why do i get placed in unfair matches when i am on a winning streak

GHOST1331·6/20/2019, 12:28:12 AM·1 votes·1,069 views

so i start my day off good winning three games, but i notice something weird in the third game. game one and two i gain 23 then 22 lp resp. then game three it goes down to 20. usually the average tier of my games is silver 3 ( slv2 slv3 slv4) but fourth game i notice in the loading screen that my team has two bronze 2 people from bronze one and one slv 4 and two slv 3 me included( one of the brnz 1 is on a huge loose strack feeding like 5 games). we got stomped and i check the enemy team and they are three slv3 one slv 2 and one slv 4.

so to recap a team with average tier slv4 at best against slv 3. how is this fair. why do they do that to me like that is not fair for me so why did this happen to me??? plus i lost 19 lp for that game so my mmr isnt even high like how can i climb like this???

3 Comments

Kai Guy6/20/2019, 3:02:27 AM3 votes

Its pretty simple. You overvalue 3 games as a streak. The time your most likely to lose is when you have mismatched teammates. Any game you have weak teammates your more likely to lose. The assumption the system has it out for you.. but randomly does not for other players is kinda… bad imo. Also you counted some of that other guys Flex loses as his solo losses. only a 3 game "streak".

There is nothing weird about it. 3 games is just a terrible number to use for a Elo or MMR system. https://www.random.org/coins/?num=128&cur=60-usd.0100c-anthony I took the 128 as the # of flips to represent how many ranked games you played and put them into a string of Coinflips. Go ahead and look at how many times you see "streaks" Last time I did this the result was over 50% of all coins where part of a "streak" And this is at "fair" 50% odds.

I think 7 is an ok number to count as a streak thou personal I use 10. 2^3 is 8. Only 1 result is 3 heads. 2^7 is 128 and 2^10 is 1024 to 1. But...that's only for out of the streak. out of 3 games out of 7 and out of 10. The total # of games impacts how many times you see combinations which drops those odds down. Its grows exponentially for some number values that require a massive top end number to just generate the results once. Which is what i am trying to show with my x to one odds, out of only 3 flips there is 8 possible outcomes only 1 is 3 heads. Now. Lets take a look at 2^4. 16 to 1 How many times could I see at least 3 heads? HHHH THHH HHHT. Of the 16 outcomes 3 of them are a streak.

I am not the best with this kind of math but it should be more or less accurate for you to get a basic idea here.

Also streaks don't matter. Consistency does. Its about overall win %. The order is meangless to your MMR. Its the sum of all games + various modifications like K factor and We. I don't want to have to explain those unless i need to because the basic concepts pretty easy to show.

Wins = adds mmr. Losses = Subtracts mmr. so WWWLL is the same as WLWLW or LWWWL. Your still getting the value of 1 W worth of MMR regardless of the order they show up in.

Now the actual MMR per match might not be the same But if you just reorder a game after its played the MMR from that game is going to be the same value.