277 replies to this topic

[Hauser]

Krzysztof z Bagien, on 20 April 2013 - 12:06 PM, said:

So you say that the using Elo system in long term should make players win/lose ratio get closer and closer to 50/50, am I right? Isn't it supposed to be exactly that with purely random matchmaking? So how can you judge if it is working properely?

With a random matchmaker the overall average w/l of every player would be 50% however for individual players this would not be so. Better players would have a higher w/l. Poor players a lower w/l. This would happen because as I showed above, they would run into more opportunities where they can carry their team.

An ideal match maker would make teams so that every individual has 50% of winning. A match maker based on elo is capable of doing this.

Krzysztof z Bagien, on 20 April 2013 - 12:06 PM, said:

Why do we even need any rating system? The answer is: not for the sake of rating each and every player (as we can't see our Elo score), but to make player experience better - to make them play against opponents of simmilar skill. And this player experience would be ultimate proof that system works well or not, but it is difficult or even impossible to measure.

Instead win/lose ratio is used to show everything works well, but there's no difference between Elo and random at the end.
Trust me, many people smarter than PGI devs spent lots of time and resources to create matchmaking system able to match people with similar skill level, and none of those systems are as simple as Elo's.

If an elo match maker is working correctly the w/l of individual people should be trending towards 1.00 and overall their elo ratings should be in a bell curve distribution. Even though the game has people playing in teams against each other. It will converge a bit slower that's about it. Most people however seem to judge match making by the score which tends t be 8-0.

Krzysztof z Bagien, on 20 April 2013 - 12:06 PM, said:

And you are not right, I'm quite familiar with how statistics work, I'm just using oversimplified examples to show what I mean, English is not my native language so I find it difficult to articulate my thoughts sometimes.

Ah apologies then.

[Redoxin]

PhoenixFire55, on 20 April 2013 - 10:32 AM, said:

Elo as it is now is being stupid, because win/loss ratio it is based off means NOTHING. Your smart curve only shows that 90% of people win about 50% of their games, meaning that its just being a random factor of haveing good/bad team. If you want to really evaluate players skill you have to use their game score related to their team average score, thus better players will always gain Elo even if they end up on a horrible team.

Such a clueless post. First of, the curve shown does NOT show that 90% of people win 50% of their games. It doesnt show anything about win percentage of individuals (I would be very keen to know this though). It just shows that the Elo values are evenly distributed among the population (which should be a given anyway).
Secondly, of course you influence the win chance of your team. It is not random if you win or lose. It does not only depend on the team excluding you, it depends on the team AND you. if you believe your play doesnt influence the outcome, why do you play at all.
If my play does not influence win chances, why did I lose 90, but win 140 games while the Elo system is in place? And that while playing solo.

And your match score does not matter as much as you believe. You can do a lot of stuff that helps your team win that doesnt get you any points. For example going back to base to stop capping, attacking the right enemy and not just randomly, luring an enemy out of their formation so that he gets killed by your pals etc.
Just last game I played with soemone who was a great shot and kiled multiple mechs, but refused to go back to base to stop the cap ("I dont care about the base"). So we lost although we destroyed most enemy mechs. Why shold this player get high Elo and play games with people that try to win hard?
In the end there is only one true way to determine how much someone contributes to the success of his team, and that is tracking if his teams win or lose, which results in a Elo score.

edit: Also thanks Hauser for the post. At least some seem to understand how Elo works. I doubt it will help though.

Edited by Redoxin, 20 April 2013 - 01:40 PM.

[Alpha087]

Weight balance might be good, but skill balance is far from it from what I've seen... Out of the last dozen or so matches that I've played, more than half of them ended with 0 - 2 kills on one side, and 8 kills on the other.

I'm currently at 273 matches played, so I'm not certain what effect that has on this invisible ELO system, but whatever was changed around has made for a much worse play experience, at least for me. I'm finding myself rage-quitting after multiple bad match-ups more and more now.. As opposed to pre-patch where I could just keep playing for hours on end.

[Redoxin]

Stoicblitzer, on 20 April 2013 - 12:16 PM, said:

It is unfortunate that Elo has very little to do with personal skill and all to do with wins/losses. What is that match score stat they added a few months ago? Seems to me that would be a decent indicator of personal skill. Maybe the mean of the match score stat would be better than the current MM system? I'm just brainstorming. Maybe put (2 high elo + 6 bad elo) vs (2 high elo + 6 bad elo) instead of (4 magnificent elo + 4 horribad elo) vs (8 above average elo) which is what it seems to do now.

Win/loss is the result of personal skill. These 2 are strongly related. You wont win EVERY game, but you can strongly increase your teams chances to win. Over a larger number of games, this higher win chance will accurately be reflected by a higher number of wins which will again result in higher Elo.
Of course people always fault others and not themselves for any loss. Thats human nature, but irrational at the same time.

[Jonathan Paine]

perhaps it is time to turn 8 man groups into:
2+2+2+2, that is to say two mechs of each class per side. You should be able to add tonnage matching and Elo to that.

[Redoxin]

Alpha087, on 20 April 2013 - 01:38 PM, said:

Weight balance might be good, but skill balance is far from it from what I've seen... Out of the last dozen or so matches that I've played, more than half of them ended with 0 - 2 kills on one side, and 8 kills on the other.

I'm currently at 273 matches played, so I'm not certain what effect that has on this invisible ELO system, but whatever was changed around has made for a much worse play experience, at least for me. I'm finding myself rage-quitting after multiple bad match-ups more and more now.. As opposed to pre-patch where I could just keep playing for hours on end.

Games going 0-8 or 1-8 is just in the nature of the game. Such scores are not necessarily the result of an uneven player distribution. If one team gets in a good position, and kills 2 enemy mechs without loosing one of their own, the game is usually over. From then on they have an advantage in fire power and can just steamroll the enemy team. I have had 8 man premade drops, where we won 8-0 in one game and then lost AGAINST THE SAME TEAM 0-8 in the next game. So no matter how even the player or mech distribution is, these steamroll games where one just destroys the other will always exist.
You enjoying the game less is probably the result of winning less games. I will go out on a limb and assume you won much more games than you lost before the Elo (and killed more mechs than you got kileld yourself). After introduction of Elo, you get better opponents and get killed more often and lose more often than before. Maybe to a point were you have a 1/1 win loss or kill/death ratio. Most people dont like it if they die as often as they kill.

This is also a reason why I am a strong proponent of showing the Elo, at least to the player himself. Because a player (like you maybe) might actually do really good, but never notice it because he gets better and betterb opponents the better he does. He gets frustrated because he still keeps on dying and no stats improve, no matter how good he plays. If he knew his Elo, he would actually know that he is doing better and better and get motivation from that. Without knowing your Elo, you always float in some kind of vacuum, never knowing how well you actually do.

Edited by Redoxin, 20 April 2013 - 02:01 PM.

[Hauser]

Alpha087, on 20 April 2013 - 01:38 PM, said:

Weight balance might be good, but skill balance is far from it from what I've seen... Out of the last dozen or so matches that I've played, more than half of them ended with 0 - 2 kills on one side, and 8 kills on the other.

I'm currently at 273 matches played, so I'm not certain what effect that has on this invisible ELO system, but whatever was changed around has made for a much worse play experience, at least for me. I'm finding myself rage-quitting after multiple bad match-ups more and more now.. As opposed to pre-patch where I could just keep playing for hours on end.

What you describe is a bit nebulous. But I think there is a good explanation. Recently host state rewind has been implemented so it becomes easier to hit people. The current popularity of PPC's increases this. Everybody has become more fragile.

edit:

And as Redoxin pointed out, you might be going up against better players.

Edited by Hauser, 20 April 2013 - 02:16 PM.

[Alpha087]

Redoxin, on 20 April 2013 - 01:56 PM, said:

I will go out on a limb and assume you won much more games than you lost before the Elo (and killed more mechs than you got kileld yourself). After introduction of Elo, you get better opponents and get killed more often and lose more often than before. Maybe to a point were you have a 1/1 win loss or kill/death ratio. Most people dont like it if they die as often as they kill.

The rest of your explanation does actually make a fair bit of sense, but as far as my enjoying it less, that wouldn't be the case at all. I actually had a win/loss and kill/death ratio that was below 1.00 up until this past week or so. If the teams aren't as unbalanced as they appear, like you say, then it could very well simply be that the Elo caught up with me and I'm getting thrown into games with players (on both sides) who are a bit better, and thus more challenging.

I guess I'll just have to look into finding an organized premade group, so I can go into a match knowing that at least half of my team will be using some sort of strategy.

[Deathlike]

If you're consistently contributing to the match somehow to generate the win, you are more likely to have higher ELO. If you're not doing much in the game (aka, the newbie), you'll probably not going to have much of an ELO to speak of.

Now obviously, in the context of kinda of matches MM tries to create, you cannot fully guarantee a win. There's plenty of other factors that work against you... as much these factors can also work for you. For instance, if your team is paired with a complete newbie, and he gets himself killed, you're probably not going to get penalized much for that... especially if your team is primarily "newbie based". Assuming you get better... chances are your teammates get better. So, what would normally cause you to "be the star" of the match in some newbie game, you have to "pull your own weight", compared to your peers when ELO is closer to them.

So.. even if it's really not your fault that the team chased after the squirrel, you're ELO probably will change little. The only time ELO really hurts you, is you crap the bed, repeatedly, due to "not getting better".

ELO is not a simple matter... if correctly implemented, it would work, but remember that not all games are created equal, and that CONSISTENCY AND GETTING BETTER will allow you to have a great chance at success. MWO is a team game.. and when you can get everyone working together, you are more likely to win. The same can be said for your opponent. It is simply not possible to rambo and claim ELO is failing you... it may simply be because you are failing ELO.

[Mystere]

Jonathan Paine, on 20 April 2013 - 01:48 PM, said:

perhaps it is time to turn 8 man groups into:
2+2+2+2, that is to say two mechs of each class per side. You should be able to add tonnage matching and Elo to that.

I respectfully disagree. That make things predictable, reducing variety, which I equate to boring.

Besides, 12x12 is coming next. What should we do then?

[Kageru Ikazuchi]

For the guys who want to track their Elo scores, to see if they're getting moving up or down, PGI had said that something like a ladder ranking is coming.

In the meantime, I recommend tracking your wins and losses in various weight classes (you have four Elo scores ... one each for each weight class) over a set amount of time. I track mine weekly.

Keep in mind that you should use your existing stats as a baseline, since they just recalculated everyone's Elo. If yours was overinflated, like I suspect mine was, it went down on Tuesday ... if it was too low, it went up.

For example, let's say that as of Tuesday, you had 100 wins and 100 losses in a particular weight class ... use these numbers as "0". Next Tuesday, you check your stats for that week, and it's 110 wins / 140 losses ... your Elo is probably going down ... the next week, it's 140 / 160 ... it's going up.

Edit: Oh - and they have also said that a stat wipe is coming when an achievement system is implemented ...

Edited by Kageru Ikazuchi, 20 April 2013 - 05:57 PM.

[PhoenixFire55]

Hauser, on 20 April 2013 - 11:53 AM, said:

Now you've got me puzzled. On one hand you say that your skill has nothing with the performance of your team, yet on the other hand you want to know this statistic based on the result of your team.

Lets put it this way ... if a total noob with no skill whatsoever just drops with 7 awesome players and always wins his Elo will be sky high. Does he deserve it? A great player who is PUG-ing and gets dropped with crappy players will always lose. Does he deserve it?

I want to see a statistic based on MY performance, and I want it to change according to MY performance, and not because I had awesome teammates who carried me or awful teammates who totally failed.

I don't understand why you defend current Elo system. What I suggest does not take Elo out of the picture completely, I'm just saying it needs to be based off YOUR personal performance, not your team performance.

Edited by PhoenixFire55, 20 April 2013 - 11:49 PM.

[PhoenixFire55]

Krzysztof z Bagien, on 20 April 2013 - 12:06 PM, said:

So you say that the using Elo system in long term should make players win/lose ratio get closer and closer to 50/50, am I right? Isn't it supposed to be exactly that with purely random matchmaking?

It already is just that. You need just 1 look at their new Elo distribution to see that vast majority of players are around that 50% win zone. Those who are not there are either really really bad players, afk'ers and griefers or players who always play in quality groups preferrably sync-dropping. As I said, 8:0 - 0:8 stomps are happening with the same rate I've had in closed beta when we had total random matchmaking with 8-man premades dropping with everyone else.

[PhoenixFire55]

Thontor, on 20 April 2013 - 01:01 PM, said:

Know what's not cool? dropping into Alpine Peaks in Conquest and having only heavies and assaults on your team, when the other team has three lights.

It seems like this happens every time I get that map/game mode combo... and every time, the other team is the only one that has light mechs....

there's pretty much no chance of winning against that. Would really like weight class matching to be back...

I'd say ... use tactics and defend bases you already have leaving people there, but ...

Thing is you don't know exactly what mechs other team has. If we knew right away that opposing team has these and these mechs we could adjust, but what we have now is that you never know. If you leave defenders on bases there is a chance that the rest of your team is gonna get rolled by an 8-man deathball. If you don't there is a chance that just 1 light from the other team is gonna cap everything without fighting and win it for his team even if the rest of them will get annihilated.

Back in closed beta we knew that if our team had n lights then the other team also had n lights, meaning we could predict their movements and counter it. IMO it was more tactical then it is now. Right now everyone does the same thing over and over on every map because there is no sense in not doing it.

If PGI really wants a 'thinking person' shooter they need to bring tactics and strategies back into game.

[PhoenixFire55]

Redoxin, on 20 April 2013 - 01:30 PM, said:

Such a clueless post. First of, the curve shown does NOT show that 90% of people win 50% of their games. It doesnt show anything about win percentage of individuals (I would be very keen to know this though). It just shows that the Elo values are evenly distributed among the population (which should be a given anyway).

You honestly think that current Elo has smth to do with your skill? What it shows is that the vast majority of people who play the game has nearly exact same Elo score. Only people who are not in that middle Elo zone are:
- afk'er, griefers (TK'ers) and just horrible players (5% "left wing")
- people who only play in really good groups and thus nearly always win (5% "right wing")
What it means that everyone else (skilled players and non-skilled players) are in the same Elo zone, which in turn means that their Elo rating just gets random adjustments up and down with equal probability when they end up on a randomly good / bad team.

You can't win against a 'far better opposition' (and visa versa) in terms of Elo beacause you never meet that 'far better' in your games. You only play inside your middle-zone. Even if you pull a rabbit out of your hat in one of your games your Elo won't go up any more then on any other of your wins. More importantly this win will quickly get negated by a few random losses you'll soon have. Everyone starts in the same zone and there is no way to get out of that same zone just as graph shows. The fact that some new players do get out of there only means that they do one of the two things stated above.

Redoxin, on 20 April 2013 - 01:30 PM, said:

Secondly, of course you influence the win chance of your team. It is not random if you win or lose.

An example for you ... completely noob player gets dropped with 7 awesome guys, does nothing in his game (0 damage etc.) but his awesome teams wins even without him. His Elo goes up. Is that fair?
Or ... an awesome players gets dropped with 7 crap players. Obviousely he can't win. His Elo goes down. Is that fair?
Or ... 4 'good' players are in team with 4 'bads', other team has 8 'average' players. Bads run away, 4 good guys are easily outnumbered by 8-man deathball. Is it fair that 4 'goods' will have to suffer Elo drop?

One person can't influence the outcome of the game. The 'difference' that you personally make is so little that it'll never really show up on current Elo graph (thats why most of us are around the middle). With 12 vs 12 coming this difference will be even less.

Redoxin, on 20 April 2013 - 01:30 PM, said:

And your match score does not matter as much as you believe. You can do a lot of stuff that helps your team win that doesnt get you any points.

Never said that the current match score is perfect. PGI will have to keep improving it. IMO you need to include factors like 'damage taken', 'distance travelled', 'locks aquired/kept', 'time on hostile cap zones', 'time on your cap zones when enemy mechs are on them' etc. But 99% of matches I've played people who really did well had higher game scores.

I don't understand why people defend the current Elo system. It has nothing (or very very little) to do with your skill. Obviousely game score has far more to do with your skill that wins/losses.

[PhoenixFire55]

Jonathan Paine, on 20 April 2013 - 01:48 PM, said:

perhaps it is time to turn 8 man groups into:
2+2+2+2, that is to say two mechs of each class per side. You should be able to add tonnage matching and Elo to that.

I'd say turn EVERY game into 2+2+2+2. If most people will only want to play assault mechs they'll have to wait forever to get a match. Same time people who play lights and mediums will find games right away. Encourage people to play different weight classes. Might not be so radical, you can do 'no-more-then-4-mechs-of-the-same-class' per game kinda thing.

[SoulROM]

I will give positive feedback for the improved MM.

I want to reiterate the importance of the global chatroom (or rooms) that is needed to maintain and further connection to groups. The C3 interface needs to be more of an inclusive VOIP that automatically is generated with each match. This would be a compliment to the MM algorithm.

Communication is key. Discussion after a battle is key.

This game is set up perfectly to literally be a "heatsink" for aggression stress and any other aspects of life's daily toil. So to make sure that I have not gone completely off topic, the new MM/ELO has created more balanced matches to allow the weekend warrior (such as myself) to have fun and improve my pilot/gunnery skills.

[Mystere]

PhoenixFire55, on 20 April 2013 - 11:55 PM, said:

Thing is you don't know exactly what mechs other team has. If we knew right away that opposing team has these and these mechs we could adjust, but what we have now is that you never know.

Knowing right away the enemy composition leads to pre-determined tactical decisions, which I equate to boring, especially because you know that your team is almost like the enemy team. I much prefer not knowing the enemy composition until they are seen, as it forces teams to make tactical decisions on a more dynamic basis.

[VXJaeger]

NOw this match-making system is working as good as it can. 4 now, GJ dudes

[Hauser]

PhoenixFire55, on 20 April 2013 - 11:44 PM, said:

Lets put it this way ... if a total noob with no skill whatsoever just drops with 7 awesome players and always wins his Elo will be sky high. Does he deserve it? A great player who is PUG-ing and gets dropped with crappy players will always lose. Does he deserve it?

Now you're treating elo as a personal ranking system rather then a component of the match making system. The critical difference is that while the match maker does require some measure of skill to assign players to a match, it does not have to be an accurate assessment of that person. The only thing the matchmaker needs is a rating that describes how the player plays.

The situation you describe is already highly improbable in a random match making scenario. You must be the (un)luckiest player in the world to drop with such good/bad people. However if both are consistently dropping with 4 friends that are significantly better or worse, then their ratings are deserved. It accurately describes how the whole group will preform.

As I explained previously to Christof, given a random match maker good players will encounter more matches where their team has the advantage. The good players team will always have 1 good and 7 randoms, while the other team has 8 randoms and thus a lower chance of getting equal or more good players.

Elo is an improvement on random match making because it puts the players who would lose allot w/ random match making with similar players, just as it puts people who would win allot together. The result being that for every

PhoenixFire55, on 20 April 2013 - 11:44 PM, said:

I want to see a statistic based on MY performance, and I want it to change according to MY performance, and not because I had awesome teammates who carried me or awful teammates who totally failed.

I don't understand why you defend current Elo system. What I suggest does not take Elo out of the picture completely, I'm just saying it needs to be based off YOUR personal performance, not your team performance.

I'm not defending the Elo system. I'm explaining how it works. I'm trying to explain that in the long run, if you play enough games, the goods and the bads will level out against each other. The only constant contributing factor that can steer your elo up or down is you.

In other words just play the game for the sake of playing the game.

Now if you want to know my criticism on the current implementation it would be dropping new players in the middle of the elo curve. This is frustrating for everybody. Dropping them at he bottom end with accelerated elo change would be a better solution.

Edited by Hauser, 21 April 2013 - 05:40 PM.