I have a game with a ranked mode that keeps track of MMR.

Players have been asking me to have "seasons" so we can reset it and help people who are "stuck in ELO hell" climb better.

My thought is that people who are stuck in elo hell are just bad and that resetting per-seasons are pointless, why not just climb? I'm going to add rewards for the highest rank achieved every 4 months but not reset MMR.

Is there any real reason to reset MMR?

  • \$\begingroup\$ What data do you have which disproves or confirms that "elo hell" is a thing in your game? \$\endgroup\$
    – Philipp
    Aug 11, 2019 at 14:53

2 Answers 2


An "even" playing field

If you put the average player at the same starting point as all other players, there's a small hope of them actually ending up on top. One's logical mind may realise it's completely unrealistic, but this hope may still be there.

If you can see all those players between you and the top, this is much more demotivating than a clean slate (even if the end result would be roughly the same).

This also applies to any rewards and your hope of getting those.

Seeing how you do against the best

This is similar to the above, but goes beyond that.

Often it would be roughly impossible for an average player to get matched against the best of the best players, due to how the matching system works.

It can simply be a nice challenge to directly face one of the best players (or at least a much better player), see how you'd actually do against them, see how they'd respond to what you do and potentially learn from it, even if you get completely demolished.

Resetting rankings gives a much better chance of this happening.

Variation in ranking is a fact, how much is hard to determine

Many games have some element of randomness, having bad days happens and the ranking system may have some issues. This may also disproportionately affect certain players, characters, roles, etc. (e.g. certain characters may heavily rely on randomness or a "support" character may have a harder time supporting teammates that can't fulfil their role, or be in a much better position to highlight the strengths or negate the weaknesses of their team).

Now you may say this effect these factors has on ranking is minor and you'd never be too far from your "true" ranking, and this may be true, but this effect may also be much greater than you imagine.

One way to potentially test this, of course, would be to reset rankings.

Mentality is reality

Believing something to be true can make it true. If a player believes they're in "elo hell", they may be demotivated and end up playing significantly worse than how they'd play under other circumstances (or they may end up saying or doing things that demotivate their teammates, even if they themselves don't play much worse).

How long would it take to climb?

Let's say you're not where you belong. What percentage of games would you win? How long would it take to climb to where you belong at that percentage? Of course you should win a lot more if you're far below from where you belong, but it may still take quite a while to get where you "belong".

There are also high ranking players playing on low ranking accounts, to either just increase ("boost") the ranking or because they enjoy the easy wins. Whether this makes up a noteworthy portion of accounts (and thus affect one's ability to climb at that level) would likely depend on the game and the ranking level.

Players need to actually play

If there's no reset and a top player stops playing or just plays less, they may actually keep getting the top player rewards for a few seasons. A reset forces players to play ever season to get rewarded, which encourages players to play more (increased engagement is good) and makes it more likely for players to rank higher (since they won't be competing with players no longer playing the game).

Making players happy

Sometimes you give in to the requests of players, instead of dismissing them, because making and keeping your players happy is important, even if what they're asking makes no sense to you. Although there is a risk that they're not being realistic and it ends up being a disappointment or your idea of what they want is different from what they actually want. Might the majority of players be disappointed enough with ending up at the same place after a reset to stop playing? I suspect not, but I might be wrong.

As an anecdote, I was quite an avid Dota 2 player a few years ago. I hypothesised that the ranking I was stuck at may not have reflected my ability (especially since I've learnt a ton since getting that ranking). So I created a new account (since there is/was no free ranking reset). With that account I ended up getting a ranking significant higher than my original account (from about 3.5k to 4.5k, if I recall correctly). I suspect demotivation was a big contributor to being stuck.

Although I do suspect most players who believe they're stuck in "elo hell" may be misjudging their own ability and resetting their ranking would likely put them in roughly the same spot.

Note some games have a season ranking separate from your regular ranking, and in some cases allow you to optionally replace your regular ranking with your season ranking (if it's better). Dota 2 is/was one example, I believe. This might be a negate the downside of resetting ranking, which is that players become attached to their ranking and resetting it would frustrate them.


Instead of fixing the symptom (people being dissatisfied with their ELO rank), let's try to find the root cause.

So there are players who feel "trapped in ELO hell". What exactly about their game experience feels like "hell"? Why don't they enjoy playing against players with a similar ELO rating? If we can answer these questions, we might find a much better solution which doesn't lead your whole ranking system ad absurdum.

You are not writing anything about your game mechanics at all, so we can only guess widely here. So here are some of my guesses what I could imagine to be wrong:

  • Players just are frustrated by their lack of advancement. So they are looking for some system to blame. The whole purpose of ELO ranking is that everyone's rank eventually stabilizes and they have a 50:50 win/lose rate. The idea is that this is the way players will enjoy the game the most. However, that's not always the case. Players want visible advancement and they want to win. If this is the case in your game, then some things you can try to tweak are:
    • Track the player's performance with other metrics which only ever go up and never down, like experience points, kill counts etc. Keep the ELO rating for matchmaking. But redesign your UI in a way which de-emphasizes the player's ELO rating and draws attention to the more motivating progression measurements.
    • Make losing fun. When you have a zero-sum game where each match has one winner and one loser, then it is of course impossible to make wins more frequent than loses. But what you can do is lessen the emotional impact of losing. Make sure that the player still has a positive experience and a feeling of accomplishment even if they don't win. This is a more complex issue which requires you to go into details about your game mechanics and how you present them, so if you are interested in more information, please open a new question.
  • Players keep playing against the same opponents. This has two problems: The players can't grow their skills because they are always confronted with the same opponents using the same strategies and they can't grow their ELO rating quickly because they rarely face much stronger opponents. To fix this problems, widen the ELO bracket from which you select opponents. Also consider to take short-term performance into account. When a player is on a winning streak, give them a much stronger opponent so they have the opportunity to outgrow their current ELO rank. When a player is on a losing streak, give them a much weaker opponent to boost their confidence.
  • Players can't get out of "ELO hell" because the game won't let them. They keep losing games against inferior opponents not because they make mistakes but because the random number generator seems to hate them or because your adaptive difficulty within the game itself goes over board and handicaps a winning player too much. To avoid this:
    • Make sure your game is less luck-based and the mechanics are more competitive. The random number generator might spice up the game by creating interesting situations, but it should not decide the game.
    • When one player dominates the other, make sure the game ends soon instead of extending it by giving the losing player an advantage.
  • Players can't get out of "ELO hell" because their teammates won't let them. This is a problem in team-oriented multiplayer games where a team of strong players can lose when there is one bad (or worse: actively obstructive) player on the team who drags them down. If an individual strong player can't decide the game if they are not properly supported by their team, then the win/lose rating of an individual player doesn't accurately reflect their skill level. If that's the case:
    • Look for a way to measure player skill which is not solely bound to the success of the whole team. (but be careful to not encourage behavior which doesn't contribute to winning the match either)
    • Have a way for players to cope with obstructive players on their team and win nevertheless.
    • Provide ways for strong players to decide the match which do not require the cooperation of their team (even if that goes against the philosophy of team-based multiplayer games).

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .