Today, ranked gameplay arrives on the Overcast Network. This is the end of casual stats farming, and the beginning of serious, structured competition. This is something the community has been craving for a long time, and something we have been planning for just as long.
For now, we will be ranking solo players, based on their skill at helping to win matches as part of a randomly chosen team.
We are launching ranked gameplay in a fairly simple and experimental form. It has been built to evolve alongside our players and our own understanding.
Where can I play a ranked match?
Ranked matches happen on special dedicated servers named
Ranked2, etc. There are currently four ranked servers on the US network and two on the EU network.
The ranked server rotation contains a selection of objective-based maps, sized around 16-24 players. The server cycles through maps in the normal way, and players can choose to play or observe each match.
Who is allowed to play in a ranked match?
First of all, you need a total of 300 enemy kills in unranked matches before you can join a ranked match. This is simply to prevent very inexperienced players from wandering into ranked matches.
There is also a daily limit of ranked matches for each player. Premium (paying) users are allowed to play up to five ranked matches per day. Non-premium players are allowed two ranked matches per day.
How do I join a ranked match? How are the teams decided?
Ranked matches can only be joined before they start. When you join, you will not be placed on a team right away, rather you will be added to a list of players who wish to participate. Teams will be chosen completely at random from this list. No player can choose their team or switch teams at any time.
After the teams are chosen, there is a "team huddle" period, followed by the actual match start.
Once the teams are chosen, no new players can join for any reason, and players who have joined are committed to finishing the match. You cannot
/leave or switch servers while playing in a ranked match. If you disconnect, you will be automatically returned to the match when you reconnect to the network.
How are players ranked?
When a ranked match ends, every player on the winning team will receive a win in their ranked stats, and every losing player receives a loss. On maps with more than two teams, the winning team will receive a win for each team they defeat. If all teams are tied, no wins or losses are given.
Overall player rankings are based on a rating that is calculated with this formula:
wins / (wins + losses)
However, if a player has played less than 30 matches, their rating will be scaled down based on how many matches they've played. After 30 matches, the rating is exactly the formula above.
What happens if I leave a ranked match?
As soon as a ranked match starts, all participating players are committed to finishing it. No matter what happens, the match will affect the ranked stats of every player who is there at the start. A player who leaves a match for more than a few minutes will receive a forfeit in their stats, as well as two losses. Players who forfeit frequently may be restricted from ranked matches. Just as you wouldn't abandon your team in the middle of a tournament match, you must make the same commitment to your team in a public ranked match.
What are the exact conditions for a forfeit?
You will forfeit the match if ANY of the following are true:
- You left the match for longer than 3 minutes in a single stretch
- You were absent from the match for more than 5 minutes in total
- You were absent for more than 20% of the match
Because of the last condition, shorter matches are easier to forfeit. For example, if the match only lasts 5 minutes, then leaving for more than 1 minute will cause a forfeit.
This may seem harsh, but there needs to be a strong incentive for players not just to stay in the match, but to take precautions against any "accidents" taking them out of the match. If you have an unreliable internet connection, or unpredictable parents, we really do sympathize with you, but there is nothing that we or your teammates can do about these problems. Ultimately, it is up to you to either ensure that you can play the entire match uninterrupted, or don't play at all.
That said, we assume that even the most reliable players will earn the occasional forfeit due to truly unforseeable events. So, don't get all upset if you end up with one or two on your record over the long term. Forfeits are an incentive, not a punishment. And they will disappear after 90 days anyway.
Are the rules any different in ranked matches?
No, the rules and punishments are the same on ranked and unranked servers.
We have removed the rule against spawn killing from our ruleset entirely. While this was an indirect result of ranked servers, the change applies to all servers, ranked or unranked.
On ranked servers, you will be expected to contribute to the match and help your team win. Though it is technically against the rules, we often allow players to goof around on unranked servers and not really play the game, as long as they aren't getting in anyone's way. On ranked servers, this is not acceptable. Neglecting your team is considered a form of team griefing.
Where does the rating formula come from?
The rating formula
wins / (wins + losses) can be thought of as a measurement of the player's probability of winning a match, which is hopefully intuitive enough.
The more matches played, the more accurate that measurement will be. For players with fewer than 30 combined wins and losses, we estimate their rating conservatively by scaling it down. The complete formula looks like this:
wins / max(30, wins + losses)
With no matches played, we assume their rating is 0. As they play more matches, their rating increases towards their actual win rate. After 30 matches (which is an arbitrary number we chose), we have "complete confidence" in their actual rating.
Another way to think of it is that we assume they will lose their first 30 matches until they actually play them.
This is similar to a "Bayesian average", which you can read about it in detail here [actually, it's not very similar anymore, but I will leave this link here for your entertainment]:
Why don't you use ELO or some other fancy ranking system?
ELO and similar ranking systems are, for the most part, designed for one-vs-one competition, where there may be a significant skill imbalance between the two competitors. They don't make sense for teams, particularly not randomly chosen teams, which should be pretty well balanced anyway. There are ranking systems that can be used to rank individual players competing on teams, but they are quite complex and it's not clear that they would serve any purpose with random teams.
We did a lot of research into this topic, and discussed many interesting ideas from ourselves and the community. Ultimately, we decided to start with the simplest system that made sense, and adapt to any problems that emerge in practice.
Why base ranks only on wins and losses? What about kills and objectives?
Because ultimately, winning the match is the only thing that matters. Kills and objectives are strictly the means to that end. That is how the game is defined, and it's how the maps are designed to be played. We have learned from experience that rewarding anything other than winning inevitably compromises the incentive to win and ruins the game.
That said, we have considered using secondary accomplishments, such as kills and objectives, to measure how much a player contributed to a win, but that is a rabbit hole we decided not to go down right now, for reasons explained in the previous answer.
Isn't it unfair that maps with 3+ teams award multiple wins?
No, there is no inherent advantage to playing these maps. They will give you an equal number of wins and losses on average, just like any two-team map.
On a map with four teams, the odds are stacked against you. You will lose three times as often as you win. So, the map must pay out three times as much for winning as it does for losing.
However, there is a slight problem with this. Even though the average wins and losses given by the map are equal to each other, they are not equal to other maps. A two-team map will give you 0.5 wins (and losses) per match, whereas a four-team map will give you 0.75 wins/losses per match (3 for every 4 matches). This has the effect of weighting the map more heavily than other maps, which is not something we particularly want.
In theory, the solution is simply to scale down the payout so it averages 0.5 per match. In that case, a four-team map would give you 2 wins for winning, and ⅔ of a loss for losing. However, that means wins and losses are no longer whole numbers, and we would have to display them as fractional values everywhere. This feels a bit ugly and confusing, though we may end up doing it anyway.
Why only count matches in the last 90 days?
Because we want rankings to reflect the current skill levels of current players. We also want to be able to make adjustments to the ranking system without instantly changing or resetting everyone's stats. The 90 day window provides a natural transition period for any change.
We are, however, strongly considering a "season" system, where the rankings would reset at fixed intervals, and the leaders at the end of each interval would be declared the winners of the season.
Why are ranked matches limited per day?
There are a few important reasons for this. One reason is that we need to make money, and we feel that the ability to play more ranked matches per day is a fair and reasonable thing to sell. The ranking system is designed to NOT favor players who play more matches, beyond the 30 match threshold, which is easy for any player to meet within 90 days. So, being able to play more matches per day should not give anyone an unfair advantage.
But if you're looking for more altruistic reasons, I'll give you a few: If we let players play ranked matches all day long, we're pretty sure they wouldn't put as much effort into them, and wouldn't find them nearly as exciting. Limiting ranked matches makes them more like special events, and prevents players from burning themselves out on competition. It also ensures that the unranked servers will not run out of players. Unranked servers are the place to no-life the night away with your friends.
What if a match gets rushed? Will I get a "refund" so I can play another match?
No, a match always counts, whatever the length, so make sure your defence is solid before you do anything else. If a match ends prematurely, it's because a team failed badly at defence. It will happen occasionally, but players will learn to stop the rush.
If it's truly impossible to defend against the rush on some map, that is an issue with the map that we need to sort out.
Why don't you limit play time per day instead of matches?
This would influence gameplay, and not in a good way. There would be an incentive to finish matches quickly, so more could be played per day, which is irrelevant to the actual game.
Why don't you use player ranks to balance the teams, instead of choosing them randomly?
We would rather avoid doing this, if possible, because it would likely have the effect of pushing everyone's rating towards the average. We suspect that randomly chosen teams will be even enough, though we can revisit this idea if that ends up being a problem.
Will you invalidate a match if something bad happens, like a really bad team griefing incident?
No, matches will not be invalidated for any reason. If this were possible, it could be exploited by a team that is behind to avoid the loss.
What will happen to the old stats?
One important purpose of ranked gameplay is to replace the old stats system with something that is much more fair, meaningful, and healthy for the network. The old stats leaderboard is gone for good, as is K/D on profiles. We have not yet decided what to do with objective stats on profiles. A few non-competitive stats like total playtime are still visible under the stats tab.
Will there be rankings for tourney teams?
Eventually, yes, but that is a considerably more complex project. We started with individual player ranking because it was relatively simple and quick to develop. We will probably want to refine and expand it quite a bit before moving on to teams.
Will more gamemodes be added to ranked servers?
Yes, we will likely add more ranked "genres" in the future, such as deathmatch and blitz. These will probably be on seperate servers and have seperate rankings. We are sticking to objective-based maps for now, just for the sake of simplicity.
How is the ranked map rotation chosen/updated?
In the same way as the unranked servers: map developers choose maps based on their own observations, community feedback, and ratings. They will be refreshed with new maps from time to time.
What is [map I don't like] doing in the ranked rotation!?
The ranked rotation is intended to cover the full spectrum of gameplay on the Overcast Network, and to appeal to all types of players. If a map is in the rotation, it's because lots of players like it, and we think it's a good venue for competition. You are free to choose which maps to build your rank on, so if you don't like a particular map, don't play it. Please don't lecture us about what true competition is and isn't, we've heard it all before.
Can you add more features to help teams communicate in-game?
We have a few interesting ideas kicking around for automated team communication, and now there is a better reason to actually implement them. So yes, we are looking into it.
Can you integrate ranked matches with Mumble?
This is probably not worth the trouble to develop, but we will keep it in mind.
Will you add a party system so I can play ranked matches with my friends?
We don't have any imminent plans for a party system, and if we ever did add one, it would not be enabled on ranked servers. We've taken great care to prevent players from having any influence over the team rosters in ranked matches, and a party system would obviously compromise that.
Why did you have to release ranked servers while I'm on vacation/in outer space/trapped in a pickle jar???