One difficulty with making even matches with newer players is that every new player who comes in is a complete unknown for the skill rating system. In order to place you in games with players of similar skill, the system needs to first learn your skill, and the only way it can do that is looking at the results of games you play. The general assumption of most Elo-like rating systems, including ours, is that better players win more, less skilled players win less often, and if a player is appropriately rating they will win about half of their games. Each game you play should be with players around your current rating, but if that rating is inaccurate the games could be uneven, and cause either winning or losing streaks for you. As your recorded rating becomes more accurate, this should level off.
We’ve seen some responses in this thread regarding a “Forced 50% win rate,” and we’d like to lay that to rest: The goal of the matchmaker and rating system is not to have every player win 50% of their matches, but rather to bring every player to skill rating where they are expected to win 50% of their matches due to being matched against other players of comparable skill. The matchmaker will not punish a player for winning a majority of their games by intentionally pairing them with other players of lower skill rating. If you are an excellent player and go on a 10 game winning streak, the matchmaker has no obligation to stop you from continuing to win. What will happen is that the game sees you performing at a higher skill level and will try to match you with and against higher skilled players to see if you can continue to win, or start to level off.
You mention wonky compositions, which is one piece of feedback we hear regularly. The Quick Match composition builder is not designed to form the absolute best composition for every player. Some heroes have higher winrates than others, are easier to play, or fit more in the current popular meta; we don’t want our systems to force compositions to always match the current professional composition trends. That means that you will see that wonkiness, especially if you play less popular, stranger heroes like Cho’Gall, Murky, Lost Vikings, or Abathur. I personally enjoy trying to find ways to make those strange compositions work, but if you prefer a bit more predictability in your compositions, I encourage you to look for parties in the New Player chat, on community sites, and among friends; you might also try the Unranked Draft mode, where you can more carefully plan a composition.
Regarding learning opportunities, I’d like to bring your attention to the series of Core Concepts blogs that we’ve recently began to publish. Heroes is a different beast from other MOBAs and puts a much stronger focus on certain aspects of the game which are highlighted in these blogs. (First and foremost, teamwork is paramount!) Much of this information has been available since the game’s inception, but the series aims to collect the major introductory points into a centralized location. To that end, we also acknowledge the feedback that this would be welcome info that’s provided within the game.
is the matchmaker working in the way we want? Personally, I’m not going to attempt to answer such a question. I see places that could potentially improve, and to be perfectly honest I haven’t come away from this with absolute faith in the matchmaker. But at the same time, I’m not sure if such improvements can come without repercussions in the form of queue timer increases or, hell, even just that esoteric “game feel” that developers so frequently chase. What we have here is, after all, somewhere close to the general hope for any matchmaker: an appropriate mitigation of the worst combinations, not their outright erasure.
Yet at the same time, I pursued this project in the hopes that it would spark some discussion. So I’m going to end this with a few juicy statements to get the ball rolling. 1) You’re wrong: that rank whatever player did not actually throw the game for you. Or at least, it wasn’t the fault of their rank. It was more likely the fault of their attitude. Probability suggests you’ve played games with lots of rank whatevers and you haven’t even noticed. 2) You’re right: the matchmaker is sending you into a stomp now and then. Not a whole lot of them. In fact, it is most likely that the majority of stomps you’ve experienced were simply a throw in a game that could have gone either direction. But if you’ve played more than fifty games, probability suggests you’ve experienced a handful of inevitable outcomes.