283 replies to this topic

[il1il]

The further players are from a 1.0 the worse their experience of the game is. Ether from being thwarted or unchallenged. In an ideal matchmaking system every one would have 1.0 win loss. However ideal systems are rarely feasible.

I propose that the best way to bring players closer to a 1.0 win/loss is invisibly to sort them by win/loss of the last 50 games with a separation in player pools at 1.0 instead of t1-3 and t5-3
If you are a great player who is receiving unfair losses your rating will drop and you will be sorted into easier games where you can prove it as well as the opposite. This means that the players who would be down at 0.1 would more likely face players at 0.1 and end up with a higher win loss pushing the average win loss closer to the 1.0.

Instead match score gives = or - rating to player who won the game for the team through sacrifice and strategy and = or + rating to a player who hid and sniped till the end and scored lots of damage while they watched their team die. I have observed this in countless games.

This could include a system to keep new players from facing the best in the game by requiring repeated high match score markers to exit a cadet protection. While serving as a stop gap/challenge to rapidly bring people out of the new player system if they are smurfing/seal clubbing. But this system would look like high number of kills in repeated games would push you out of the system and into the win loss only category. While a series of horrible games might place a player (like a returning player) back into that system (temporarily).

Match maker could be sorting even weight into teams and then switching players to keep weight distribution even while balancing the number of players at specific win loss benchmarks. Not totaling their performance, while assessing premade groups at the highest win loss rating of that group for all of the players. Why the highest and not assess them as individuals? Because the highest player will most likely make the largest impact and their win loss will even out as they play games any ways. It will also remove the incentive of grouping as a means to gain performance instead of playing with friends because you want to. and solve the "soup queue" complaint while allowing the group functionality still.

[Nightbird]

il1il, on 30 November 2020 - 07:05 AM, said:

I propose

Other than the 50 matches rollover, that's how the red bars are simulated.

Nightbird, on 26 November 2020 - 12:33 PM, said:

Since we're way past the promised update, here's the 240 games comparison. I missed the snapshot for 240 exactly so I had to use the full season 51 data, which puts us 15 games over.



The accuracy in the fringes are a bit lacking but the main columns in the center are much closer to the prediction than I hoped for. I predicted 44.6% in the center 3 columns, and the result is 44.2%. Even the shape holds true.

If anyone forgot why we're using average games instead of total games, the reason was posted here: https://mwomercs.com...ost__p__6340950

[il1il]

Nightbird, on 01 December 2020 - 08:09 AM, said:

Other than the 50 matches rollover, that's how the red bars are simulated.

I did not mean your simulation but in game replacement of the matchmaker

[Nightbird]

il1il, on 01 December 2020 - 07:16 PM, said:

I did not mean your simulation but in game replacement of the matchmaker

...

You know the simulation was of the MM proposal right?