Content of the article: "Let me tell you, why you really hate stacks!"
Once in a while we are getting this "omg I hate stacks" posts. Backed up with bazillion "oh right, me too" responses. And then comes old mantra "that's not the way game was meant to be played" and "only solo for ranked" ideas.
What most people DON'T understand is, that problem isn't in 3/4/5-stacks, problem is in how MMR and matchmaking works. Let's take a closer look here:
1) MMR of given account/mode is actually very inaccurate metrics. If you think about it, the same number describes how good you were in games with random people, with friends, on tank, on damage dealer, on healer, on each map.. it is your game history averaged to single value.
2) matchmaker DOESN'T know if you are going to play tank, heal, dmg dealer.. your preferred roles are not a part of matchmaking process (they are just visual indicator for your teammates). All matchmaker knows is that you are XYZ MMR player.
So yeah, when matchmaker is composing team of random people, you get people within some MMR range, sure. But that's all that is sure – matchmaker won't guarantee you get player who is main tank, or main healer, or actually does understand given map dynamics. Then comes draft – you have two sides with (more or less) equal MMR average. But what if player X has that 4000 MMR from playing mostly tanks, but (just to fill) has to play damage dealer? Or main support fills tank role? For 99% of community it means they won't perform on such off-role as well as on their main. So, your team isn't anymore worth MMR they were composed around!
Now, let's go deeper into MMR itself. You may find it shocking, but it isn't very accurate for 80% of playerbase. Where it is accurate are two groups – "GM potential/level" players and "potato" players. Players who are that good they can solo carry even a team of potatos, and on the other side people that are so bad (or trolling) that they can single-handed ruin game after game. For all other cases, with each match, MMR gets a bit of false (positive or negative) data. It results from two wrong vectors:
player was bad, but got carried by their team. Such player gets MMR increase where they shouldn't!
player was good, but got burned by their team. Such player gets MMR decrease where they shouldn't!
You may think that it will average itself. No, it won't. If you ever studied system stability topics, it is obvious – it is constant (for 80% of community) computation error introduced in feedback loop. One error affects composition of next team (so affects next 9 people) – because they get underrated/overrated player.
But hey, I was supposed to talk about stacks, and I am talking about MMR/matchmaker? Yes. I needed this part, to show you how random matchmaker (so, something you get in soloQ) works. It simply has very very high chance to be wrong when composing teams.
Now, about stacks. What changes if people play in n-person group? First of all, they are eliminating that wrong matchmaking. They decrease/remove (depending on group size) chance of getting "overrated" person. They decrease/remove risk of people forced to play off-class. And since they are playing together, their after game MMR increase/decrease is actually more proportional to their actual performance (less errors of false MMR increase/decrease I've described above).
What I am saying is – it isn't stacks that ruin your games. It is how bad matchmaking/MMR works for random teams that ruins your games. Get good five randoms who understand game, who will play their main roles and who can communicate – and you have good chance to win against "stack" (look disclaimer at the end of the post!). But get people who fail to communicate and/or who are forced to play off-class and/or have their MMR overrated – and yes, "stack" will win. Not just because it is stack, but because your teammates are actually worse then they should be.
Knowing that, you should understand why "let's remove stacks from SL" isn't the right solution. It wouldn't introduce "fair games" – because there is nothing fair in having bad random compositions on both sides – team that randomly gets more bad players is still in bad spot.
So, what's the right solution? It would require rebuilding matchmaking.. which, at this point, probably won't happen. It is easy to say "fix matchmaking", but it is very complex problem. Yet, I would have few suggestions, or directions, we could go with:
let matchmaker consider our preferred roles when composing teams. That decreases chance someone will play outside of their main class(es).
let's end with single MMR value to describe all roles. Minimal change would be to track MMR per role (so separate for tank, separate for damage..).
let's try to remove/decrease false MMR input after the game. We can calculate how well player performed during the game (in case you doubt it – HotsLogs did it years ago, HerosProfile is doing it now). That means we can adjust MMR decrease/increase after the game based on how well/bad player performed.
when stack enters matchmaking, use their HIGHEST mmr as stack MMR. That will hurt random/friends stacks a bit (but will correct/average itself quite fast), but in return it will remove boosting stacks
disclaimer: stacks CAN be a problem on GM level where people playing (semi)competitive teams are also playing in SL as stack – because there is a huge gap between (semi)competitive player and just master/GM SL player. But for rest of community (and that is above 90% of population) that isn't the issue.
- I’m a new player to Paladins, and here is my feedback and rants I guess.
- An argument against the removal of SBMM. Tonight I played 25 games with my clan as a 6 person fireteam, we won all 25 games and this doesn’t make sense as we look further at the data.
- Current matchmaking is forcing extremely bad games after a series of good games and vice versa
© Post "Let me tell you, why you really hate stacks!" for game Heroes of the Storm.
Top 7 NEW Games of June 2020
Quite a few exciting games are releasing for PC, PS4, Xbox One, and Nintendo in June. Here's what to keep an eye on.
Top 10 NEW Open World Games of 2020
Video games with open worlds continue to roll out in 2020 on PC, PS4, Xbox One, Nintendo Switch, and beyond. Here are some to look forward to!
Top 10 Best New Upcoming Games 2020-2021
The best selection of games which will be released in 2020 and 2021 for PS4, PS5, Xbox One, Xbox Series X, Google Stadia and PC - and you can watch in amazing UHD 4K and 60FPS with latest updates about all of the games in this list!