[Theory] Riot's matchmaking intentionally attempt to put people on "streaks"
Major CaveatI have never (nor will I likely ever) played ranked. These observations are from my non-ranked play on both live and PBE.
None of this is data-driven, but I would be interested in what an analysis of the data would show.
Based on my personal observations (as limited as they are) I feel that the matchmaking system in use by Riot is intentionally tuned to cause win/lose streaks. It may seem counterproductive, but there is logic behind this.
my reasoning is simple: similar to a random chance game (like loot crates/hextech chests/ etc) it causes a high. When you get on a win streak, don't you get a good feeling? It is that exact high that Riot wants you to experience when playing that game. It is what keeps you coming back to the game and striving to feel that feeling again.
Now, on to the counterproductive part: not everyone can be on a win-streak. For a large enough group of people to be on win-streaks, there would need to be a (roughly) equal number of people to be on lose-streaks. You might ask "wouldn't lose-streaks cause the opposite effect and cause a negative reaction?" It isn't simple, but the answer is no.
Think about your most recent lose-streak, or perhaps your worst lose streak. If you are anything like me, when you lose a string of games in a row, you probably go "dammit, I just want to WIN ONCE!" So you play, and play, and play, until finally you get that high of winning again. Once there, you either try to make it into a streak, or you end your gaming session, and end on a high note.
Both win and lose streaks causes a higher level of interaction with the game (aka, you play more). It is in Riot's best interest to get you to play more. If everything were simple matchmaking based entirely on skills alone, there would still be a chance for streaks, but they would be less prevalent**.
If you have had a similar, or different, experience, please do share!
** I know this will be controversial from a statistical standpoint, which is why I am keenly interested in knowing what others have experienced, and I would really like to see what kind of information/patterns we can extract from the data.
