What actually determines your elo?

{quoted}

My Normals W/L is 73 over. Does that mean I'm masters?

Kinda depends on the amount of games played. The algorithm moves accounts up and down the MMR ladder more slowly the more lifetime games they have. If you've only got 500 total normal matches played? Yeah, you'd have about 3000 normals MMR. If you have 5000 normal matches played? More like 1850-2100 normals MMR.

{quoted}

It isn't ELO but Elo. It's the name of a guy not an acronym.

Quite right, a system originally designed for chess.

And if you are familiar with it, you know that LoL's MMR has very little in common with the real Elo system. Not just things like it's for 1v1 competition and when they use it for teams they use it for the team as a whole and never for individual players, simply because it is unable to do that job

Pretty spot on so far. Elo systems are often used as the framework for competitive games. For team games they cannot, and will not ever, function properly. The very use of one for League is a failure on Riot's part to understand the system.

but more importantly, in the Elo system you only gain or lose rating when you beat a higher Elo or get beaten by a lower Elo player. Otherwise your Elo stays the same as that was the expected result. But when you win/lose that match, your Elo can change significantly right away. As the system is designed to place you accurately as quick as possible. Though, after you got your proper rating, it is very hard to move anywhere because you have to really play better then before.

Mostly correct. Although you DO actually gain and lose rating points for any match played. Its just that playing players of a relatively similar range nets very low yields, particularly at the top end of the spectrum where it might be a 2500 player versus a 2520 player and both lose or gain only 3 or 4 rating points upon match completion. Large swings in score are usually only possible when opponents are of widely varying skills and the result is an upset, for example a 1750 player beating a 2300 player would probably mean a change of 50-70 rating points for either player.

LoL's system is very different as it's designed to make you grind a damn lot to change your rank but it's designed to always give you some room to advance.

That's actually just the veneer or cover sheet for the MMR system. And that system (the League System) can be ignored in its entirety. Choosing to acknowledge its existence at all can only hurt you as a player since that system is responsible for nothing except displaying your rank in non numerical terms and the two systems do not align properly due to the "room given to advance/decline" factor.

Of course, with a proper Elo system people wouldn't play LoL day and night as it would turn out in a month that you are S2 or G3 or whatever and you couldn't move up without seriously improve your gameplay in many ways.

In essence that is how it actually works. Most players who grind up the ladder and play more than 200 ranked games on an account are actually seeing an INFLATED displayed ranking while they maintain a deflated or tanked MMR. And Riot continues to make the system MORE forgiving for some reason.

And if they use LoL's system in chess, nobody would ever get any rating as they simply don't play enough matches a whole damn year :)

HA! Hadn't thought of that. Pretty funny.

Other interesting difference about the systems is the curve of ratings. In Elo most of the players are in the middle ground with just a very few at the bottom anmd at the top. In LoL it would mean that 80%+ of the playerbase are around G4-G2. It wouldn't be good for matchmaking at all. You can check the FIDE curve if you fancy: http://2.bp.blogspot.com/-edJfz2u1XoM/VUNV3aNwU0I/AAAAAAAAAFE/HMtF0Dk0-ZA/s1600/men_gauss.png In LoL most of the players are bronze, a bit less silver, and it goes up like this. You can't have a piramid-like system like this without forcing it, so some manipulation (I don't mean cheating).

And that there in is the major problem with ranked for the last 6 seasons. A point I've made repeatedly.

Btw, some math-guys did a detailed analysis and simulation about how MOBA ranking systems work, like LoL, DotA, etc. there's even a video about it somewhere, and it turned out that in theory it indeed works, but you need about 800-1000 matches to start to get roughly accurate results.

Yeah, they did all that based on my research.

=P

I figured all that out and posted it to the forums 4 some odd years back. Lol.

So, in practice, if you aren't way way better or way worse than everybody in your matches, the guy above was pretty spot on that it is mostly about luck.

And full circle here we come, luck actually matters folks. Managing it can be important. Yes, you CAN manage your luck.