38 replies to this topic

[DaZur]

I'd rather "new players" get dropped into a separate queue, using the same Elo mechanic as used now, comprised of same/similar "new players" for the duration of their cadet bonus.

Once they finish their cadet bonus allotment, they are effectively considered "done with their cadet training" and they carry that Elo score (the one they fostered in the cadet arena) into the public pool.

This guarantees new players only drop with same/similar players, they cultivate their Elo score in that closed environment and once they effectively graduate to the public pool, their Elo is reflective of their actual skill and game-play proficiency.

No more dropping "fresh meat" in the deepest / most ruthless draw pool with zero experience... Once they hit the public pool, by virtue of their cultivated Elo score they are dropping accordingly without having taken a beating about the face and neck by veteran players.

Edited by DaZur, 14 March 2014 - 08:36 AM.

[wanderer]

Honestly, newbies should not even be part of the main ELO system, but rather be in a -second- ELO system.

Call it tier 0, the happy hunting grounds, whatever. When a player reaches a certain ELO rating in the newbie's ELO system, they are given a new ELO rating of zero (yes zero) and put into the main player pool. This means that players that hit an exceptional level of skill for a newbie and they'll end up in the regular game, while people who can't muster basic proficiency will no longer be cannon fodder until they do gain said basic skills.

[Amsro]

Elo = Team Win/Loss ≠ Player Skill

I don't understand why people think this makes ANY sense.

[Charons Little Helper]

I'd agree with new players being in their own sandbox for their cadet bonus or so - except that I don't know if MWO has a big enough player base to support newbie only matches without too long a wait time.

[Almond Brown]

Roadkill, on 14 March 2014 - 08:03 AM, said:

Suggestions like this come from a misunderstanding of how Elo works.

It isn't possible for new players to start at any point other than the middle in an Elo system. Elo is a zero sum system*, so if you move the entry point for new players all that does is move the "average" and all rankings then migrate to that new average. Doing so accomplishes nothing.

If you want new players to start low and work their way up you'll have to invent an all-new system.

But what could be done is to segregate new players (say, less than 50 matches) from experience players. Use the same ranking system for both groups, but force new players to play with each other until they have some experience and then move them into the pool with the sharks.

*not technically true in most implementations due to limits placed at either extreme, but accurate enough for our purposes.

The bigger issue is not keeping the New away from the Sharks. It is somehow keeping the Sharks away from the New. Now that would be a neat system.

[Trauglodyte]

Well, we can at least look forward to ELO buckets with the launch module. How much that changes our on personal game views, I don't know. But, it should group players from 0-1000, 1001-1500, and 1501-2800 or something like that. And, the games will only allow a single group per side whose ELO will be an average of the 2-4 players in that group. Unfortunately, without any of us knowing what our ELOs are, we might end up grouping with someone that is awful or we could be that person that is awful and accidentally causes the group average to go down thus pitting you with others that don't know how to walk and chew bubble gum at the same time.

[Charons Little Helper]

Trauglodyte, on 14 March 2014 - 11:07 AM, said:

Unfortunately, without any of us knowing what our ELOs are, we might end up grouping with someone that is awful or we could be that person that is awful and accidentally causes the group average to go down thus pitting you with others that don't know how to walk and chew bubble gum at the same time.

I do hope that they at least tell us which 'bucket' we're in.

[Trauglodyte]

Charons Little Helper, on 14 March 2014 - 11:34 AM, said:

I do hope that they at least tell us which 'bucket' we're in.

That would be nice. I don't need to know how close I am to either end of the specture within that bucket. Just let me know what bucket I'm in.

[Deathlike]

My thing about said buckets, is if the buckets have precedence over stuff, there's going to be a weight class logjam.. particularly at the higher Elo buckets.

If the buckets don't have as much precedence and simply more a "refined" version of what we have right now (MM v4+), we'll literally have more of the same now, except different weight brackets will have a completely different concentration of players (newbies having mediums, veterans having assaults, and variations in between).

So... it's not going to work out ideally... let alone optimally.

[Charons Little Helper]

Deathlike, on 14 March 2014 - 11:44 AM, said:

...we'll literally have more of the same now, except different weight brackets will have a completely different concentration of players (newbies having mediums, veterans having assaults, and variations in between).

I disagree - once taking a medium won't hurt your team / chance of winning (since they other team can have a highlander vs your hunchie) I think more veterans are likely to play mediums etc.

[Roadkill]

Trauglodyte, on 14 March 2014 - 08:22 AM, said:

At any rate, Champion mech drivers and new players need to have their ELO set to a much lower level than that of everyone else.

Your suggestion makes no sense. You cannot arbitrarily set someone's rating in an Elo system - it disrupts the entire system, which then self-adjusts to correct for the arbitrariness that you've introduced. It ultimately has no effect other than to screw things up (and not in a good way) for a while.

What you're really suggesting is that PGI use a system other than Elo.

[Willard Phule]

Charons Little Helper, on 14 March 2014 - 09:08 AM, said:

I'd agree with new players being in their own sandbox for their cadet bonus or so - except that I don't know if MWO has a big enough player base to support newbie only matches without too long a wait time.

This. Right here.

Why doesn't MW:O have a big enough player base to support a "newbie queue" for however many rounds or whatever you want to do to separate them from the seal clubbers until they can handle themselves?

I'll tell you why. MW:O is hemorrhaging players faster than it can replace them. New players find some kind of advertisement for the game somewhere, download and join in, drop a few bucks....get sick of being facepounded repeatedly, uninstall and move on to a more enjoyable experience.

My 12 year old is a PRIME example. He liked the game at first...even spent some of his allowance on a Heavy Metal. He hasn't played since UI 2.0 came out because "it gives him a headache to try to use it."

This game has been officially released but should still be in Beta. Promises have been made over and over and over with regard to content and when it will be added to the game...and those deadlines have been shattered, crapped all over and kicked to the curb.

Yeah, yeah...UI 2.0 was the choke point....blah blah blah. You've got your UI 2.0 now. Keeping to a 2 week patch cycle where you only release the bare minimum of any type of content or fix (besides more shiny new mechs to buy) is going to kill this thing.

And as much as I'd love to see MW:O become the realization of all the dreams and promises that have been made....maybe it's best to just let it crash and burn. Enough interest has been made with regard to the franchise...someone competent is bound to pick it up as soon as they can.

[Deathlike]

Charons Little Helper, on 14 March 2014 - 12:06 PM, said:

I disagree - once taking a medium won't hurt your team / chance of winning (since they other team can have a highlander vs your hunchie) I think more veterans are likely to play mediums etc.

That's not exactly what I said. Think about it in terms of popularity and usage. Is there an equal chance that a medium would be taken over a heavy or an assault by a "elite Elo" player, given the state of the game? It's not exactly about how much you may pine over mediums (and I pine over them as it is), but this is about playerbase we're talking about. If you took a poll over their goto mechs, mediums are not going to be at the top of the list.

The assumption on PGI's side is essentially saying that everyone will be carrying a medium and are more than willing to drive them like everything else. The problem is, do you think you have the playerbase, let alone the mechbays to field them on a consistent basis across all Elo levels? You're easily going to cause problems with the guys that won't drive anything but an Assault or Heavy.

I'm not saying there aren't people more than willing to drive a medium, but that's not enough of them to go around.

[Adiuvo]

Amsro, on 14 March 2014 - 08:45 AM, said:

Elo = Team Win/Loss ≠ Player Skill

I don't understand why people think this makes ANY sense.

Matchmaker is random. Everyone has the same chance of getting a good quality or poor quality team as you. Since everyone is under the same conditions, individual contributions will be exposed over time, and will be reflected in your W/L.

So yeah, W/L is pretty well linked to player skill. You won't be able to tell if someone with say, a 2.0 is better or worse than someone with a 4.0 since that's likely a result of player group quality, but something like .8 vs 1.5? There's a good indication on player quality in those kinds of numbers.

Edited by Adiuvo, 14 March 2014 - 02:59 PM.

[Willard Phule]

Adiuvo, on 14 March 2014 - 02:58 PM, said:

Matchmaker is random. Everyone has the same chance of getting a good quality or poor quality team as you. Since everyone is under the same conditions, individual contributions will be exposed over time, and will be reflected in your W/L.

So yeah, W/L is pretty well linked to player skill. You won't be able to tell if someone with say, a 2.0 is better or worse than someone with a 4.0 since that's likely a result of player group quality, but something like .8 vs 1.5? There's a good indication on player quality in those kinds of numbers.

Ok...I can see how you'd think that, but you're incorrect on so many levels that it's hard to set you straight....

Elo is married to but separate from the Matchmaker.

Elo starts out at 1400 when you're done with your Cadet Matches and tops out at 2900. Those are numbers you REALLY need to keep in mind. There are 1500 points between "Derp" and "Elite."

The Matchmaker is allowed to go up or down to a maximum of 1400 points to find players to fill a match. That means if you have a 2900, you could very easily be playing with people at a 1500 level. And vice-versa....if you're at 1500, you could very easily get clubbed by a 2900 because...that's fair.

The Matchmaker, when putting a match together, puts two teams together and predicts which one will win and which one will lose. This is IMPORTANT.

If you are on the team predicted to win and you do, in fact, win...your Elo will NOT CHANGE. If you're on the team predicted to win and you actually lose, then your Elo will drop by up to 50 points.

If you are on the team predicted to lose and you, in fact, lose...your Elo will NOT CHANGE. If you win, it will go up by up to 50 points.

Basically, your Elo only changes when you're on a team predicted to do something and you don't go with the prediction.

The Matchmaker is DESIGNED to put two teams together and try to make teams as even as possible by "average Elo score." I won't insult you by explaining how averages are figured out. We'll just assume you get it.

So...in any given match, since the vast majority of players are in the middle of the bell curve (many of which are 'new players' with artificially inflated Elo scores), the matchmaker is going to put the true middle players on one team...the team it predicts to win....and it's going to put the lower scores, (along with a few higher score "ringers" to even them out) on the team it predicts to lose.

Ultimately, the Matchmaker is doing it's absolute best to keep YOU and everyone else at a 1:1 win/loss ratio. Don't believe me? Look at your stats. You can't tell me you've got anything like 2500 wins and 1000 losses. It doesn't happen. If it does, then there's something wrong or you're dropping in 12 mans all the time....because the matchmaker doesn't effect 12 man drops at all.

It is what it is. It's half-finished, a "work in progress", something that drives new players away faster than you can say "fat woman".....and it's what we've got. Unless someone here is willing to go to the twitter feed and massage their egos long enough to get their attention, nothing is EVER going to be fixed. And...even if it IS fixed, odds are good it'll be half-assed and broken anyway.

Welcome to Derpwarrior Online.

Edited by Willard Phule, 14 March 2014 - 03:19 PM.

[Adiuvo]

Willard Phule, on 14 March 2014 - 03:18 PM, said:

snip

Well for starters, you never really addressed my argument. You're also wrong about the actual numbers. Elo starts at 1300 once you're done with your cadet matches. During cadet it's locked at 1100, and the maximum Elo possible is 2800. 100 off doesn't really matter I suppose, but considering the way you started your post it's kinda funny.

Matchmaker starts its search with a goal of putting players together who are of the same Elo, and it expands the range of acceptable values over time (up to 3 minutes), stopping at a total variance of 1400. You mentioned this. You are also correct that MM will assign a predicted winner and a predicted loser, and that your Elo will change in accordance to these predictions. I also 'get' averages, thanks for the vote of confidence

Higher Elo people are indeed balanced out to average with lower Elo people. This is very obvious in my, and other people's, matches. Now, I think it's this point where we differ. Matchmaker does indeed try its hardest to keep people at a 1:1 ratio. I'm guessing your own W/Ls are somewhere near that. Maybe 1.5, maybe 1.1, whatever. Doesn't really matter. In any case you're presenting your argument as if it's impossible to get above that. Especially with the 2500 wins and 1000 losses comment. That's only a 2.5 W/L. Pretty good, but not exactly amazing. There are many people with that or better.

Since you asked, here's the stats of my Ember. Since this mech just came out I feel it accurately reflects my current performance as a pilot.

EMBER  412  333  75  4.44  836  136  6.15  188,430  434,633  1 day 12:52:21

4.44 W/L, 6.15 KDR. Anyways, I do group and 12man in that thing a bit (I'd say about 60% of my matches are with at least 1 other person, typically 2, or in a 12man, so the W/L is a tad inflated. I did however, do 25 solo drops in a row once I got it mastered and put all the data in an Excel file. Since I can't really copy paste that, I'll just put down the notable stats.

W/L = 3.00
KDR = 12.2
Avg Kill/Match = 2.90
Avg Death/Match = .238
Avg Dam/Match = 520.857

That's solo dropping around 1:00pm PST on a weekend. Lower W/L, higher KDR makes a good amount of sense when solo dropping, since you don't have other good groupies sucking up some of your potential, but they also aren't there to help out and win the match. You can only carry so hard in a light mech, unfortunately My stats aren't even that amazing compared to other people. I know a few assault pilots with 15 W/Ls. For matchmaker's purposes we get the same games though.

Now as for if matchmaker is 'half-finished,' I have a few problems with it namely that it completely fails to take into account weight or chassis selection. A top assault pilot is always going to be able to carry harder than a top light pilot. A Dragon Slayer will always perform better than a Dragon. But with matching skill? That it usually handles pretty well. This is very obvious during top Elo prime time. All the good players literally match against each other constantly and the lances shuffle around. One match it may be a Lords 4man vs. an SJR 4man, or Lords + SwK vs. SJR + BSK. I'll be seeing Ladykiller, Carnage, JagerXII, TwinkyOverlord, Gman, etc.. in matches often. We play at the same times, and we're in the same Elo, so matchmaker just sticks us together. In short, it works.

Edited by Adiuvo, 14 March 2014 - 04:06 PM.

[Amsro]

Adiuvo, on 14 March 2014 - 02:58 PM, said:

Matchmaker is random.

Nope it isn't, it is very much picking players, for team A players that will be best suited to lose against team B.

What team you get placed on and how many newbs or veterans are on your team will determine if you are A or B.

This is the leading factor to all the "rolls" people claim in happen in matches. This is working as intended. No idea how the "buckets" will change this.

At least if its making team A and B from the same bucket you have a chance at a closer match. *crosses fingers*

[Amsro]

Adiuvo, on 14 March 2014 - 04:06 PM, said:

One match it may be a Lords 4man vs. an SJR 4man, or Lords + SwK vs. SJR + BSK. I'll be seeing Ladykiller, Carnage, JagerXII, TwinkyOverlord, Gman, etc.. in matches often. We play at the same times, and we're in the same Elo, so matchmaker just sticks us together. In short, it works.

I feel sorry for the newbs that get tossed into your matches.

ggclose

[Sarsaparilla Kid]

Amsro, on 14 March 2014 - 04:33 PM, said:

I feel sorry for the newbs that get tossed into your matches.

ggclose

Agreed, and it does happen. I set up an alt account a few weeks ago and shortly after my first 25 matches, was getting put into Charlie lance several times with members of SJR and SwK showing up in Alpha lance, and my W/L ratio at the time was just slightly over 1. Matchmaker is using fairly low Elo players to balance very high Elo players to even out the team Elo to something it can manage balancing against another team. Watching Jager's stream, in which Adiuvo has seen some action, they are often shaking their heads at either the matchmaking disparity between teams, or the disparity in the levels of skill and tactics within their team.