Why do new players start with low gold mmr?

Sspencer··5 votes·7,140 views

New players should not start with with low gold mmr. The average player is below low gold 4 so why would a new player start with higher mmr than an average player who has likely been playing for multiple seasons. Yeah maybe if there is a smurf they will be better than that but the system shouldn't try to cater to smurfs. If a new player were to start with low silver mmr that would likely place them much closer to the rank they truly deserve off the bat. Currently if you win 3/4 of your placement games you will be out of bronze. It would ruin far less games for someone who is genuinely new to the game to start low and climb as they improve rather than start higher and fall down to where they should start ruining games for others in the process. And it really sucks for them as well because they just get flamed for trying their hardest despite the problem being a flawed system.

54 Comments

Saezio4 votes

So, new players need to be placed at 50% of the MMR ladder. Why? Because if they are placed somewhere else, given enough time THAT PLACE WILL BECOME THE MIDDLE OF THE MMR LADDER. It's math. If you want players to be placed lower, we need an entire new matchmaking system and a new "skill value assigning" system.

Sorry English is not my main role ;)

Ambitious Minnow2 votes

That's not math at all. New players, because of their lack of xp and statistical Math, should be the worst of the worst, aha should be at bottom. Why would new players be average players???

Saezio1 votes

I am not saying they ARE average players, I am saying you need to place them at 50% and let their wins/losses place them higher or lower. That's why placements exist, to swiftly move all accounts closer to their real skill. If all players started at iron 4 MMR then according to math, half would end up LOWER than iron 4 MMR and that's not possible. That's why you need to place them at the middle. Because if you are not placing them there, then you are artificially moving the middle point towards the elo you are inserting players.

Does that make sense?

Ambitious Minnow1 votes

Think about what you're writing! No it doesn't make any sense at all! UNLESS EVERYONE IS RESET EVERY SEASON, there's no reason to be placing "new" players from the middle of the system! New players should be the worst potential players barring one place for people who are just below average starting bad, they should be starting from bottom.

Instead with this system, and the "ability" to get placed fast, it encourages people to make many accounts diluting the ability of the system to make any good judgements. This is the point we're at now. The system caters to players who cheat it in every way. Look up players and dodge when it's your turn to get the obvious crazy ones? PROFIT. Want to place higher than your current rank, set up ten accounts take one that gets best placement carries. PROFIT. You've made it to platinum, but it's fun to beat low ranked players? Make a new account and you are rolling in two days flat, PROFIT.

It should take time and success to move up any ladder, not cheap gimmicks like getting 87lp / win on a new account. This is the reason our matchmaking sucks.

Subdue1 votes

[{quoted}](name=Ambitious Minnow,realm=NA,application-id=3ErqAdtq,discussion-id=T3EXvMfE,comment-id=0005000000000000,timestamp=2019-11-03T17:21:41.506+0000)

Think about what you're writing! No it doesn't make any sense at all! UNLESS EVERYONE IS RESET EVERY SEASON, there's no reason to be placing "new" players from the middle of the system! New players should be the worst potential players barring one place for people who are just below average starting bad, they should be starting from bottom.

Instead with this system, and the "ability" to get placed fast, it encourages people to make many accounts diluting the ability of the system to make any good judgements. This is the point we're at now. The system caters to players who cheat it in every way. Look up players and dodge when it's your turn to get the obvious crazy ones? PROFIT. Want to place higher than your current rank, set up ten accounts take one that gets best placement carries. PROFIT. You've made it to platinum, but it's fun to beat low ranked players? Make a new account and you are rolling in two days flat, PROFIT.

It should take time and success to move up any ladder, not cheap gimmicks like getting 87lp / win on a new account. This is the reason our matchmaking sucks.

I think the problem is you've got it backwards. New players have always started in Silver. All the way back in Season 1 and every season afterwards, they started in Silver. Silver is defined by the skill of new players. There aren't new players in Veteran Silver Land; there are veteran players in New Player Silver Land.

Unker1391 votes

[{quoted}](name=Ambitious Minnow,realm=NA,application-id=3ErqAdtq,discussion-id=T3EXvMfE,comment-id=00050000,timestamp=2019-11-03T16:21:02.599+0000)

That's not math at all. New players, because of their lack of xp and statistical Math, should be the worst of the worst, aha should be at bottom. Why would new players be average players???

New ranked players are unknown. They might be bad...they might also have a thousand normal games or it might be a new account for a good player.

A new ranked account is 0-0...think of it as the same as 500-500.

Starting new players at the bottom would be like staring them at 0-500. Its not fair to give them automatic losses just because. The bottom players have proven themselves bad. New players haven't proven anything good or bad.

iMidg3t1 votes

{quoted}

Starting new players at the bottom would be like staring them at 0-500. Its not fair to give them automatic losses just because. The bottom players have proven themselves bad. New players haven't proven anything good or bad.

It's actually fair. If they're better than the bottom players, they will climb. If they're not, they wont.

Ambitious Minnow1 votes

{quoted}

It's actually fair. If they're better than the bottom players, they will climb. If they're not, they wont.

I love, LOVE, when someone goes on for pages about math and statistics, then says this. With as many players as lol has, even if the system was great(which it isn't) luck would benefit some players and room others. It certainly IS possible to get bad rolls of teammates over and over(has to happen to somebody).

Buckeye20021 votes

{quoted}

So, new players need to be placed at 50% of the MMR ladder. Why? Because if they are placed somewhere else, given enough time THAT PLACE WILL BECOME THE MIDDLE OF THE MMR LADDER. It's math. If you want players to be placed lower, we need an entire new matchmaking system and a new "skill value assigning" system.

Sorry English is not my main role ;)

yes please, new players should start at the bottom

Saezio1 votes

You have to put new players in the place you soft-reset to. OR, imo, Riot need to hard reset the entire ranked ladder because really it's been 10 years... it's about time.

Now, if you make that hard reset to the BOTTOM of the MMR ladder, it means that by definition, half the player base will end up LOWER than the bottom of the MMR ladder (negative MMR? lol) you can't have that. And even if you don't hard reset, if you feed the new players to the bottom of the ladder, it means that given enough time, the average MMR of the entire ladder will fall due to new players.

P.S. it also means that every single smurf has to pass through iron bronze silver and gold. Where as atm they only start existing in plat as a somewhat regular occurrence.

Buckeye20021 votes

{quoted}

You have to put new players in the place you soft-reset to. OR, imo, Riot need to hard reset the entire ranked ladder because really it's been 10 years... it's about time.

Now, if you make that hard reset to the BOTTOM of the MMR ladder, it means that by definition, half the player base will end up LOWER than the bottom of the MMR ladder (negative MMR? lol) you can't have that. And even if you don't hard reset, if you feed the new players to the bottom of the ladder, it means that given enough time, the average MMR of the entire ladder will fall due to new players.

P.S. it also means that every single smurf has to pass through iron bronze silver and gold. Where as atm they only start existing in plat as a somewhat regular occurrence.

Actually I agree with a hard reset and for smurfs having to go through Iron, bronze and silver, that is fine...they should be able to cruise through it. It will also stop them from throwing so they can demote into those leagues. As for the negative, that doesn't make sense, stop with the bell curve. Just let people play, if they win they go up, if not they go down, if they are iron 4, that's all they go down to.

Saezio1 votes

even at iron 4 they would have MMR equal to the mid of the ladder.

Smurfs won't have a problem cruising through those tiers, but considering iron 4 is as populated as challenger, you understand how it will be actual elo hell.

And what happens when you introduce someone to the ladder at iron 4 MMR like you want, and they lose because they are essentially playing only with new accounts because the actual iron 4 accounts are next to non-existant? How do you value that loss? They played vs new players with new players in their team. Sooo... how can you evaluate the skill of all 10 players in that game, since you haven't evaluated any of those players previously?

Buckeye20021 votes

{quoted}

even at iron 4 they would have MMR equal to the mid of the ladder.

Smurfs won't have a problem cruising through those tiers, but considering iron 4 is as populated as challenger, you understand how it will be actual elo hell.

And what happens when you introduce someone to the ladder at iron 4 MMR like you want, and they lose because they are essentially playing only with new accounts because the actual iron 4 accounts are next to non-existant? How do you value that loss? They played vs new players with new players in their team. Sooo... how can you evaluate the skill of all 10 players in that game, since you haven't evaluated any of those players previously?

Wait so you're saying new accounts will lose because they are only playing with new accounts? I have no issue with that. that's where they learn to carry a game. If they are any good at all, it won't take long. I just don't want them in my gold and silver games feeding or behind 50 cs by 10 min. It's not about evaluating the skill, if they do well they go to iron 3 and if they win, they go to iron 2, ect, eventually they will stop winning.

Rikght now the ladder does not take performance into account, so It would not be missing anything, otherwise take their kda/cs/tower damage, objective damge ect into account.

Saezio1 votes

Wait so you're saying new accounts will lose because they are only playing with new accounts?

I am saying you can't evaluate an unknown variable correlating it to another unknown variable.

If they are any good at all, it won't take long.

If they are any good at all, they belong at least in silver. and again, you don't know who they are playing against if they are playing vs other new accounts.

And to be completely blunt, put them at 50% of the mmr ladder, silver is piss easy to carry even as a new player so it's not unimaginable they are way better than silver.

Buckeye20021 votes

{quoted}

I am saying you can't evaluate an unknown variable correlating it to another unknown variable.

If they are any good at all, they belong at least in silver. and again, you don't know who they are playing against if they are playing vs other new accounts.

And to be completely blunt, put them at 50% of the mmr ladder, silver is piss easy to carry even as a new player so it's not unimaginable they are way better than silver.

That is not true, I've had newbies, not a single one of them came close to carrying a game, they feed their asses off and make the game an autoloss. This game takes time to understand and rushing into ranked at level 30 after playing a bunch or arams does not make you a good player. I'm gold, low silver I can dumpster, higher silver sometimes I can, but not always and it really sucks having new players on your team. Just put them in the bottom and they will climb out. Who cares about who they are playing against, just win and get that 100 LP to get into promos.

Saezio1 votes

So, you always check all 9 players in your games to see if they are new?

MagicFlyingLlama3 votes

Gold 4 is the 50% mark, supposedly.

It is a bit odd, though. A higher level than 30 should be required, to give the system time to calibrate their ability

Kai Guy1 votes

Normals are not used for ranked. Silver 2 looks to be the 50% mark.

VampiricLlama1 votes

Gold IV is the top 51.709% of players.

https://www.leagueofgraphs.com/rankings/rank-distribution

Also fun fact: there are fewer players in Bronze than there are in Platinum, and fewer players in Iron that there are in Diamond!

Kai Guy1 votes

Today 49.2% of population is iron to silver 1 and G4 is 14% So that's 49.3% to 63.2% So Yes The 50 percentile mark is inside gold 4 now. (but G4 is gonna be a range of players. Min value and Max value.

Its not a Zero sum system so folks going for the season rewards are impacting the ladder. It was different a while ago when I made an example.

Curve Skew is normal when low ranked players rage quit and expected. Heres the graph mentioned when Distribution of players was different and had the 50 percentile marker in silver and an example of Skew. https://imgur.com/a/EF3y0LZ

Riot adds more skew as their division 4 is insanely hard to demote from and their ranks are not locked to MMR.

Anyhow the reason I saw Silver 2 looks to be the starting MMR is that tends to be the match average for the 1st promo game on a fresh account.

Kovorix3 votes

Because you assume the players are average so you start placing them at the average mmr.

Mahoh2 votes

Didn't people used to start at like bronze?

Kai Guy1 votes

Silver for as long as I remember.

FioraWillCarry2 votes

You always started in Gold elo. Even with the introduction of Iron and Grandmaster, this hasn't changed. What may have changed is the actually starting rank but not the starting MMR. Before Iron and Grandmaster, your rank would start somewhere around high Silver but your MMR was Gold. Since the introduction of Iron and Grandmaster, the starting rank is somewhere around Iron or low Bronze but with Gold MMR the same way.

Zeyphel2 votes

For the sake of imbalance.

iMidg3t2 votes

Apparently thats how it should work, although in reality it just makes games imbalanced.

Ambitious Minnow2 votes

Funniest part of all the people here that think spouting off about K values makes them sound smart: People that dig math and science love and always reconsider their values when PROOF that contradicts their claims is staring them in the face.

People in this elo range could post hundreds of matches that were decided simply by having the troll new to ranked on their team, if it weren't for Riot deleting the posts because they "shame a player." Even though they are really about shaming Riot's system.

We all know that proof exists, see it personally, Riot just won't let us discuss it honestly.

Kai Guy1 votes

And for every team that lost to a troll there are 4 players who suffered and 5 who gained. How exactly will that ever force some one to stay in the wrong ranking?

People that dig math and science love and always reconsider their values when PROOF that contradicts

https://en.wikipedia.org/wiki/Law_of_large_numbers ^. Does this work. Yes or No.

Should the Law of large numbers be ignored and say that hey a lot of people whine about MM in a team game cuz its not fun to lose because a bad/troll teammate. That's the argument being made here. Frequency of Complaints = Validity of complaints. Its Pretty Dang Common Must Be A Global Conspiracy

Shame there's no math concepts that deals with shifting that out. That's why everybody knows casinos are fair and a great way to invest your Money and you should buy a lot ticket cuz you win or you don't so the odds are 50/50, right?

Alternately? You guys could listen to my advice of, Please take a moment and study Elo and MMR system theory? https://www.microsoft.com/en-us/research/wp-content/uploads/2007/01/NIPS2006_0688.pdf https://imgur.com/a/lNVMjgd https://en.wikipedia.org/wiki/Elo_rating_system

By all means feel free to show me where the Anecdotal "proof" provided by angry posters is relevant? I gave you the full list of variables Elo used.

Meanwhile when some one says Riot Riggs MMR to make a climb longer, simply lowering K variable LITERALLY does that for the entire ladder with 0 effort and 0 time spent trying to find random players and examine all accounts in the que to attempt to load a match that may or may not even get the result you wanted.

Its a great excuse if your stuck, just like Elo hell is, but looking over the systems used for rankings its pretty clear its just an excuse. No matter how many folks claim otherwise. (World is flat, Man will never fly. Leaches only suck out bad blood)

But hey, don't take my word for it. Go, learn, study. Information is out there for any one who wants to find it. Or get upset at me and others for providing information on the topic and using the terms relevant to it. Be shure to punch your mechanic in the face for being so pretentious tell you that the Service Engine light is appearing as a result of a fault detected in the emission for your vehicle. The Idiot should just use simpler terms for it like "car no vroom vroom good!"

Kai Guy1 votes

The way Elo and MMR ladders work is the starting value is the center of the ladder, more or less. Its not zero sum in MMR so some inflation or deflation can happen but its basically the middle and soft resets are a opportunity to completely re-center things. Its not a 0 to 100 style system where you go bottom to top. Its made so the average level of skill is used to build a bell curve. Ranks are just arbitrary segments of the curve. Riot can change what ever Ranks they want or what portion of the curve is represented, LP will slant to adjust, but the bell curve remains the same and is built off the populations skill. Ranks don't represent skill unless accurate.

Ambitious Minnow1 votes

How does a system where new players start in the middle and not the bottom possibly make sense to you?

Kai Guy2 votes

Bell curves. You need a system that measure skill in a population? You want a rank to have a actual relation to skill and not just a representation of a investment of time? Then its on a curve.

The basics. There are only 2 outcomes in a game. Win or loss. Equal probability. You have 2 games? Game 2 has same possible outcomes. Win or loss, But probability becomes a thing. Because not every combination is equally likely.

1 game. 2 outcomes. W or L. 50% 2 games? WW Wl LW LL. 4 possible combinations, 1 win and 1 loss overlap for WL and LW. So 50% you see a 1W1L Player. 25% for WW 25% LL. 3 games, 8 possible combinations of results. 12.50% 37.50% 37.50% 12.50% 3-0 and 0-3 are far less common then 2-1 and 1-2. 4 games. 16 combinations. WWWW LLLL being 6.25% each of the possible combinations.

So for a "fair" game you would assume 50% odds. Both sides equal ability to win. When that's not true? That's how and why some one sees a raiting change. Your results are not matching what most likely to occur. Your above or below average.

At Low Data? Yes its pretty RNG. However you can't magically know how good some one is with out measuring it, judging it, Rating it. So As you add more and more numbers the curves get more pronounced. This is literally how Casinos make their money, "House always wins" The numbers are set in games to give the House an edge and make it so the most commonly occurring event is them making a profit.

Roulette for example. You have Red or Black that would be a 50/50 Bet if not for house advantage, the table also has 0 or 00 at which point house wins. So instead of 50%, (Reduced 18/36 and you get 1/2 which is 50%) You see It be 18/38 no matter if your betting on Red or Black, Odd or Even, High or Low.

Kind of a drawn out way to say something that's pretty "duh", the things that are most likely to happen are the things that happen most often.

So regardless of what "Rank" it is. The starting value in Elo and MMR will be the majority player, the most common level of skill, the average level of skill. Below average will sink, Above will climb. If you have that math in place and make the "bottom" of the ladder your starting point?

Roughly 50% of all the games players are going to be stuck in that division and there is no title for the gap in skill. Iron 4 would be a majority of players, and the "skill gap" between iron 4 would be the equivalent of everything from current I4 to roughly Silver 2 (MMR is not made to be zero sum.)

It gets more complicated as you go on, not less.

Ambitious Minnow1 votes

You added literally nothing of value to the conversation here, like one of this people that gets all confused when the math is rolled into a word problem.

The starting point for ranked being in the middle of the skill distribution only makes sense if all players are regularly starting from scratch in ranked. But that's not how it works.

The overwhelming majority of "new"players are exactly that, new or generally inexperienced. These players should be placed in the middle of the new player skill curve, they should not be placed in the middle of the general player skill curve. That. Makes. No. Sense.

The exception to this system are the smurfs they obviously do exist. The obvious solution is to handle them specially, not let them break the whole concept of a ranked system.

The obvious solution is to measure the skill of these new players before introducing thru into the general system.

Kai Guy2 votes

The overwhelming majority of "new"players are exactly that, new or generally inexperienced. These players should be placed in the middle of the new player skill curve, they should not be placed in the middle of the general player skill curve. That. Makes. No. Sense.

The quickest way to find if some one belongs in a Range for MM to use which lowers the total # of mismatched games that player is in and causes.

When you change the starting point the ladder? It will redistribute. All the folks who broke even and treaded water because they get more wins from weaker players and newbies lose that source of wins. They then will start to sink. The people who beat them to break even. They start to sink.

Give it time, the ladder then reflects the "newbie" starting point as the center of the curve and hey people are again being put there. The odds favor you to see the worst player in the match be on the enemy team. Pretty dramatically. Remember Roulette with its 18/36 distribution odds? If a game losingly bad player is in a match and its not you, then you get the house advantage.

5 players per team. 4 slots on your team due to you being in 1. 5 on their team. Always will be 4/9 vs 5/9. Aka 0.44 0.55 Thats 56% to 44% odds rounded.

By your own admission you think a majority of these players are free wins.

LLN works. Riots "Ranks" are shiny distractions, they have no meaning outside of what segment of the curve they represent.

They could have a majority of players not earn any rank until you hit bench mark percentiles. Top 30% minimum to earn bronze for example. They can make a majority be in iron or bronze. Its not going to change how a Curve is made and the impact of changing what the starting MMR is.

The obvious solution is to measure the skill of these new players before introducing thru into the general system.

Until they start showing up at higher randomness over the ranked ladder creating longer streaks of chaos and mismatched low quality games.

If it was a tournament then yea it would be easy to have some test or min requirements. Its a 24/7 MM que. Killing population kills que quality. More people is better for tighter ranges. Downside is having to deal with some god awful mismatched high uncertainty games.

If I wanted to ensure a min level of skill then I would make Minigames that teach players basic mechanics that are needed to function Cant beat a arbitrary Minimum value cant que for ranked. Like CS simulator and you struggle to LH 1 per wave? GTFO no one wants an adc doing that. On paper its fine. In practice people wont behave the way I want.

So I get it. I understand mismatches suck.

But... Is it really Riots job to hold everybody's hand and spoon feed them skill and knowledge or is it on the player who wants to get better to improve?

iMidg3t1 votes

[{quoted}](name=Kai Guy,realm=NA,application-id=3ErqAdtq,discussion-id=T3EXvMfE,comment-id=00030000000000000000,timestamp=2019-11-03T18:32:17.847+0000)

Give it time, the ladder then reflects the "newbie" starting point as the center of the curve and hey people are again being put there. The odds favor you to see the worst player in the match be on the enemy team. Pretty dramatically. Remember Roulette with its 18/36 distribution odds? If a game losingly bad player is in a match and its not you, then you get the house advantage.

How is hoping that you get a bad player in the enemy team even remotely competitive and fair?

Kai Guy1 votes

Never said it was? Just that it happens. People act like correct placement is easy with 0 data. That's bullshit. Skill is something that exists even at unknown values but measuring it requires proof. So,Knowing some ones skill is not the starting point. MMR MM needs to locate where some one is going to fit into the curve first before you see them not make games into mismatches. Being wrong and self correcting. That's why its good for something like "skill" that changes over time people can improve their skill at things from their starting point. Especially in a game like Leauge where skill can shift outside of player control. (balance thru patches, reworks, meta shifts, item changes.)

Consistent positive Win trends in your Class interval move you away from it higher. Negative trends sink you down. Anything inside the standard deviation expectations floats there for eternity till they have positive or negative trends.

After 30 games the odds of some one winning all or losing all are basically 0. 1 set in 1,073,741,824 I believe?

If you remove the people who are providing a lot of wins to more players then the create losses for, you sink the ladder. Until the people who need to have those wins start seeing them again just to tread water. Just how its going to go.

I feel its fair to tell people that as a rule of thumb, High uncertainty matches = Low gameplay quality.

"stronger" MM is lowering the time it takes to correctly locate accounts in the curve so you don't have them at high uncertainty. Again, there are only so many possible combinations of W or L. There will be an average and thats why I keep talking about a curve.

That's also why I am constantly point out that gameplay issues are best solved by fixing it by considering gameplay. MM wont care about gameplay experience. Its looking at wins and losses. That's really it, all it needs and takes from a match played.

Fun, Close, feeders,strong team,weak team. It really does not care about that.

Its competitive because it will get people to the correct part of a curve and then you see games are with and vs peers. Your not mismatched and single handedly stomping or mismatched and single handedly losing. You do still see mismatched games, some one might come and stomp or come and flop. That's not you thou, and variance in performance is expected. MMR is a mean raiting not an absolute one. It's a range of expectations not a single whole number, its just an averaged representation of skill.

its fair because it moves people in relation to what the accomplish and it looks to lower mismatched games by design.

Gameplay thou is gameplay. People confuse the issue. Like, If how a majority of games are played is feast or famine resulting in 1 sided stomps. Nothing will fix that in MM.

iMidg3t1 votes

[{quoted}](name=Kai Guy,realm=NA,application-id=3ErqAdtq,discussion-id=T3EXvMfE,comment-id=0003000000000000000000000000,timestamp=2019-11-03T19:20:25.082+0000)

Never said it was? Just that it happens.

The thing is that it shouldnt happen. It just ruins games. No matter what you say to justify it, it's just fucking over actual silver/gold players.

Its competitive because it will get people to the correct part of a curve and then you see games are with and vs peers. Your not mismatched and single handedly stomping or mismatched and single handedly losing. You do still see mismatched games, some one might come and stomp or come and flop. That's not you thou, and variance in performance is expected. MMR is a mean raiting not an absolute one. It's a range of expectations not a single whole number, its just an averaged representation of skill.

In oder for game to be competitve you need a competition on both sides. How is playing against a literal newbie and sweeping the floor with him a competition? Spoiler alert, it's not.

its fair because it moves people in relation to what the accomplish and it looks to lower mismatched games by design.

Yea, if people actually manage to escape from an MMR of a newbie. But thats kinda hard for some of us, since it wont stop until mid/high gold. So losing a game because we were unlucky enough to get a newbie whos barely able to move, let alone play properly, on top of other bs, while having 0 control over it, is not fair.

If you think it's fair then it's the same as thinking having a troll in your team is fair aswell. But at that point you would lose all credibility.

The point is that, if I go in ranked and I'm silver 2, I want to play with actual silvers. I dont want to play with a fresh lvl 30 who got his silver MMR by default. I want players of my skill level which should increase as my elo increases.

Kai Guy1 votes

Skill is not a known variable with zero data.

You can not run a skill ladder or a skill MM system with 0 information on some ones skill. So MMR MM is not just "skill biased matchmaking" its also the system that has to measure out skill.

There would be no need to Rate players if we could instantly know how good they are. Skill is not a constant, its changes, people can improve or get worse. MMR ratings represent a mean performance average.

If you don't want to see "newbies" in your games you only need the ability to climb above the MMR they exist in. If you cant do that despite the fact the player your complaining about will with LLN provide more wins then losses, you lack the ability to play at that level.

The thing is that it shouldn't happen. It just ruins games. No matter what you say to justify it, it's just fucking over actual silver/gold players.

The populations skill curve of Rating wont change. Swaping the titles wont fix it. Its going to be a system built around averages. So, change the starting value, ladder will distribute to match that change. This fucks over ALL players. Those silver/Gold folks are now sinking slowly into the new starting MMR, and now not only do they still get the games your calling unacceptable their rank is also dropped to what ever you set as "starting" Raiting. Iron, bronze. Where ever you put it, that's where a majority of the player base will be playing.

In oder for game to be competitve you need a competition on both sides. How is playing against a literal newbie and sweeping the floor with him a competition? Spoiler alert, it's not.

The have some time to learn the game prior to queing for ranked. If some one thinks they have the ability and goes into ranked lacking that ability? bo hoo. That's on them. You compeate with out the ability to back it up then you pay the price. That's life. No devs are obligated to hold some ones hand and spoon feed them skill.

Again, that's only gonna happen in a fraction of a players games. Better MM lowers the Frequency. I wont make any claims for riots system apart from I do believe there is room for them to improve their MM.

If you do something stupid, run a secondary skill curve, the exact same issue happens.

Its already something you can observe.

Normal are an independent curve. If you don't have the "skill" from normal going into ranked you get demolished.

Same issue, If you make a secondary curve for high uncertainty players for ranked, then when they start to blend into the curve, if they cant cut it, they get demolished. The main difference is that because they spend time in a separate que that gives them inaccurate raitings, when they blend they do so at a wider range of the ladder. Impacting more games, and because you only blend when uncertainty would be lower? they spend more time IN the main population curve fucking up more games.

Its better to run a high K, dead weight will sink from that. People who can stomp will climb.

No attempts to match skill are not competitive. MMR is the best option. It has its flaws, but its proven to work. Just needs the data, and it will have mismatched games from high uncertainty.

Yea, if people actually manage to escape from an MMR of a newbie. But thats kinda hard for some of us, since it wont stop until mid/high gold. So losing a game because we were unlucky enough to get a newbie whos barely able to move, let alone play properly, on top of other bs, while having 0 control over it, is not fair. If you think it's fair then it's the same as thinking having a troll in your team is fair aswell. But at that point you would lose all credibility. The point is that, if I go in ranked and I'm silver 2, I want to play with actual silvers. I dont want to play with a fresh lvl 30 who got his silver MMR by default. I want players of my skill level which should increase as my elo increases.

I agree, A majority of games with a good MM should be exactly what you want. Your an established silver 2 player? Majority of what you play should be with established players and mismatches from uncertainty are not a majority.

I can only relay information as it stands. The curve will adjust to a change of starting MMR value to make that become the majority range. This is how its designed and how it measures player skill. The way the math functions for averages and probability is going to create the same end results with LLN. Same way

Because Its multiple games, not single ones. I do consider having a troll on my team "fair." I think it makes the match complete shit and not fun. I am not an idiot whos going to claim it makes "fair gameplay" But I will have more wins then lose because of trolls. If the enemy team had the ability to get that troll or what BS toxic behavior some one chose to do. That's fair. Randomness is fair in how it treats people. Acting like things that only impact my team are the only things that matter is unfair.

Gameplay is completely irrelevant to Elo systems. 1 or 0. win or loss. Quality or concern for how much fun it was are not tracked or considered variables.

That's why I direct people to get annoyed at Riots behavior systems. To request improvements to ban and remove trolls and toxic accounts ASAP. they suck all the fun out of a game. I don't enjoy losing because of egos. I don't find winning because some one is running it down mid to be fun. This is not something MM will care about.

I know that sounds weird, so please don't blindly trust me or disbelieve me. Go and look it up. Its well documented.

Think for a second why can Elo be used for so many different sports and games and even abstractions like a MM algorithms for literally Matchmaking dating sites like Tinder?

What I am trying to explain, its probability systems not game design. Look into it, you should be able to see if this is the truth behind the reason it can be used to make ratings for songs on sites like pandora, popularity on Facebook, chess, scrabble, football, or curling. Its breakdown of possible outcomes into likely hood of those outcomes then comparison to see if there's a gap between expectations vs reality.

If skill could be accurately judge with out data? that would be nice. The treatment of all players is equal, the behavior of the system is always the same. That's about all the "Fairness" you will get and should expect.

You want to see Truly competitive? Its not gonna be a RNG que in a team game. Its gonna be premade 5 man teams. Proplay is a pro team for that reason.

You want to play a majority of your games vs your peers? Elo and MMR systems do that but are not at all perfect and cant accurate rate people with low data. LLN needs time to shine.

iMidg3t1 votes

[{quoted}](name=Kai Guy,realm=NA,application-id=3ErqAdtq,discussion-id=T3EXvMfE,comment-id=000300000000000000000000000000000001,timestamp=2019-11-04T00:18:23.663+0000)

If you don't want to see "newbies" in your games you only need the ability to climb above the MMR they exist in. If you cant do that despite the fact the player your complaining about will with LLN provide more wins then losses, you lack the ability to play at that level.

And again, thats kinda hard since that shit keeps happening all the way to gold 2 (source: my friend who peaked gold 2 this season was getting fresh lvl 30's while playing solo, and I also kept getting them when I was gold 4) It's kinda acceptable in the elo im currently in (or at least used to be during the old lvling system, runes and IP) but ffs that should not be happening in fucking gold. And btw this is a team game so rly that shit youre saying aint that valid. If youre smurfing then ye, you can climb with ease, but thats because youre WAY above those players skill-wise. Thats not the case for your usual low-elo player. I'm not saying I deserve to climb. If I dont deserve it I dont, thats fine. But at least show me I dont deserve it because I cant beat the opponent of my skill level, not because of some out-of-my-control shit like fresh lvl 30 getting shit on, or a troll, or a dc is fucking me up.

The populations skill curve of Rating wont change. Swaping the titles wont fix it. Its going to be a system built around averages. So, change the starting value, ladder will distribute to match that change. This fucks over ALL players. Those silver/Gold folks are now sinking slowly into the new starting MMR, and now not only do they still get the games your calling unacceptable their rank is also dropped to what ever you set as "starting" Raiting. Iron, bronze. Where ever you put it, that's where a majority of the player base will be playing.

And how do you know it? By reading? Sry bro I only believe in it when I see it.

The have some time to learn the game prior to queing for ranked. If some one thinks they have the ability and goes into ranked lacking that ability? bo hoo. That's on them. You compeate with out the ability to back it up then you pay the price. That's life. No devs are obligated to hold some ones hand and spoon feed them skill.

  1. It's too little time because you dont have to invest any games past lvl 30...no need to grind them to afford champs and runes. Youre set for ranked resource-wise before lvl 30 (will seind ss if you want to see what I mean)
  2. by the time those new players realise they aint ready they will drag so many players down with them.

I agree, A majority of games with a good MM should be exactly what you want. Your an established silver 2 player? Majority of what you play should be with established players and mismatches from uncertainty are not a majority.

True, they're not the majority. But when you pair it up with leavers, trolls, autofills...that becomes a common enough to the point where it impact your games quite a bit, and not in a positive way.

Because Its multiple games, not single ones. I do consider having a troll on my team "fair." I think it makes the match complete shit and not fun. I am not an idiot whos going to claim it makes "fair gameplay" But I will have more wins then lose because of trolls.

But you see, I dont need free wins from trolls. I need quality games. Idc if I lose so much to the point where I fall in Iron, as long as I'm losing simply because the enemy team is better than mine.

If the enemy team had the ability to get that troll or what BS toxic behavior some one chose to do. That's fair. Randomness is fair in how it treats people. Acting like things that only impact my team are the only things that matter is unfair.

If I wanted to play games based in randomness, guess what? I would go and gamble on csgo, at least there I have a chance of getting something with a high RL value.

The treatment of all players is equal, the behavior of the system is always the same. That's about all the "Fairness" you will get and should expect.

In that case it's no suprise that no1 is taking ranked seriously xD

You want to see Truly competitive? Its not gonna be a RNG que in a team game. Its gonna be premade 5 man teams. Proplay is a pro team for that reason.

That's true. But that doesnt mean ranked should be left in this state where you need to hope that each game the enemy team will be fucked over by all kinds of bs more times than you so you can climb.

Kai Guy1 votes

but ffs that should not be happening in fucking gold

Ranks only have significance inside of the system. Riot sets that and can make it represent what % the wish.

And how do you know it? By reading? Sry bro I only believe in it when I see it.

Learning is important for understanding. If you never learned addition, or hypothetically don't even how to read? Then you would not be able to Solve "one plus one equals?" Make the effort to learn about a topic. I get seeing is believing but any one who pretends like the seahawks did not win the superbowl 2014 because they did not watch the game is an idiot. Information is out there. You want sources for elo and MMR I can provide the links so you can see it. Your unwilling to spend time learning and ignore reality? Not my fault or issue.

2,by the time those new players realise they aint ready they will drag so many players down with them.

Do some basic math. We are talking about 1 game by 1 game so with RNG 4 teammates losing and 5 enemys gaining. More players gain then lose. Then as Next game they have different teammates and 5 new enemys gain. More people go up then down, more RNG players gain then lose. Only person sinking every single game a completely outclassed player is in... is the outclassed player. The opposite holds true for over skilled players as they climb.

Not hard. Long as its not you, worst player is on the enemy team more times then your team. If its you, every game its you, every game its on your team.

True, they're not the majority. But when you pair it up with leavers, trolls, autofills...that becomes a common enough to the point where it impact your games quite a bit, and not in a positive way.

How is a non majority not by definition uncommon?

Also, I do have issues with Riots MM. I think mismatches are to fucking common. I don't think its a MM failure, just some of their other systems not doing a good enough job. Duoing for one, not removing toxic players quickly enough for another.

But you see, I dont need free wins from trolls I hate those games as well. Like, Losing and winning because some idiot refused to play competitively sucks. I am just relaying information, stating how things are to the very best of my knowledge after looking into the topic and learning. That does not mean I embrace or endorse it. How it breaks down is pretty simple. If its not you, the weakest or the guy who wants to lose is on the enemy team more games then your team. that's just what happens because your a slot on your team 100% of games and it leaves 4/9 remaining slots to go to your team and 5/9 slots to be on the enemy side. All I am saying, this is how it is.

If I wanted to play games based in randomness, guess what? I would go and gamble on csgo, at least there I have a chance of getting something with a high RL value.

If you don't control your teammates there will be randomness. Shit, even if its a sports team there is some fucking randomness. Humans are god dam inconsistent. A top tier QB can have off games. How it goes. Go play solo player PvE games with no RNG if its an issue for your game aesthetics.

In that case it's no suprise that no1 is taking ranked seriously xD Life in general is unfair. You wanna win then you play for wins. If the "underdog" could never win then nobody would give a shit about competition. Again, The idiots who hard lose games end up on the other team more often then your side. Take it seriously and you have a house advantage. Don't? You deserve every loss you coulda won but fucked around instead of winning.

That's true. But that doesnt mean ranked should be left in this state where you need to hope that each game the enemy team will be fucked over by all kinds of bs more times than you so you can climb.

They will if its not you being the problem. LLN works. My attitude is remove the shit that makes games BS. Assholes who want losses. Restrict # of accounts. Make duoing less of a huge % range of population. Improve gameplay and behavior systems to target and fix the things MM can not reliably impact.

I want a better game, I play it. Improvements only help me have more fun. Ignoring reality is dumb so I refuse to do it.

Gameplay in PVP is Player generated. Make shure players generate good gameplay form behavior.

Riot can improve on things. That's how it goes. Those improvements might be really unpopular, but fk it quality first right?

There will be ruined games. Take steps to lower and reduces that for quality is what I want.

iMidg3t1 votes

[{quoted}](name=Kai Guy,realm=NA,application-id=3ErqAdtq,discussion-id=T3EXvMfE,comment-id=00030000000000000000000000000000000100000000,timestamp=2019-11-06T09:40:37.111+0000)

Learning is important for understanding. If you never learned addition, or hypothetically don't even how to read? Then you would not be able to Solve "one plus one equals?" Make the effort to learn about a topic. I get seeing is believing but any one who pretends like the seahawks did not win the superbowl 2014 because they did not watch the game is an idiot. Information is out there. You want sources for elo and MMR I can provide the links so you can see it. Your unwilling to spend time learning and ignore reality? Not my fault or issue.

If that "source" of your comes from Riot or is confirmed as a legit source by them then yea, drop it and I'll give it a red. If it's not, I rly dont care since for all we know they might use something different for a MM system. And it's not the system that bothers me, but the issues inside of it. Take a look around the Boards, do you see any1 saying ranked is okay this season? Posts like this one should give you an answer. Ask streamers aswell. There is a reason they prefer smurfing in low elo rather than playing in high elo.

Not hard. Long as its not you, worst player is on the enemy team more times then your team. If its you, every game its you, every game its on your team.

And again, thats not making ranked feel better to play. Get that in your skull, not every1 wants cheap wins. Some of us want challenge. Stomping or getting stomed aint it chief.

I hate those games as well. Like, Losing and winning because some idiot refused to play competitively sucks.

If you rly hate those kind of games you wouldnt say shit like "you will climb as long as the enemy team has an idiot in their team, assuming its not you". You just contradicted yourself.

If you don't control your teammates there will be randomness. Shit, even if its a sports team there is some fucking randomness. Humans are god dam inconsistent. A top tier QB can have off games. How it goes. Go play solo player PvE games with no RNG if its an issue for your game aesthetics.

And like I said, I dont mind losing if we just got outskilled fair and square. In those games I can learn something, since despite being evenly-matched they still had something that gave them an edge over us. What can I learn in a game where some1 inted my ass, or was just way below the skill level of other players, aka, new player in a gold game? NOTHING.

Life in general is unfair.

Are you fucking serious??? Pls stop pissing me off with stupid shit like this. If this is the best thing you can type, dont type at all. This is a videogame, not a real life.

I want a better game, I play it. Improvements only help me have more fun. Ignoring reality is dumb so I refuse to do it.

I dont rly see you want it, since you type dogshit things like "its just life lmao". It just seems like you enjoy the cheap climb because of how fucked ranked is.

There will be ruined games. Take steps to lower and reduces that for quality is what I want.

Yea right xDDD If that was rly the case you wouldnt defend stuff like this. But w/e keep being a cheap player. I'm done here.

Buckeye20021 votes

[{quoted}](name=Kai Guy,realm=NA,application-id=3ErqAdtq,discussion-id=T3EXvMfE,comment-id=00030000000000000000000000000000000100000000,timestamp=2019-11-06T09:40:37.111+0000)

Ranks only have significance inside of the system. Riot sets that and can make it represent what % the wish.

Learning is important for understanding. If you never learned addition, or hypothetically don't even how to read? Then you would not be able to Solve "one plus one equals?" Make the effort to learn about a topic. I get seeing is believing but any one who pretends like the seahawks did not win the superbowl 2014 because they did not watch the game is an idiot. Information is out there. You want sources for elo and MMR I can provide the links so you can see it. Your unwilling to spend time learning and ignore reality? Not my fault or issue.

Do some basic math. We are talking about 1 game by 1 game so with RNG 4 teammates losing and 5 enemys gaining. More players gain then lose. Then as Next game they have different teammates and 5 new enemys gain. More people go up then down, more RNG players gain then lose. Only person sinking every single game a completely outclassed player is in... is the outclassed player. The opposite holds true for over skilled players as they climb.

Not hard. Long as its not you, worst player is on the enemy team more times then your team. If its you, every game its you, every game its on your team.

How is a non majority not by definition uncommon?

Also, I do have issues with Riots MM. I think mismatches are to fucking common. I don't think its a MM failure, just some of their other systems not doing a good enough job. Duoing for one, not removing toxic players quickly enough for another. I hate those games as well. Like, Losing and winning because some idiot refused to play competitively sucks. I am just relaying information, stating how things are to the very best of my knowledge after looking into the topic and learning. That does not mean I embrace or endorse it. How it breaks down is pretty simple. If its not you, the weakest or the guy who wants to lose is on the enemy team more games then your team. that's just what happens because your a slot on your team 100% of games and it leaves 4/9 remaining slots to go to your team and 5/9 slots to be on the enemy side. All I am saying, this is how it is.

If you don't control your teammates there will be randomness. Shit, even if its a sports team there is some fucking randomness. Humans are god dam inconsistent. A top tier QB can have off games. How it goes. Go play solo player PvE games with no RNG if its an issue for your game aesthetics. Life in general is unfair. You wanna win then you play for wins. If the "underdog" could never win then nobody would give a shit about competition. Again, The idiots who hard lose games end up on the other team more often then your side. Take it seriously and you have a house advantage. Don't? You deserve every loss you coulda won but fucked around instead of winning.

They will if its not you being the problem. LLN works. My attitude is remove the shit that makes games BS. Assholes who want losses. Restrict # of accounts. Make duoing less of a huge % range of population. Improve gameplay and behavior systems to target and fix the things MM can not reliably impact.

I want a better game, I play it. Improvements only help me have more fun. Ignoring reality is dumb so I refuse to do it.

Gameplay in PVP is Player generated. Make shure players generate good gameplay form behavior.

Riot can improve on things. That's how it goes. Those improvements might be really unpopular, but fk it quality first right?

There will be ruined games. Take steps to lower and reduces that for quality is what I want.

These don't really work. Way too many streaks in this game for me to buy into it's RNG and 50% of the time blah blah blah. And why does matchmaking put 4 golds on one team and 4 silvers on another....they should even those out, but most of the time it's like that.

Kai Guy1 votes

I get it. That's a common view. Let me guess, You call it a streak at 3-4 games? That's also a common. You tell me when you start to count a streak.

But 2 games is technically a streak. Why not use that? Does it feel just a bit low? Don't want to use the absolute minimum value. How much better is 3 at as the second lowest option?

I ask folks to learn the math becuase it matters. If you want here is the breakdown.

To make it easy in the examples I am using 2 as a streak. The combinations of possible results in 2 games are WW,WL,LW,LL. Now 1 win 1 loss as a result is higher odds, as the order does not impact it. 50% see a result of 1w1l. 25% 2w 25% 2l. But 2l and 2w is a streak so 50% streak vs non streak. Elo is probability. Its a curve. 25% 50% 25%. Streaks? 50% 50% they dont match up at all. Its going to get worse not better.

3 games? 2^3= 8. 3w or 3l is 2/8 reduced thats 1/4 = 0.25=25% to see a streak of 3 going eather direction. If we stop here, looks like the % went down right? But I am counting and looking for 2 game streaks not 3 games. That's also WWL.LWW.WLL.LLW. 3w 3l are still streaks and WLW.LWL. are not good. So 2/8. Are not a streak and 6/8 are. 75% you will see a streak. Probability curve is a mix made of 3l 2l1w 2w1l 3w. So. 1/8 3/8 3/8 1/8 =12.5% 37.5% 37.5% 12.5%

4 games 2^4. 16 possible combinations. Only 2 combinations don't have a streak. WLWL and LWLW. So 14/16 = 7/8 = 87.5% odds to find a streak.

At 5 games you hit 32 possible results. And you dont just see 1 streak but can find 2 streaks in 1 set.

More games more you see streaks. You ever consider relevance of how many games?

You want to hear how people fuck up on WR?

You ever stop and ask yourself why you think streaks matter? Who told you they did? WLWLWLWL is a nice pattern.... Its not a common one. 1/256.

Buckeye20021 votes

[{quoted}](name=Kai Guy,realm=NA,application-id=3ErqAdtq,discussion-id=T3EXvMfE,comment-id=0003000000000000000000000000000000010000000000010000,timestamp=2019-11-06T22:40:33.109+0000)

I get it. That's a common view. Let me guess, You call it a streak at 3-4 games? That's also a common. You tell me when you start to count a streak.

But 2 games is technically a streak. Why not use that? Does it feel just a bit low? Don't want to use the absolute minimum value. How much better is 3 at as the second lowest option?

I ask folks to learn the math becuase it matters. If you want here is the breakdown.

To make it easy in the examples I am using 2 as a streak. The combinations of possible results in 2 games are WW,WL,LW,LL. Now 1 win 1 loss as a result is higher odds, as the order does not impact it. 50% see a result of 1w1l. 25% 2w 25% 2l. But 2l and 2w is a streak so 50% streak vs non streak. Elo is probability. Its a curve. 25% 50% 25%. Streaks? 50% 50% they dont match up at all. Its going to get worse not better.

3 games? 2^3= 8. 3w or 3l is 2/8 reduced thats 1/4 = 0.25=25% to see a streak of 3 going eather direction. If we stop here, looks like the % went down right? But I am counting and looking for 2 game streaks not 3 games. That's also WWL.LWW.WLL.LLW. 3w 3l are still streaks and WLW.LWL. are not good. So 2/8. Are not a streak and 6/8 are. 75% you will see a streak. Probability curve is a mix made of 3l 2l1w 2w1l 3w. So. 1/8 3/8 3/8 1/8 =12.5% 37.5% 37.5% 12.5%

4 games 2^4. 16 possible combinations. Only 2 combinations don't have a streak. WLWL and LWLW. So 14/16 = 7/8 = 87.5% odds to find a streak.

At 5 games you hit 32 possible results. And you dont just see 1 streak but can find 2 streaks in 1 set.

More games more you see streaks. You ever consider relevance of how many games?

You want to hear how people fuck up on WR?

You ever stop and ask yourself why you think streaks matter? Who told you they did? WLWLWLWL is a nice pattern.... Its not a common one. 1/256.

No I'm talking like 8 games+ I've been on win streaks of 15 games, and some losing streaks of 10 games, I learned support just because of how bad the bot lane feeds. It's usually squishy supports, you play Leona or naut, very rarely feed, why don't more people do that?

Kai Guy1 votes

Then lets take a look. 8 Game minimum, that's the value you gave me so lets use it over this year for you, Start of season Jan 23. Your history is available for ranked games here. Going back to Season start I found streaks counting at 8+ to be 9W streak Starting on (S)7-23 Ending on (E) 7-22 9l Streak S: 7-31 E: 8-4 9L S: 9-14 E: 9-16 8l S:10-7 E 10-8
8l S:10-27 E: 10-29

So a total of 34 for games total in a streak of 8 or more. Out of W262 and L267
Break down into %. Overall 34/529 so 0.064 Rounded into 6% of your total games. Given loss streaks seem to be the majority, and are also the ones that suck the most to experience seems fair they stand out to you as negatives.

If we used 10 games it would be 0%. If we used 7 as a start instead of 8 there would be a more streaks. 2 would add a fuck ton of streaks and 3 is just the next one up.

I am not seeing 15 or 10 for ranked play this year but I think that was likely you going off memory then you trying to intentionally be deceptive, its a pain in the butt to go over every game in some ones history. I think your values aren't really a big deal of an exaggeration because personally "loss streak"of even 3 feels like an eternity of shitty matches and i am some one who expects seeing that.

I don't know why more people don't play support or at least learn the basics for the role, but im of the opinion you should learn the basics for each and every champion in the game, otherwise your gonna need to learn it on the fly mid game.

So, I tried to ensure the numbers I gave you are in line with reality. Please double check them and alert me if I made any errors.

After that, ( its a hassle but please do that. I want to be checked and held accountable for accuracy) Do you think 6% "in a streak" of your matches qualifies as high frequency?

Buckeye20021 votes

[{quoted}](name=Kai Guy,realm=NA,application-id=3ErqAdtq,discussion-id=T3EXvMfE,comment-id=000300000000000000000000000000000001,timestamp=2019-11-04T00:18:23.663+0000)

Skill is not a known variable with zero data.

You can not run a skill ladder or a skill MM system with 0 information on some ones skill. So MMR MM is not just "skill biased matchmaking" its also the system that has to measure out skill.

There would be no need to Rate players if we could instantly know how good they are. Skill is not a constant, its changes, people can improve or get worse. MMR ratings represent a mean performance average.

If you don't want to see "newbies" in your games you only need the ability to climb above the MMR they exist in. If you cant do that despite the fact the player your complaining about will with LLN provide more wins then losses, you lack the ability to play at that level.

The populations skill curve of Rating wont change. Swaping the titles wont fix it. Its going to be a system built around averages. So, change the starting value, ladder will distribute to match that change. This fucks over ALL players. Those silver/Gold folks are now sinking slowly into the new starting MMR, and now not only do they still get the games your calling unacceptable their rank is also dropped to what ever you set as "starting" Raiting. Iron, bronze. Where ever you put it, that's where a majority of the player base will be playing.

The have some time to learn the game prior to queing for ranked. If some one thinks they have the ability and goes into ranked lacking that ability? bo hoo. That's on them. You compeate with out the ability to back it up then you pay the price. That's life. No devs are obligated to hold some ones hand and spoon feed them skill.

Again, that's only gonna happen in a fraction of a players games. Better MM lowers the Frequency. I wont make any claims for riots system apart from I do believe there is room for them to improve their MM.

If you do something stupid, run a secondary skill curve, the exact same issue happens.

Its already something you can observe.

Normal are an independent curve. If you don't have the "skill" from normal going into ranked you get demolished.

Same issue, If you make a secondary curve for high uncertainty players for ranked, then when they start to blend into the curve, if they cant cut it, they get demolished. The main difference is that because they spend time in a separate que that gives them inaccurate raitings, when they blend they do so at a wider range of the ladder. Impacting more games, and because you only blend when uncertainty would be lower? they spend more time IN the main population curve fucking up more games.

Its better to run a high K, dead weight will sink from that. People who can stomp will climb.

No attempts to match skill are not competitive. MMR is the best option. It has its flaws, but its proven to work. Just needs the data, and it will have mismatched games from high uncertainty. If you think it's fair then it's the same as thinking having a troll in your team is fair aswell. But at that point you would lose all credibility. The point is that, if I go in ranked and I'm silver 2, I want to play with actual silvers. I dont want to play with a fresh lvl 30 who got his silver MMR by default. I want players of my skill level which should increase as my elo increases.

I agree, A majority of games with a good MM should be exactly what you want. Your an established silver 2 player? Majority of what you play should be with established players and mismatches from uncertainty are not a majority.

I can only relay information as it stands. The curve will adjust to a change of starting MMR value to make that become the majority range. This is how its designed and how it measures player skill. The way the math functions for averages and probability is going to create the same end results with LLN. Same way

Because Its multiple games, not single ones. I do consider having a troll on my team "fair." I think it makes the match complete shit and not fun. I am not an idiot whos going to claim it makes "fair gameplay" But I will have more wins then lose because of trolls. If the enemy team had the ability to get that troll or what BS toxic behavior some one chose to do. That's fair. Randomness is fair in how it treats people. Acting like things that only impact my team are the only things that matter is unfair.

Gameplay is completely irrelevant to Elo systems. 1 or 0. win or loss. Quality or concern for how much fun it was are not tracked or considered variables.

That's why I direct people to get annoyed at Riots behavior systems. To request improvements to ban and remove trolls and toxic accounts ASAP. they suck all the fun out of a game. I don't enjoy losing because of egos. I don't find winning because some one is running it down mid to be fun. This is not something MM will care about.

I know that sounds weird, so please don't blindly trust me or disbelieve me. Go and look it up. Its well documented.

Think for a second why can Elo be used for so many different sports and games and even abstractions like a MM algorithms for literally Matchmaking dating sites like Tinder?

What I am trying to explain, its probability systems not game design. Look into it, you should be able to see if this is the truth behind the reason it can be used to make ratings for songs on sites like pandora, popularity on Facebook, chess, scrabble, football, or curling. Its breakdown of possible outcomes into likely hood of those outcomes then comparison to see if there's a gap between expectations vs reality.

If skill could be accurately judge with out data? that would be nice. The treatment of all players is equal, the behavior of the system is always the same. That's about all the "Fairness" you will get and should expect.

You want to see Truly competitive? Its not gonna be a RNG que in a team game. Its gonna be premade 5 man teams. Proplay is a pro team for that reason.

You want to play a majority of your games vs your peers? Elo and MMR systems do that but are not at all perfect and cant accurate rate people with low data. LLN needs time to shine.

So what data is needed to determine skill? I know it's not kda/csing because I can see those numbers of teammates and they blow most of the time.

Kai Guy1 votes

So what data is needed to determine skill?

Elo uses W/L , The # of games, And the context of the rating of who your playing as the core for directional change in the "skill" measurement. MMR adds on an at minimum change of keeping each individual on a teams adjustment specific to their values rather then treat it as 2 collective wholes.

There are basic expectations of Player at X rank has % odds vs Y rank. When that not met then rating changes to reflect it. When there is a consistent trend of failed expectations you see some one climb or sink. More dramatic early on, lowers with # of games.

For Elo adjustment is Rn=Ro+K(W-We) at its most basic representation. Rn is your Rating after a game. Ro was what MM used to make the match. W is win or loss, its a positive or negative value. We is system expectation and its for mismatches. I believe in MMR its independent to the game average cuz MMR is not Zero sum but I do not have full confidence to state that as an established fact or assure you Riot does so. It might only be a factor if the teams are uneven rather then relevant every game. K just multiplies the result of W-We and thus controls how much the Max R you gain/lose will be. Higher K = faster movement. Used for games with Low # of games, typically lowers over # of played games (seems to be the case in Riots system) And can also lower by ladder position.

If you want, I can link you the images of the formulas in more complex unreduced format. Or scroll up I am about to drop a bunch of links to reply to Minnow.

Streaks are not exactly relevant as you would see something like w-l-w look the same as l-w-w both are 2w-1l. Why i kept bring up averages and a curve.

FioraWillCarry1 votes

[{quoted}](name=Ambitious Minnow,realm=NA,application-id=3ErqAdtq,discussion-id=T3EXvMfE,comment-id=0003000000000000,timestamp=2019-11-03T17:55:42.310+0000)

You added literally nothing of value to the conversation here, like one of this people that gets all confused when the math is rolled into a word problem.

His explanation is the best I've ever seen on why the system is designed the way it is. I used to think it's better to start everyone at the bottom until I saw an explanation like this and I actually agree. Starting everyone at the bottom just wouldn't work. I'm not gonna bother to attempt an explanation since the average person lacks the capacity to understand even the simplest explanation of how the system is made to work.

Kai Guy1 votes

If my explanation is the best you have seen... Then you should take a look at http://www.lifewithalacrity.com/2006/01/ranking_systems.html (its better then me but people ignore links pretty dam often.) which I found from Microsoft TrueSkill section on other rating systems. https://www.microsoft.com/en-us/research/project/trueskill-ranking-system/

FioraWillCarry1 votes

Very interesting reads. Good stuff.

More from Gameplay

More by Sspencer