30 replies to this topic

[PPMcBiggs]

I am not quite sure how to word this post.

I have a feeling either elo is broken or elo is incredibly ineffective to the point of being useless.

I am getting really, really tired of stomps. That is where the score is 8-2, 8-1, 8-0. Happens all the time. Whether I am on the winning or losing side this is how at least 90% of the matches I observe end.

This is not fun.

This is happening because one team seems to get all the guys who have obviously earned their elo, while the other team has guys that are playing on their big brothers account or something. I don't know.

It sucks. Yeah it is nice to win when this happens but it REALLY sucks to lose when this happens. This means one team walks away with a hollow victory, while the other team is going to best buy to get a new mouse. Not fun.

Please PGI do something to balance the teams against each other.

I understand you are limited by groups and things like that, but please do something to make it better.

If we could see what our elo is we might be able to understand what is happening a little better. The game is beta right? That is one of the first things to get thrown in someones face for discussing problems. If it is beta, and we are therefore beta testers, give us some more tools to test with. Yeah yeah if people could see their elo they could game it or whatever. Who cares, its beta. Give me some tools to help me understand what is happening and I will report my findings.

Help me understand what is going on and I may, magically, decide I am actually having fun because now at least things make sense.

I want this game to be fun just like you guys do.

[Eldarstorm]

Definetly agree there is a problem that needs fixing and I would even suggest that its getting worse.

And when your on the loosing side in an assualt mech you get to the fight and it's all over.

Hopefully in these cases my elo doesn't suffer to much as I usually score highest. But as has been said it would be nice to know my elo and how it compares to other players.

Edited by Eldarstorm, 05 June 2013 - 05:51 AM.

[Star Captain Obvious Kerensky]

PPMcBiggs, on 05 June 2013 - 03:45 AM, said:

I have a feeling either elo is broken or elo is incredibly ineffective to the point of being useless.

This is the problem yes. The Matchmaker loosens the Elo and Tonnage matchmaking in order to ensure you find a match. Unfortunately there are not enough players to ensure a fair match of equal tonnage and Elo on a regular basis. So the Matchmaker just tosses some fairly abysmal teams together on a regular basis.

Your team down by 150 tons? Got several plays that don't know what double heat sinks are? That's the new normal.

Edited by Eldragon, 05 June 2013 - 06:22 AM.

[xDeityx]

OP, your frustration is happening because we don't have the population for an Elo-based matchmaking system to function properly much less one that also wants to incorporate other factors such as weight.

It wouldn't be so bad if our Elo was public because then you could see that losing those severely imbalanced matches that the matchmaker gave you a high probability to lose doesn't impact your Elo so much. Honestly though making a broken system transparent would just highlight PGI's incompetence. As a beta tester though I would want the following information available at the end of each game: what is my current Elo rating in the 'mech I piloted, and how did this game's outcome change it? What are the two team's average Elo ratings? What are the two team's combined tonnages (without having to sit there and add them up)?

That will never happen though, because PGI is all about obfuscation of data rather than transparency. That was evident when they took away the player counter that used to be in the mechlab client.

[PPMcBiggs]

Eldragon, on 05 June 2013 - 06:21 AM, said:

Unfortunately there are not enough players to ensure a fair match of equal tonnage and Elo on a regular basis. So the Matchmaker just tosses some fairly abysmal teams together on a regular basis.

xDeityx, on 05 June 2013 - 06:44 AM, said:

OP, your frustration is happening because we don't have the population for an Elo-based matchmaking system to function properly much less one that also wants to incorporate other factors such as weight.

I know what you guys are saying, but where is the proof? I recognize names from time to time, but it is not like every round is with the same 15 other guys.

I am not as concerned with tonnage as I am with skill, mainly aim. As I spectate my team (when I am losing) I am appalled at how bad these guys aim is, and how they obviously get shell shocked and can't manage to just pull the trigger. IMO tonnage is less important than basic skill. To illustrate I have gotten 1200+ dmg in everything from a Raven to an Atlas, so tonnage matters less than popular opinion makes it seem. The converse of this is absolutely depressing when you consider 8 awesome dudes against 7 losers and one guy of equal skill.

What can be done about this? Even if elo just arranged the teams differently the outcome might be different but it seems like there is always a major bias of skill towards a single team.

From what I can gather elo works on W/L. This is inadequate. It needs to take into account K/D, accuracy, DMG/round, AVG time alive. Those factors taken together may produce better results. Maybe it already does this, but I can't tell, and it is impossible to know.

[PPMcBiggs]

Eldarstorm, on 05 June 2013 - 05:50 AM, said:

I would even suggest that its getting worse.

I would say you are dead on. The worst part about it is that as you are teamed up with a bunch of lower skilled guys and, predictably, you lose the nature of elo keeps you with the losers despite the fact that you regularly are number 1 on the losing side by a large margin. Seems to me like a self perpetuating issue. This is why I think elo needs to work off more stats than W/L.

I wish screenshots were implemented a week ago when I did 1278 damage in my atlas, had 0 kills and my team lost the game. The rest of my team combined had less than 60% of my damage, I had twice the damage of the highest scoring enemy. Our team as a whole had 2 kills at the rounds end. WTF? This is the most glaring example that has happened to me. Conversely there have been plenty of rounds where the enemy team lost with less than 400 dmg total. That is pitiful, they were obviously out classed in every way.

Who has fun in those scenarios? One team might as well be on the testing grounds.

[Joseph Mallan]

I played for 2 hours last night. I didn't see a single name repeat nor did I recognize anyone... That is a sad thing seeing how long I have been playing.

Edited by Joseph Mallan, 06 June 2013 - 04:31 AM.

[Krzysztof z Bagien]

Elo system just can't work properely in a game like this. I suppose it assigns scores properely, but the problem is that those scores don't have anything to do with players' skill.
It was designed for chess, which is 1vs1 fully symmetrical game - so whether you win or lose depends only on your and your opponent's performance. That is the reason we can assume that winning player is better.
MWO is 8vs8 (soon to be 12vs12), both teams are of different composition and maps have different advantages and disadvantages depending where you start. That means the outcome of each match is a derivative of so many factors and we can't simply assume that player who is on the winning side is better. After all - good player with 8 pawns only would most likely lose against weaker player with 4 queens, 2 bishops and 2 knights.

Edited by Krzysztof z Bagien, 06 June 2013 - 04:25 AM.

[Innocent]

It seems to me there is a basic fix for the matchmaker (then again it may already do this). Have the matchmaker build a group of 16 people, then split the entire group into 2 teams of 8. Prioritize premades onto both teams first. If the original list of 16 has 2 groups of 4 then each team gets a premade 4 man, if there are more than two 4 man groups then put the highest elo 4 man with the random 4 people.

I think this would help with the expanding search problem. Right now if one team is built and it needs 2 more people those late draws might be pulled from much lower or higher elo players.

[Snuglninja]

One of the problems not saying the only, is people come into a match with expectations that piloting an assault with ppcs or lrm or whatever fotm is all you need to win spamming the whole time stick together, focus fire and these are tactics to them.
The get whatever map walk to the designated fighting area and start poping over the ridge getting killed 1 by 1 or just rushing single file into the opening getting wiped out. Then the QQ starts this is op that's op premades ruin the game.
Never once does anyone consider pulling fast mediums around a flank, lights capping - squirreling the enemy. Just rock'em sock'em robots with people so blood thirsty they rush for the kill steal shooting their teammates in the back.
Tl;dr Elo might not be the reason people are losing

[Blackadder]

Snuglninja, on 06 June 2013 - 04:45 AM, said:

One of the problems not saying the only, is people come into a match with expectations that piloting an assault with ppcs or lrm or whatever fotm is all you need to win spamming the whole time stick together, focus fire and these are tactics to them.
The get whatever map walk to the designated fighting area and start poping over the ridge getting killed 1 by 1 or just rushing single file into the opening getting wiped out. Then the QQ starts this is op that's op premades ruin the game.
Never once does anyone consider pulling fast mediums around a flank, lights capping - squirreling the enemy. Just rock'em sock'em robots with people so blood thirsty they rush for the kill steal shooting their teammates in the back.
Tl;dr Elo might not be the reason people are losing

bad players do not negate the fact that ELO should never be used in this system, given its heavy emphasis on team based gameplay, along with the low population that MWO has always had. its a bad system for this game, its been as bad for WOT at times, except wot has always had significantly higher numbers of players then MWO, so the larger player pool offsets some of the negatives.

[Lupin]

Yes team balance is worse than I have ever seen it, to the point I have reduced my game time again on MWO.
I wonder if this is due to pre-made teams being allowed to drop without another pre-made team on the other side. Or pre-made team tonnage not being correctly accounted.

I have put an post in "Ask The Devs - 40!" to see if they will reply:
http://mwomercs.com/...13#entry2424913
I hope I have asked the right questions.

[xDeityx]

PPMcBiggs, on 05 June 2013 - 03:01 PM, said:

I know what you guys are saying, but where is the proof? I recognize names from time to time, but it is not like every round is with the same 15 other guys.

PGI intentionally hides the data that would be needed to provide the proof.

PPMcBiggs, on 05 June 2013 - 03:01 PM, said:

From what I can gather elo works on W/L. This is inadequate. It needs to take into account K/D, accuracy, DMG/round, AVG time alive. Those factors taken together may produce better results. Maybe it already does this, but I can't tell, and it is impossible to know.

If Elo were to include those factors then people would game the system. Factoring only based on win/loss ratio and taking into account the disparity of the teams' average Elo ratings to weigh those wins and losses is the best way to measure player skill. Simplicity is also best when working on such a complicated task as matchmaking.

[SgtKinCaiD]

It's been months that players complain about balancing matchs with ELO ratings and it is always the same. In fact, this system has been controversial since its inception.

Even if we do not take into account ELO ratings, I do not understand PGI on this one : they created a game where you have to grind xps and cbills, test new builds / mechs, and you can do that only in ranked matchs ... derp !
Mastering blackjacks was horrendous last month. Now every time i have a new mech, i play only a couple of matchs before being fed up and switching to an already mastered mech, just in order to have decent matchs.

They threw us ELO rankings when everybody was asking for a basic CW. Plus, the matchs were better balanced when it was only taking in account the mech tonnage and they are still doing nothing.

[PPMcBiggs]

xDeityx, on 06 June 2013 - 05:45 AM, said:

If Elo were to include those factors then people would game the system. Factoring only based on win/loss ratio and taking into account the disparity of the teams' average Elo ratings to weigh those wins and losses is the best way to measure player skill. Simplicity is also best when working on such a complicated task as matchmaking.

I completely disagree.

How would one game the system? These factors can only be increased through skill. The only gaming one could do is practice. Which is my entire point.

Factoring on W/L is not even close to the best way to measure skill. I know there are people out there with artificially high elos simply because they are lucky enough to drop with a great team quite often and their team carries them to a win. The evidence is all around you while you play. Have you ever spectated or examined the post round report? You can witness it with your own eyes. Conversely there are guys out there with artificially low elos because they are unlucky and drop with crap teams quite often. The really sad thing for these guys is they are stuck in elo hell forever because if elo actually works it sticks them with people with similar elos and the cycle repeats. Put more data in the equation like K/D, DMG/match etc and now you can make some really educated guesses about a persons actual skill.

I agree that simplicity is best. That is why I would like to see elo die in hell. Give me completely random drops. That has got to be better than the current elo hell.

Or, how about this: since the game is a beta let me do some testing and turn it off if I like so that I may compare. I will report my findings.

[xDeityx]

PPMcBiggs, on 06 June 2013 - 07:30 AM, said:

I completely disagree.

How would one game the system? These factors can only be increased through skill. The only gaming one could do is practice. Which is my entire point.

Factoring on W/L is not even close to the best way to measure skill. I know there are people out there with artificially high elos simply because they are lucky enough to drop with a great team quite often and their team carries them to a win. The evidence is all around you while you play. Have you ever spectated or examined the post round report? You can witness it with your own eyes. Conversely there are guys out there with artificially low elos because they are unlucky and drop with crap teams quite often. The really sad thing for these guys is they are stuck in elo hell forever because if elo actually works it sticks them with people with similar elos and the cycle repeats. Put more data in the equation like K/D, DMG/match etc and now you can make some really educated guesses about a persons actual skill.

I agree that simplicity is best. That is why I would like to see elo die in hell. Give me completely random drops. That has got to be better than the current elo hell.

Or, how about this: since the game is a beta let me do some testing and turn it off if I like so that I may compare. I will report my findings.

The idea behind Elo rating systems is a very polished concept, it's not like people haven't thought about basing skill on metrics other than win/loss. But at the end of the day that's what it all boils down to and basing skill ratings only on win/loss accounts for all of the unmeasurable things that lead to wins and losses.

For example, let's say that your Elo was based on damage per round as well as wins and losses, but you are somebody who always wins your games by being a sneaky base-capper. You may be the single most important player on your team in regards to the reason why your team won, but your damage will be very low. Meanwhile that Atlas who was oblivious to the overall situation but just stayed in combat and pounded it out despite the fact that he's in Conquest and about to lose in points gets rewarded for his obliviousness.

Also, "Elo hell" is a myth. Given enough games, your Elo will be an accurate metric of the skill you have demonstrated while playing the game.

[PPMcBiggs]

xDeityx, on 06 June 2013 - 07:47 AM, said:

The idea behind Elo rating systems is a very polished concept, it's not like people haven't thought about basing skill on metrics other than win/loss. But at the end of the day that's what it all boils down to and basing skill ratings only on win/loss accounts for all of the unmeasurable things that lead to wins and losses.

For example, let's say that your Elo was based on damage per round as well as wins and losses, but you are somebody who always wins your games by being a sneaky base-capper. You may be the single most important player on your team in regards to the reason why your team won, but your damage will be very low. Meanwhile that Atlas who was oblivious to the overall situation but just stayed in combat and pounded it out despite the fact that he's in Conquest and about to lose in points gets rewarded for his obliviousness.

Also, "Elo hell" is a myth. Given enough games, your Elo will be an accurate metric of the skill you have demonstrated while playing the game.

It could be a polished system for a 1 on 1, symmetric scenario. When you are unlucky enough to get teamed up with terrible guys all the time how is this a skill problem for the player? How is this one guy supposed to hold up all 7 guys on his team round after round until elo places him in the correct spot? It does not seem possible. It seems to me that it is very likely that if you get a certain elo (most likely right in the middle of that pretty bell curve PGI is so proud of) you will be stuck there forever.

Elo hell is not a myth. If your data set is broken elo hell is the most likely outcome.

Lets face it, when you are in elo hell you are not fighting 8 guys, you are fighting 15. How can 1 guy do this consistently enough to dig his way out? If it were 1 on 1 I can totally see elo working.

[MeatForBrains]

PPMcBiggs, on 05 June 2013 - 03:45 AM, said:

I am not quite sure how to word this post.

I have a feeling either elo is broken or elo is incredibly ineffective to the point of being useless.

I am getting really, really tired of stomps. That is where the score is 8-2, 8-1, 8-0. Happens all the time. Whether I am on the winning or losing side this is how at least 90% of the matches I observe end.

This is not fun.

This is happening because one team seems to get all the guys who have obviously earned their elo, while the other team has guys that are playing on their big brothers account or something. I don't know.

It sucks. Yeah it is nice to win when this happens but it REALLY sucks to lose when this happens. This means one team walks away with a hollow victory, while the other team is going to best buy to get a new mouse. Not fun.

Please PGI do something to balance the teams against each other.

I understand you are limited by groups and things like that, but please do something to make it better.

If we could see what our elo is we might be able to understand what is happening a little better. The game is beta right? That is one of the first things to get thrown in someones face for discussing problems. If it is beta, and we are therefore beta testers, give us some more tools to test with. Yeah yeah if people could see their elo they could game it or whatever. Who cares, its beta. Give me some tools to help me understand what is happening and I will report my findings.

Help me understand what is going on and I may, magically, decide I am actually having fun because now at least things make sense.

I want this game to be fun just like you guys do.

8-1 or 8-2 win doesn't mean that you lost to 8 undamaged mechs. Battles only tip one way once all the armor is stripped, and then it's a domino effect. Don't be fooled by the number of surviving mechs.

[SgtKinCaiD]

MeatForBrains, on 06 June 2013 - 08:03 AM, said:

8-1 or 8-2 win doesn't mean that you lost to 8 undamaged mechs. Battles only tip one way once all the armor is stripped, and then it's a domino effect. Don't be fooled by the number of surviving mechs.

When half of your team did individually less than 100 dmg, it's not a domino effect.

[xDeityx]

PPMcBiggs, on 06 June 2013 - 08:00 AM, said:

It could be a polished system for a 1 on 1, symmetric scenario. When you are unlucky enough to get teamed up with terrible guys all the time how is this a skill problem for the player? How is this one guy supposed to hold up all 7 guys on his team round after round until elo places him in the correct spot? It does not seem possible. It seems to me that it is very likely that if you get a certain elo (most likely right in the middle of that pretty bell curve PGI is so proud of) you will be stuck there forever.

Elo hell is not a myth. If your data set is broken elo hell is the most likely outcome.

Lets face it, when you are in elo hell you are not fighting 8 guys, you are fighting 15. How can 1 guy do this consistently enough to dig his way out? If it were 1 on 1 I can totally see elo working.

Here's an article on why Elo Hell is a myth.

Here's a video.

You have 7/8 chances to get a bad player on your team. You have 8/8 chances for a bad player to be on the enemy team. Statistically speaking you play more often with idiots on the enemy than you do with idiots on your own team assuming you are a good player. The common denominator across thousands of matches will be you.

People think that Elo doesn't work because of the Dunning-Kruger Effect and because of perception biases that make it seem like the idiots are always on your team because you don't notice the enemy idiots as much. You always spec your own team when you die, and idiots on the enemy team you just kill and move on rather than examining their badness in detail like you do with your own teammates.

edit: i diot is a filtered word? Really? Really? /facepalm

Edited by xDeityx, 06 June 2013 - 08:41 AM.