78 replies to this topic

[KharnZor]

Mischief stop bagging a mech you cant play. Its tiresome.

My W/L has been largely the same for the past few months so its hard for me to comment on the state of MM.

[cleghorn6]

Ok, lets clear this up. If Elo matchmaking was working PERFECTLY, no flaws, no problems, exactly as specified, everyone would have a 1.0 W/L. Everyone would be losing exactly the same number of matches as they win. This is what Elo matchmaking is FOR.

Obviously there will still be runs of wins and losses, potentially quite significant ones (particularly for new users of especial skill, or who are abnormally useless), but over time, your W/L will tend to 1.0.

This fact is commonly overlooked.

[MischiefSC]

I'm going to quote some info from the original post on matchmaking from Paul:

Quote

Some people also asked to have the description simplified. Here's the summary:

• The Match Maker uses a scoring system to determine if your team is more likely to win or lose based on your team's average Elo rating.
  
• If the Match Maker determines that you're going to lose, but you actually win, then your Elo score is going to go up and the enemy's score is going to go down.
  
• If the Match Maker determines that you're going to win and you actually win, then your Elo score isn't going to change very much (if at all). The same applies to a prediction of loss and you actually lose, your score may drop but it will be slight.
  
• The more games you participate in, the more accurate the Match Maker becomes and you will start seeing that you are playing against people of relative equal skill.

What does that mean?

The better you get, the more likely you're going to be playing against lower scored players, that means even if you win you don't get an Elo bump. If you lose, your score will drop.

If you lose to a higher skilled team you will lose only a tiny number of points, if any.

So if based on the Elo scores involved if team A is better than team B, the matchmaker will predict that team A will win. If team A wins, team A doesn't get much if any boost to their Elo. Team B doesn't lose much if any points.

If team B wins anyway then team A absolutely loses points, as many as 50 (that's a big chunk in a system that ranges from starting 1300 to uberninjagrandmaster 2800) while team B gains points, as many as 50.

Here is how you calculate how many points are gained or lost -

Team A Elo is 1500 average
Team B Elo is 1400 average

The Matchmaker predicts that team A will win.

If Team A wins, they gain no Elo.

If team A wins, Team B loses no Elo. It went as predicted.

If Team B wins, you get MATH.

Here is the math. First, you solve for the probability of winning (variable P).

P = 1/1+10(difference in Elo scores/400)

First you get the difference in Elo scores:

Team A: 1500 - team B 1400 = 100 point difference.

You divide that 100/400. You get 0.25.

You finish the 1+10(difference in Elo scores/400) first. So you plug the0.25 in and you get 1+10(0.25), which is the same as 1+2.5, which = 3.5.

So you plug that back into the equation and you get P = 1/3.5, which means P = 0.29 rounded up.

Next, you calculate how much you gain. Max is 50 points but the actual gain is calculated thus:

50 x (1 - P), which we now know is 50 x (1 - 0.29). So you do the bit in the parenthesis first, so you get 50 x 0.71 = 35.5.

Result? Team B changes from 1400 to 1435 (dropping the decimal). Team A change from 1500 to 1465.

Does that make a bit more sense? Variance in Elo scores between teams is actually required for people to really advance.

The bigger the difference in teams Elo scores the bigger the change in points. A match between exactly equal teams results in no points either way for a win or a loss.

Movement in Elo score is only going to happen when you lose to a lower scored team or beat a higher scored team.

[Lightfoot]

Well, you want your team to form up, at least the assaults, heavies, and slow mediums. Then you figure what your teams are missing and try to build that in mechlab. That's what I do anyway. I just expect ELO to hand me a challenge each time. Mechlab is important, getting that little bit extra on your mech makes all the difference.

[Agent 0 Fortune]

cleghorn6, on 26 January 2014 - 08:25 PM, said:

Ok, lets clear this up. If Elo matchmaking was working PERFECTLY, no flaws, no problems, exactly as specified, everyone would have a 1.0 W/L. Everyone would be losing exactly the same number of matches as they win. This is what Elo matchmaking is FOR.

Obviously there will still be runs of wins and losses, potentially quite significant ones (particularly for new users of especial skill, or who are abnormally useless), but over time, your W/L will tend to 1.0.

This fact is commonly overlooked.

Finally someone gets it. There is no need to give someone advice to "beat the matchmaker". If someone wants he can play with a HOTAS (joystick, throttle, pedals, etc), and run nothing but medium mechs. It is the matchmakers responsibility to find appropriate matches.

[Joseph Mallan]

Superslicks, on 25 January 2014 - 09:13 AM, said:

like I said 9 team loses, and this seems quite the norm now, something is not quite right with this game.....
I will retire just for today and see what tomorrow brings.
regards
josh

If you are PUGging you are worrying about the wrong stat. What did you do in each f those losses Josh? That is more an indicator of if you are doing good. Team work and everyone working together is handicapped by a lack of Comms, and cooperation.

[MischiefSC]

Joseph Mallan, on 27 January 2014 - 10:30 AM, said:

If you are PUGging you are worrying about the wrong stat. What did you do in each f those losses Josh? That is more an indicator of if you are doing good. Team work and everyone working together is handicapped by a lack of Comms, and cooperation.

Seriously, Joseph gets the Matchmaker way better than just about everyone.

It's not about winning or losing more or less. That's going to shake out in the averages over many games anyway.

It's about how well you do in every match, one after another. What mistakes you made, what you could have done different, did you carry all the weight you could or at least have fun.

[wanderer]

I find plenty of PUG losses are simply because the team, against all logic, does something tactically insane.

They "chase the squirrel" while being sniped to death. Then they ignore a badly damaged target to randomly pour fire into the undamaged guy next to him. Back off from a 4vs1 when they're the 4- but run in solo to a lance crossfire one at a time and die whining.

PUGs make no logical, predictable sense. Doing "well" in a match is difficult to rate when your team loses 5 members in under two minutes, leaving you to be one-shotted or backshot to death- your choice.

The only measure of skill is your performance in an organized group. Anything else may as well be rolling dice and hoping your number comes up on the "who doesn't get hosed by the random" chart.

[Mechteric]

I think being a no-respawn game accentuates the feeling of losing more than a respawn style game (like previous Mechwarrior games). Many people just aren't cut out for the risk versus reward of a no respawn game.

I'm not saying there's nothing wrong, I think tonnage limits will be a good start, maybe respawn style game modes could help those out who prefer that style also.

[Joseph Mallan]

CapperDeluxe, on 27 January 2014 - 11:16 AM, said:

I think being a no-respawn game accentuates the feeling of losing more than a respawn style game (like previous Mechwarrior games). Many people just aren't cut out for the risk versus reward of a no respawn game.

I'm not saying there's nothing wrong, I think tonnage limits will be a good start, maybe respawn style game modes could help those out who prefer that style also.

best observation I have read in a while!

[Mechteric]

Joseph Mallan, on 27 January 2014 - 11:17 AM, said:

best observation I have read in a while!

Thanks! And I had some more ideas while pondering that line of thought:


To really get an idea of what makes it unique and yet similar to other more popular no-respawn games I think we should see how Counter Strike compares (one of my personal favs back in my college days).

1. Counter Strike also had stomps. If you found yourself on a server against a "premade" team of opponents, you will lose, and you will lose hard. But, sometimes there would be some kind of auto-balancer running or you'd just leave the server after a few-several tough losses and find another server to hop on. Obviously you don't have this server choice in Mechwarrior Online, but the matchmaker kinda serves as a sort of auto-balancer (tonnage limits notwithstanding).

2. Dying early in a match in Counter Strike meant you were a spectator for usually no more than 3-5 minutes. In Mechwarrior Online, I think the average match time is double that somewhere. Of course the nice thing is that you can choose to just launch another mech immediately (more or less), which is nice for those who don't wish to wait that out.

3. Killing an opponent in Counter Strike was usually the result of somewhere between 1 and 10(ish) bullets hitting their body. Time to kill less than a second. Additionally every enemy had about the same amount of hitpoints (assuming everyone is smart and buys the armor). In Mechwarrior Online, the variables for time to kill are staggering. Sure an enemy could be taken down in less than a second, but most likely that would require that person to have made the mistake of being engaged by the entire enemy team (or piloting a Locust).


Therefore, I think when you combine the similar yet completely different factors you get a game that harshly penalizes hugely bad decisions, and yet at the same time allows for you to make some small bad decisions and still come out on top, I think mainly because of having a larger time to kill than any other no-respawn FPS out there.

Edited by CapperDeluxe, 27 January 2014 - 01:28 PM.

[Roadkill]

MischiefSC, on 26 January 2014 - 09:07 PM, said:

Movement in Elo score is only going to happen when you lose to a lower scored team or beat a higher scored team.

I think (hope) that Paul made a mistake when he said that, otherwise we're not really using Elo ratings. Your Elo should move every time you win or lose, but it should move MORE if the results was not what was predicted.

But you know that already.

[Ghogiel]

MischiefSC, on 26 January 2014 - 03:08 PM, said:

Yep, becaues Cbills and XP is how you get paid and rewarded. Elo and the matchmaker just try to keep the effort you have to put in to earn it the same with everyone else.

I guess people play for different reasons

[Solis Obscuri]

9 team is at disadvantage vs. 12 team.

[Procyon Alpha]

Solo Puging can go either way and you should just enjoy the ride....lol

[A banana in the tailpipe]

The popcorn has been delicious as I watch more and more elitists turn on each other with fewer pugs to stomp because of a horrible matchmaker.

[Corbon Zackery]

That's not even the record it takes about 14 to 20 losses in a row to really start to complain.

I had one match last night were we had 6 Atlas. 5 of the Atlas scored 2 kills. For a total of 10 kills

Dmg of other teammates:
0
37
67 6th atlas
100

12-10 loss

So the match was rigged by ELO for us to fail by giving us 4 guys to tank the game.

In the end it was really 8 on 12.

So sorry Paul the matchmaker is still fubar.

[Sarlic]

I have alot of lost matches too couple of weeks. Teams are utter bad. Really bad.

[Mudhutwarrior]

My record is 43 in a row many months back. It is better now as I rarely hit 10-15 in a row.
Losses never matter if its a good fight though. If its a stomp then its just rotten.