11 replies to this topic

[VirtualRiot]

Why not use a league system? Here is how I think it could work;

All players Win/Loss Ratio is calculated. Everyone is listed from lowest WL/r to highest WL/r.

There are 5 leagues, with 20% of the entire player base in each league. Just like other games... Bronze, Silver, Gold, Platinum, Diamond.

You are not placed in a league until you complete a certain number of placement matches. How this would work could be argued, but my idea would be 10 placement matches. You play 2 matches with every tier, in a randomly assigned order. Then your WLR is calculated and you are placed in a tier. I'm not saying 10 matches is good enough, but you get the general idea.

How is this system not infinity better than this ****** "PSR" system we have in place now where everyone and their grandma is in T1 and it doesn't matter?

Maybe this is all a moot point anyways, the player base might not even be large enough to match only people within relevant leagues together.

But something has to change.

[El Bandito]

Depends. If PGI is forced to let players play 2 ranks above/below them (like current PSR) due to population issues, then MM will still suck.

[Zergling]

As El Bandito said, the low population would force players in one rank/league/tier to fight with/against players of other tiers.

And even in the top 20% of the playerbase, there would still be a huge variation in skill levels. To get players of similar skill levels would require the top ranks to be 5% and 1%, and there obviously isn't nearly enough population for that.

[kyfire]

Ah another thread with a MM based off W/Lr.
What's up with everyone being so hooked on a useless metric as W/L? Yes, I said useless!! Why is it useless you ask? It's quite simple, you drop into a match, you die in the first 30 seconds of the match. The rest of the team goes on and wins the match.
Pay attention now, cause here comes the useless part. You get credited for a win. Did you win? No you were dead after 30 seconds! How can you win if you're dead?

[Mystere]

Are you asking PGI to waste more effort, time, money, and other very limited resources on another matchmaker?

**** it! Just go:

[Ted Wayz]

Just use leaderboard by chassis. Match people by their top 10 scores by chassis and viola mucho variables no longer matter including chassis imbalance.

[RAM]

kyfire, on 19 April 2017 - 07:24 PM, said:

Ah another thread with a MM based off W/Lr.
What's up with everyone being so hooked on a useless metric as W/L? Yes, I said useless!! Why is it useless you ask? It's quite simple, you drop into a match, you die in the first 30 seconds of the match. The rest of the team goes on and wins the match.
Pay attention now, cause here comes the useless part. You get credited for a win. Did you win? No you were dead after 30 seconds! How can you win if you're dead?

Mathematics: clearly not everyone can get it


RAM
ELH

[PhoenixFire55]

Anything is better than PSR, but then again, each "phase" PGI "improved" the MM and each time it made matters worse. Sorry if I'm not holding my breath for yet another "phase".

[El Bandito]

kyfire, on 19 April 2017 - 07:24 PM, said:

Ah another thread with a MM based off W/Lr.
What's up with everyone being so hooked on a useless metric as W/L? Yes, I said useless!! Why is it useless you ask? It's quite simple, you drop into a match, you die in the first 30 seconds of the match. The rest of the team goes on and wins the match.
Pay attention now, cause here comes the useless part. You get credited for a win. Did you win? No you were dead after 30 seconds! How can you win if you're dead?

Good players will have better than 1:1 WLR. It is all about consistency in the long run.

[Skipmagnet]

It's not just a +/-2 bandito. Tier 1 can play with Tier 4. Worse yet, a Tier 1 comp-level player with more than 8k games, can be in the same match, on the opposite side, from a Tier 4 player with less than 20 games total.

That ain't right. PGI don't care, t4 dude isn't a paying customer. Tier 1 don't care, t4 dude is padding for his epeen. T4 dude? We'll never know because he quit before he got to 25 games because no one should have to face the best players in the game while they are still learning to play, and anyone that thinks that is okay is just an *******.

[VirtualRiot]

eyeballs, on 20 April 2017 - 03:10 AM, said:

It's not just a +/-2 bandito. Tier 1 can play with Tier 4. Worse yet, a Tier 1 comp-level player with more than 8k games, can be in the same match, on the opposite side, from a Tier 4 player with less than 20 games total.

That ain't right. PGI don't care, t4 dude isn't a paying customer. Tier 1 don't care, t4 dude is padding for his epeen. T4 dude? We'll never know because he quit before he got to 25 games because no one should have to face the best players in the game while they are still learning to play, and anyone that thinks that is okay is just an *******.

I agree, the variation in skill levels I see in the average match is just too wide.
I want to do well in matches, I want to give each match my all. But my all is way to much for new players to handle by far, its not fair to them. They need to stick around so they can get better and actually pose a threat.
If they quit after their first 10 games because they just keep getting mutilated every match, the cycle can never start.
Making a matchmaker that would work if there was a proper population is the first step.
Then PGI can focus on attracting new players. Because then more of them will stay.
You wouldn't put bait in a trap before you fixed the holes in the cage.

[Mawai]

VirtualRiot, on 19 April 2017 - 04:44 PM, said:

Why not use a league system? Here is how I think it could work;

All players Win/Loss Ratio is calculated. Everyone is listed from lowest WL/r to highest WL/r.

There are 5 leagues, with 20% of the entire player base in each league. Just like other games... Bronze, Silver, Gold, Platinum, Diamond.

You are not placed in a league until you complete a certain number of placement matches. How this would work could be argued, but my idea would be 10 placement matches. You play 2 matches with every tier, in a randomly assigned order. Then your WLR is calculated and you are placed in a tier. I'm not saying 10 matches is good enough, but you get the general idea.

How is this system not infinity better than this ****** "PSR" system we have in place now where everyone and their grandma is in T1 and it doesn't matter?

Maybe this is all a moot point anyways, the player base might not even be large enough to match only people within relevant leagues together.

But something has to change.

You realize that this is essentially a very simplified Elo system right? Elo is slightly more complicated since it takes the rating of each pilot, combines that into a team rating, compares the team ratings to determine which side would be predicted to win, takes the actual match outcome compared to the expectation and adjusts the individual player ratings based on the actual outcome compared to the expected one. It isn't as simple as just W/L ... since your rating goes up more when you win a game where you are the underdog.

The biggest problem with their previous Elo system (which so many people apparently disliked that we had to get the (actually much worse) PSR system) in my opinion was that it did not include a factor for mech choices and mech skills.

How can the matchmaker correctly estimate the likelihood of one team winning when it has no factor for whether one player is in a Kodiak-3 and another is in a Victor (using current mech balance as examples)? How can it predict an outcome when one player is in a fully mastered mech and another is just starting a new mech? The answer is that it can't and as a result the Elo changes would not be very consistent. Under the old Elo system, folks dropping in Kodiak-3s would end up with a higher Elo long term, not because they were inherently better players but because of mech choice.

However, no matter what rating system is used, game population will always be an issue for folks at the positive and negative extremes of the distribution .. these folks will frequently have to play against others with very different ratings.

Anyway, in my opinion again, PGI could do a lot better than currently ... if they don't have the resources in house then I am sure that they could find folks in the community willing to code up an improved matchmaker under contract and NDA for a modest sum.