Boards›Gameplay

Ways Ranked could be Improved for next season

InfinitexMortal·6/20/2019, 10:42:31 PM·3 votes·1,217 views

Rework the ranked system to actually reflect a ranked system.

  1. Pair people with ONLY the same rank. No b3s in a silver 3 game. No gold 4s in a plat 2 game. No diamonds in a master game. Ranked needs to be same rank vs same rank. I don't care if someone is "silver 3 skill in b2" so they are "paired against silver 3s in their b2 game". If they are truly silver 3 skill, pair them with and against their rank (b2). if they win games move them up. If they lose games move them down. Why make them have to be better than silver 3s to move up in b2? Makes no since.

  2. No more pairing 1 team with most people on win streaks or higher winrates vs other team with most people on lose streaks or lower winrates. It's a stupid way to force 50% winrate. RANDOMLY match people from the same Rank. Yes this may still occur randomly do to it being randomized, but it should be far more infrequent and no longer feel illegitimate or forced.

  3. Get rid of Promos. They DO NOT belong in an MMR based ranked system. They are an artificial gate to extend play time and slow climbers. If someone hits 100 lp move them up. If they hit 0lp move them down. Rank would be much more fluid and not be gated with "if you don't win 2 out of 3 of these specific games none of your other games mattered".

  4. Stop unnaturally forcing 50% winrates. As i said in 1 and 2 pair people against their same rank and have it be randomly made teams. If they win they win if they lose they lose. Stop pairing people above a 50% winrate with people more likely to lose in order to force them down to 50%.

  5. (Not sure how possible this one is) Make rank be based more on personal performance per role rather than W/L. You are 1 person out of 10 therefore, your rank only reflecting a W/L seems inaccurate to me. So many games I've seen people who are horrible in the game but get to move up in rank because their team carries them, or really good players in lower ranks because they get afkers and people who have no idea how to not feed (or simply cant win promos). Would seem logical that if they were based more around performance the games would be more fairly matched with less games where people are far better or worse than others on their teams.

[sg-ezreal]

1 Comments

Kai Guy6/21/2019, 1:29:50 AM1 votes

Let me start by saying I do appreciate that you took the time to try and find ways to improve the game. Please don't get discouraged, Despite the fact im literaly about to disagree with every thing you said... I hope you keep posting and keep looking for solutions to issues you see.

One needs a basic understanding of Elo systems and the interactions. They are probability theory biased and made to be self correcting. Its a long explanation behind the details on how this is done. If your curious I will do my best to convey the basics. Let me know via a reply if you want that. I don't feel you have the best grasp on how they function given what your suggestions are.

  1. I don't quite know how you think Tiers are made. I know of 2 easy ways. The bad one would be simply off player populations. For me i feel this fails because it actively looking to keep players in ranges near the middle and its influenced strongly by player population size. Active and inactive accounts make wacky impacts. Does not do a great job of representing skill because of that. The better way is to have each tier represent static and set range of MMR values with the basic assumption that only when that value is accurate does the Tier represent a players skill. When accurately placed it does not matter how many players are around because each correctly placed player will meet the system probability expectations regardless of populations.

Tiers themselves are not magical good indicators of skill, and because they are wider ranges they are also less accurate then using the specific MMR value on an account. It can get pretty ugly in terms of MM. because the gap between exact values between all players can get wide and the system thinks nah this is fine. Inaccurately placed accounts are not great to experience in MMR but they become even worse if only Tier is used by MM. A minor issue is pools of players. The bigger problem is that a range of numbers where each individual number is already expected to represent a range of results is higher uncertainty so you just end up with higher chaos and uncertainty in each and every match. And that's if all 10 accounts are accurately placed which.. they wont be always.

  1. Far to often people get upset by streaks and win %. Team A op team B was trash, RITO STOP RIGGING GAMES! Then unironically the solution players ask for is for MM to balance teams more equally. I find it entertaining that they are unaware of the impact, If some one has an 80% win rate and you want the MM to balance this... wouldn't that force it to intentionally sabotage that player by trying to find enough players with sub 50% win rates to drag it down? If some one was at 5%... wouldn't they be gifted amazing teammates? How would a player even achieve these dramatic % in the first place if the systems looking to rigg teams?
    But... people ask for it. They want rigged teams to be balanced to 50% overall. or the ask each team to be similar win%... unaware this interferes with Elo systems from moving players up and down the ladder.

The entire point of a competitive skill biased ladder is to sort players by weak, normal, strong. Why should your system sort players in a single Tier by weak, ok,strong. Your taking the entire concept of a ranked ladder then applying it a second time for each individual tier for what? It takes away the value of a Tier or rating in the first place. there's no good reason to have 1200 point players be sorted into. weak, ok, strong. Instead let it run and adjust so that weak players are 1000, average are 1200, and strong would be 1400. Yea a strong will beat weak hella badly.. that's kinda why and how they move up and why and how weak sinks. Let it run fairly.

I don't get why players go on about Streaks as much as they do. Its an arbitrary value players pick to have meaning that most the time does not have any relation to MMR. https://www.random.org/coins/?num=100&cur=60-usd.0010c Here is what 100 50% odds looks like RNG. Flip some coins. When I counted how many coins are part of a 3 or higher game streak win or loss I had 67% of the coins I go as part of a streak. If I said that 10 games made a streak I had 0. So... 3 games is pretty common even at a "fair" occurrence. A win streak should be even more common to see if your better then your current MMR.

Why do people think its rigged to lose because of bad teammates? Like... Duh? You think your going to have a loss because you had great players on your team? All 5 of us played amazing and CRUSHED the enemy team... then we lost? Or is it more likely that bad teammates failed and preformed abysmally which resulted in a loss outside of your control? Just feels like the logic people use when the complain about how something missing is "always in the last place you check" Really Freddy? your whining its the LAST place you looked? Gee... how about next time you misplace your car keys. After you find them. Why don't you spend an additional 10 min checking other places you might have left them. You stop looking when you see them its that easy. A win streak is ended once its made by any loss made arbitrary value players picked and ends most often due to bad teammates. This is just a logical assumption... not that it was rigged to end your streak. If I used 10 I had 0 broken streaks

Also MMR is a balance so why do you care what order wins and losses come in? Example. 1000 points. 10 per game added or taken away. 10 games. 6 wins 4 losses. No matter how I place them the end result is the same, I have 1020 points. can be WWWWWWLLLL or WLWWLWWLWWLWWW. Most the folks on the boards complaining about streaks have a negative # of games... And yea I get it. K factors make it harder to understand for people who don't understand what a K factor is and that it applies weight to early games.

  1. Pretty much the only positive for players outside of perception, not seeing their accounts sink below their starting MMR. Is that LP enables riot to theoretically have more aggressive MMR changes in the starting MMR. This lowers the # of games required to move Newbies and smurfs out of regular players games. Promos are a good way to counteract the impact of early luck when players have a high K factor. Personally I woulda just had tiers unlock only on an account after it has a minimum required # of games. If you don't put in enough points of data for my system to be able to be reasonably confidant it has you in the right MMR value or a close range to it then you don't get a Title, Not even silver or bronze. the problem is... players would not like this system much imo. 1. Accusations of making it "to grindy" (why id be a bad dev and get fired super fast. id tell folks to fk off I don't own a crystal ball and cant magically predict skill with 0 data.) I assume there's also a chance folks treat to many games casually prior to unlocking a title because they think its just extended provisional matches. So promotional are a necessary evil if you want to use LP for any Actual MMR related quality changes... I have no clue if riot is or is not doing so. Take it with a grain of salt as a lot of its just my opinion.

  2. Again... No good way to show its doing this. If its RNG then folks will have variance in win% and you see all kinds of mismatches between teammates and enemy's. Most players only check it when they lose a game... and usually if they had been on a "streak". 1) has issues and 2) is most commonly going to result in players accidently demanding riot Riggs games to deal with their perception games are rigged. Because they overvalue 3 games in systems that usually need 100+. 3% is nothing significant given the amount of RNG in the system.

  3. has issues if its not tied to win or loss. It has issues if it IS tied to win or loss. The problem is... if a game of skill means skill impacts your ability to win. Is there any metric better then winners and losers? if you play 100 games. Lose 100 games. But had great damage and KDA... do you deserve to climb? Considering that its very easy to pad stats in a moba. Karthus who spams R every time its off cd and he has 5 targets generates more damage then a karthus who casts it only to help fights or secure kills. CS can be stolen from teammates who need it to pad your income. KDA can be maintained by playing excessively cautiously and abandoning any team fight that looks like a risk.

If the nexus does not blow up... What else matters exactly? If I need to body block Blitzcrank Q for my adc or mid. and it costs me my life but wins us a team fight. Was that a bad play despite the fact I generated negative stats? How do you make a subjective system automated that it can account for padding or good plays with negative values?

Even if its got win and loss as the core and adheres to Elo principles, is there any reason to assume players wont pad to protect ratings any time they feel they have bad teammates? If you think its already a loss but know you can save some LP/MMR by just padding till game ends... well I don't trust players to not do exactly that.

Thank you for your time if you go over this wall. Please use box quotes for anything I said you want clarity on or simply object to. I hope you keep looking for solutions.