players are between 900-1500, and so for the vast majority of LoL players, ELO is meaningless. A 900 player might be as good as, or better than a 1500 player, there is no way to tell because wins and losses depend on a random team, and not on any individual player.

How to Fix:

With the upcoming league/ranking/ELO changes, some might think this issue will soon be solved, but their new system (as I have read and understand it) has the exact same problem as the current one. Which is that you move up or down in the system based on team results only, and your individual performance is mostly irrelevant, assuming you are not at an extreme end of the skill spectrum. The way to solve this is by incorporating a carefully designed *individual* point system like in Dominion, where there is some measure of your personal performance regardless whether your team wins or loses. This would be an imperfect solution of course, as it would encourage players to go for points rather than wins, but if carefully designed so that gaining points correlates to doing helpful things for your team, it would be far, far superior to the current system, which rewards/punishes you based almost exclusively on factors beyond your control.


Happy to answer any questions about the methodology of my little test, or reply to any intelligent comments/questions that make it clear the person actually read/understood my post. Others will be ignored. Thanks for reading!

You're exactly right.

its almost as if the ranking system is largely based on a win/loss outcome, if you get a team that wins you get elo, if you get a team that loses you lose elo. with that said you now have a **** shoot of 5v5, now you can say **** like "your odds are better at winning cause you are 1 of the 5, they have a 5out of 5 chance of people all bads" this is SLIGHTLY true as if you have a team of 5 and they are relatively average players they will most likely **** stomp your team if you are also an average player but have 4 herps you are going to lose, hands down. 1 person cannot beat 5 people of the same skill level, they wil have a hard time beating 2 people of equivalent abilities as is. it just isn't feasible.

Something that might invalidate your experiment is that you have to play differently at different elos; you have to take advantage of people's mistakes more and more the lower Elo you are, or you won't be able to carry your team. Inversely, the higher your Elo, the higher the average number of farmfest games will be where late-game teamfights are most important. So, if you play how you might as a 1500 player at 900 Elo, of course you'll lose more often, but if you play more aggressively, you can bring your winrate way above 50%. I don't think people give individual skill enough credit. To say "your individual performance is mostly irrelevant, assuming you are not at an extreme end of the skill spectrum" seems glaringly wrong to me, and yet so many people agree with you.

Try playing 500 games instead of 40. A mere 40 games isn't enough to judge anything.

Implying 40 games means the long run (it doesn't).

Whether you like it or not, it take many more games for most people to get to a more accurate Elo range. Your experiment is not that astounding. If you can get to 100 games on both accounts, then you will have a lot more credibility.

40 games isn't enough information to base anything on. Many people find their true elo after about 300 games.

Since there is no real reflection of individual skill, a lot of great players who should be higher are stuck in lower brackets and vice versa