> First of all, let's gather all facts we have on MMR distribution. When ranked matchmaking (RMM) was introduced, Valve released an MMR distribution list based on the hidden, unranked MMR that used to determine who everybody was being matched with.

> The initial distribution looked like this:

> 5% 1100

> 10% 1500

> 25% 2000

> 50% 2250

> 75% 2731

> 90% 3200

> 95% 3900

> 99% 4100

> Source: http://de.dota2.com/2013/12/matchmaking-2/

> So you were better than 50% of the player base at 2250 MMR.

> Now, according to an active Dotabuff employee the distribution of matches looks like this:

> 72.5% Normal

> 15.5% High

> 11.9% Very High

> Source: http://www.reddit.com/r/DotA2/comments/2weu3l/merlini_is_serious_about_this_shit/coqf00c?context=3

> The cut off points for MMR brackets are roughly:

> <3100 Normal

> <3700 High

> >=3700 Very High

> Now let's consider some things: 72.5% of all matches happen at below 3.1k MMR which is to be expected. The 11.9% very high bracket matches seem a little odd given that initially barely 5% of the player base was in that bracket - so is there an actual MMR inflation that increased the % of players in the VH bracket? No, probably not. VH bracket players are still a minority who, however, play a lot of games which is likely the reason they are this good and the reason why the number of matches at +3.7k is so high. Very highly skilled players will have more games per week on average than lower bracket players so that number is inflated.

> Since Valve is very secretive about these things, a lot of stuff said about RMM is speculation but with the vague data at hand, we can absolutely and certainly say that 3k for instance is NOWHERE NEAR the average Dota 2 skill level. In fact, you are likely among the top ¼ or top ⅕ of players (or probably better) in all of Dota if you are at 3100 MMR or higher.

> EDIT: I'm really happy this thread took off like this. I think we can finally bury the myth of 3k being average (or even remotely close to it). Ya'll are quite a bit better than you might've thought.