Dota Seasonal Rank distribution and Medals

Dota 2 seasonal rank distribution based on the data of millions of players. The MMR by medal is an estimate and is constantly adjusted. Updated to the last season.

I have been asked numerous times about the current rank distribution in Dota 2. We don’t have official and specific info from Valve, but thanks to the work of the community and some tools, it is possible to find a reasonable answer.

The following dataset has the purpose to create a plausible MMR distribution by collecting data from as many players as possible.

A fraction of the player base might not be available as you can’t gather data from Dota 2 profiles that keep their info hidden, but the tool used for this distribution claims that “it covers nearly all ranked players and is not affected by the third-party sharing option” because “it gathers data directly from the Dota 2 system”.

With over 3 million (season 4) 5 million (season 3), 6 million (season 2), and 5 million (season 1) players analyzed, we can consider this distribution very realistic.

I will update the data monthly, so we will have an idea on how the rank distribution evolves over time, and if there is any change.

Quick Links

May 2020

Core

The table is responsive and sortable. If you are using a phone, you may have to turn it sideways for a full view. Otherwise, zoom in or click the grey area to swipe and sort it.

Support

Summary of the data

• According to these statistics, the average player is a Crusader 5, while a Legend 1 is already way above the average hitting the 75 percentile.

• Just a minuscule portion of the player base reached the top ranks, and there is also a huge skill difference between players belonging to the same medal but with a different number of stars.

• Only 10% of the players are spread from Ancient 1 to Immortal, with Divine starting from above the 97 percentile. Many high ranked players often complain about the current state of the matchmaking, but with only 3% of the player base at Divine, and even split in multiple servers, matching together individuals with similar skills isn’t an easy task.

• Only 0.7% of the players have a rank higher than Divine 5! Pro players are a restricted elite group. I am not surprised by their long queue times and that the matchmaking is forced to place them in team with less skilled players.

Learn more: Competitive hero tier lists

April 2020

Core

Support

March 2020

Core

Support

Ranked roles update

Valve updated Ranked Roles matchmaking on March 2nd, 2020. The patch notes confused several players, so I have outlined the key points:

• Your MMR has been consolidated into a single value. (note: the API still shows two separate values, as you have noticed from the graphs)

• When queuing for Ranked Roles matchmaking, you MMR value will be altered depending on the role selected. (Classic Ranked is not affected).

• You can track your current role performance in the Role Queue menu, before searching for a game.

Ranked roles performance – Valve

• The roles marked in green will use your full MMR value. According to the image above, if your rank as a soft support is 4000 (marked in green), you may be placed in a 3500 MMR match when playing as a mid laner (marked in red). (these numbers are just examples)

• Regardless of the role, you will lose or gain 30 MMR in each game.

February 2020

Core

Support

Matchmaking updates

January 16th

• We’ve removed variability in the amount of MMR you gain or lose from a match. This means that there will no longer be matches that give results like +10/-40.

• Instead of matches giving 25 MMR, solo games will now give 30 MMR and party games will now give 20 MMR.

• We have removed the Fast/Slow queue and replaced it with Ranked Roles and Ranked Classic.

  • To play in Ranked Roles you need to earn Role Queue games by playing matches while selecting all roles – this is unchanged from before.

  • Ranked Classic has replaced the Slow Queue, and it does not let you select roles.

February 11th

“We have banned over 40,000 accounts for players who were found abusing matchmaking. These bans will now appear as game bans in Steam as well as being matchmaking bans in Dota 2.”

— Valve

January 2020

Core

Support

Matchmaking updates

December 17th

• Added the Strict Solo Matchmaking option back for fast queue games

• For players with large spreads between their core and support MMRs, there is now a one medal (5 stars) max delta clamp. When the ranks for these players are maximally apart, the two ranks will fall and rise together.

• Increased the variety of party combinations that are valid, to help improve matchmaking quality and queue times in some cases, in part as a result of the strict solo queue addition (for example this means that makeups like 2-2-1 will valid)

December 19th

Role selection has been disabled for very high-rank games, to improve match quality and matchmaking time. 

Right now it’s set for games that include a 7.5k+ MMR player, but we’re tuning that value and watching queue times and match quality to see what the best setting is.” — Jeff Hill – Valve Software Developer

December 2019

Core

Support

November 2019

Core

Support

November 11th Matchmaking update: As a result of the recent changes to legal party combinations, we sometimes have scenarios where three person parties have a lot of trouble finding compatible players.

This is because 3 person parties have recently been disallowed from matching against 5 stacks or solos. Matching against solos in particular had allowed for a much bigger spectrum of compatible matches, but we know that that is undesirable to some players.

So with this update, we are allowing 2s and 3s to on occasion match with a 5 stack when the need is high.

October 2019

Core

Support

Starting from Season 4, the graph and percentile tables in this article will show the rank data split between Core and Support MMR.

From Season 3 to Season 4

• The average rank dropped from Archon 4 to Archon 1.

• The 70 percentile moved down from Legend 3 to Archon 5.

• The percentage of players ranked at Ancient and above diminished from 15 to 11%.

Season 4 matchmaking update

Valve completely overhauled the Dota 2 matchmaking system in Season 4 with a gigantic update in September, and improved it with more drastic changes on October 10th.

Their work is not concluded because – while the new system is working well for the majority of the players – there are still big flaws, especially at the highest ranks. I am certain they will continue gathering data and improve it in the months to come.

October 11th: We are aware of matchmaking queue times taking a very long time and are working on improving it. We may need to slightly relax a few of the new quality settings for the weekend and then we’ll continue tightening things on Monday.

Below, a quick summary of the current situation after the matchmaking updates released so far.

Party matchmaking

• Five-player parties are now matched only against other five-player parties.

• Solo players will be matched against other solo players or at most against one party of two.

• The acceptable MMR spread within a party has been reduced from 2800 to 2000 in order to improve the match quality of both teams.

Party at high MMR

For any party that includes an Immortal player, every player in that party will be considered the same rank as the highest player in that party.

This new rule is necessary to counter the high frequency of matchmaking abuses at high ranks, but on the other hand it will also affect “regular” parties with large MMR disparities, which might impact the match quality.

October 18th: Immortal players will only ever be matched with other Immortal or Divine players in Ranked. For example, that if there is a Divine player in a party with players below Divine, that Divine player will as a result never end up in a match with other Immortal players.

Role symmetry

Misaligned roles have a meaningful impact on game quality. An example of this is when the mid player for one team has the highest rank on their team, and the mid player on the other team has the lowest rank. This tended to cause games to be really volatile and not enjoyable.

It is not fun when a core player is able to stomp your whole team, so starting from this season there will be rank symmetry between players on the same role.

Also, the matchmaker will try to avoid having supports as the highest MMR players in a game, whenever possible.

Bans and punishment

• Players that have a behavior score below 3000 are no longer able to use chat or voice until their score rises above the acceptable threshold.

• Booster and exploiter detection algorithms are constantly updated, and there will be regular ban waves.

  • Lots of users have been banned for violating the Steam Service Agreement that prevents the purchase or sale of Steam accounts, and others because they reached an extraordinarily low behaviour score.

• You can now report players who do not play the role they queued for.

Valve is really trying their best to fight matchmaking abuses, in fact even pro players are punished. In the video below, the famous Jacky “EternaLEnVy” Mao got banned for six months, most probably due to a combination of role stealing and low behaviour score.

“I started at 9k Conduct summary somewhere during July. Then i started to role steal all the way to 1 conduct summary within a span of 2 weeks.

Then after they implemented the stealing role report at start of game I stopped and since then I’ve won 90% of my games playing unconventional heroes and being PMA.

During this time my conduct summary kept going up up but i would always end up in low priority anyway. Finally today after raising my conduct summary from 600 to 1.6K I am banned for 6 months.”

— Jacky “EternaLEnVy” Mao

EternaLEnVy on the six-month ban – Reddit

Smurfs

Smurf accounts are a plague infesting our games as the ruin the match quality at most ranks.

There is not an easy way to deal with them, so the developers are working to improve the detection mechanism and to put them on the right rank area.

We are also changing the mechanism to grant Ranked access. Rather than a fixed number of games, we are changing the metric to be a time based one and increased that amount to 100 hours of playtime.

MMR distribution by Medal – Season 4

Valve rebalanced what MMR range each medal represents, and probably they will continue adjusting it during the season. We are also back to 5 stars per medal as 7 just created confusion among the players, and did not improve the matchmaking quality.

In the past, once a player reached a certain rank medal, they would retain that medal for the rest of the season. Now, if you drop more than one star away from your actual rank, your medal will be adjusted to reflect the new MMR.

One of the best benefits brought by this change is the fight against account buyers: a player can no longer buy an account and keep the high medal for the whole season.

Finally, when winning or losing a match queued as one type of role (Core or Support), you will also gain or lose MMR of the other type by a certain percentage.

Below, the season 4 distribution used by the Dota 2 wiki:

• Herald: 1 – 769

• Guardian: 770- 1539

• Crusader: 1540 – 2309

• Archon: 2310 – 3079

• Legend: 3080 – 3849

• Ancient: 3850 – 4619

• Divine: 4620 – 5420+

All the Dota 2 ranked seasons

I had to split this analysis in several articles because of the excessive amount of tables and graphs, which caused very long loading times.

All the seasons

Stats and seasonal rank graphs: dota.rgp.io – Header image (The Havoc of Dragon Palace for Monkey King): Valve

Source