Lp gain and Lp loss

KakarotSS3··1 votes·893 views

So I'm a plat 4 player who has been climbing the ladder, I have an above 50% winrate yet I lose more than I gain. Just curious if this is a common thing or is it just me in Low elo. I don't think its fair to have it that some people at the same rank and about equal the amount of games played, should have to play more than the other person. I thought Riot was making it so climbing was a tiny bit easier, but its way harder.

2 Comments

Kai Guy1 votes

MMR gains are related to the skill of your enemies. LP gains are slanted to move your title to match the MMR range. If your gains are lower and losses are higher the easy test is to look up your match avrages. If your consistently being placed in matches lower then your current title expect to see lowered gains, regardless of WR.

The opposite is true if your in matches above your current rank.

MMR values are off probability distributions. For example lets use Arpad Elo's system with a Logistical Distribution and the Class interval value will be set to 200 points. With Arpads Arbitrary values when you have 1200 v 1400 is roughly 3 to 1 odds the 1400 player wins. Assuming accuracy then If they play 10 games you expect the 1400 player to win 6-8 of them. This is also true for 1400 V 1600 but the 1600 is now favored. The gap between 1200 V 1600 would be something like 90% odds the 1200 player loses.

Inside the system these values are absolutes. Now, When **incorrect **that's how and why your rating changes. If your beating 1400 and your rated 1200. Your rating improves. If for example you beat 1400 3 out of 4 games consistently, your rating would eventually reflect as 1600. The end goal is to get your rating to represent a proper probability expectation for your skill biased off who your playing and beat/get beat. If you cant break even at your rating your MMR sinks.

For Team games, theres a bit more to it. Your looking not at 1 player on paper playing 1 player over and over, theres not multiple rounds. Instead its about beating the average player at that raiting to build the probibilty expectations. There is a lot more uncertainty and less accuracy so it takes longer to establish the correct expectations on an account. You Track each player as an individual and you don't run the system as a Zero sum system so that you can account for variable K factors per team. K factor is the variable used to represent uncertainty. It impacts how many points you gain or lose per match. Higher value more dramatic Gains/Loss.

The Titles Riot gives are often misleading. They are applied even when your high uncertainty. Demotion is very difficult, Across the bord you find division 4 has a wide range of MMR to lose before they get demoted down a Tier.

Titles are best made by using class intervals and Ranges. Personaly, id want a title to represent my systems Class intervals and what ever % I set for expectations. You do this Via Ranges. Silver 4 for example could represent 1200-1250 points. Gold 4 would be 1400-1450 points.

So you then just slant LP gains to move titles to the direction your MMR value is at to put your title as what is represented by your Raiting. Or... lower MMR then title = negatively slanted gains.

HEALTH FOUNTAIN1 votes

I win 16 lose 17 I'm coming down with you it sucks still managed to get 90 lp though but i would've climed so much faster if it was normal rate of 22 win 15 loss

More from Gameplay