The way LP gains work are entirely in relation to your hidden MMR in relation to your division, Your hidden MMR gains are affected by win loss and have modifiers on gains directly related to your mmr and teams over all mmr as well as if it made a game it feels does not favor team.
To add a system that helps players with bad teammates on losses you must also create on that punishes bad players on wins, The way MMR works is it takes from the losing players and adds to the winning players so that the ladder is not inflated, some one has to go up and some one has to go down at each win and loss. Theres just not a great way to always check for how much value a player did to a win or loss just off the #s, the 20-4-11 adc could be the guy that lost your team the game by face checking barron bush 45min. The 0-9-12 support could be making hero plays body blocking key ablitys and CC which result in his carrys winning fights for his team. Im shure you can think of more examples.
The biggest issue with MMR is that its a modified ELO system, This requires a pretty dam large sample size of games for it to get accurate when it places players if they do not start near their actual skill. If your MMR system is very basic it would require for example 200-400g games to have a reasonable relation of its assigned player skill to the actually skill of a player. Riots does stuff like Placements and higher MMR variance for lower game accounts to speed this up, but it does create some issues for players where they move up and down ladder to rapidly as they just don't have a high # of games so their MMR fluctuates rapidly. End of the day its going to require players to put in a decent # of matches for the system to have a good grasp on that players skill, until that happens expect a wide range of variance. That being said the advantage of an ELO system like this is that it will over relevant span of games be self correcting and get player to a point they are close to their level of skill when the RNG of grossly mismatched teammates is lowered and balances out from law of large numbers. When that happens ideally a "good player" Would have above average MMR if they got grossly misplaced below their skill and had streaks of unwinnable games, once the odds even out under the assumption they are better then their division by a decent amount they will start to climb and their MMR will grow at a faster rate then their LP gains, this results in a situation where they will get more LP per win then they get per loss.
Mathematically the odds are very much in a "good" players favor, the only issues if that player will have the time to grind out ranked games. The fairness is that you get free wins for when the enemy team has the same bad game losing players that cost you a match, how ever the odds are higher they will be placed on the enemy's team.
A common problem I see is that If you easily tilt, or tilt teammates, or exhibit behavior that makes the game more toxic your shooting yourself in the foot, you load odds that previously favored you into bad odds. Being able to play around teammates is a skillset that will improve your odds for a win, don't let teammates freak out and flame, be civil and keep folks focused on the game rather then a bitchy finger Pointing flame fest.
One solution to this variance is to have the # of games played start putting you into matches with only players with similar # and reduce the amount of low game players on each team. the fact that Duo exists as well as wanting to keep que times reasonable does make this a good deal harder.
so TLDR the system already Favors "Good" Players in their ability to climb to where their skill is at, it just requires you to play enough of games. Mechanical skill alone does not make for a good player.
