Guild Wars 2

Hi ANet and Evan Lesh,
Pre HoT, I was able to get a (55%) win-rate in Ranked matches. Unfortunately, I don’t have the “skill” necessary to compete in the expansion. Is it possible to lower my MMR artificially? I’m afraid I’m going to have to keep losing like this for a long time for the system to lower my MMR, and it’s really not fun.

I’m not being snarky. Thanks.

To the forum trolls: I tried to “L2P,” but clearly it wasn’t enough for me.

Oh hey, you won a match.

Are you playing Reaper, or standard Necro? Just curious.

Those games were all played on Reaper.

Our streak continues! Next time I see you I’ll send you an invite so we can lower our mmr’s at the same time.

Haha, count me in meow!

All I got from that Screenshot is that over the course of roughly 2000 unranked matches, you are at about a 50% win rate.

Could it possibly get more balanced than that?

Cogbyrn.7283:

All I got from that Screenshot is that over the course of roughly 2000 unranked matches, you are at about a 50% win rate.

Could it possibly get more balanced than that?

99% of those ~2k games were from before the expansion; my win-rate was balanced pre-HoT. Now, it’s not.

Salamander.2504:

Cogbyrn.7283:

All I got from that Screenshot is that over the course of roughly 2000 unranked matches, you are at about a 50% win rate.

Could it possibly get more balanced than that?

99% of those matches were pre-HoT.

Also, watch this:

Cogbyrn.7283:

Salamander.2504:

Cogbyrn.7283:

All I got from that Screenshot is that over the course of roughly 2000 unranked matches, you are at about a 50% win rate.

Could it possibly get more balanced than that?

99% of those matches were pre-HoT.

Also, watch this:

http://www.youtube.com/watch?v=4T0e1V1qovI

I’m not really sure the point you are trying to get across. By stating in my first post that I’d like my MMR lowered, I’m openly acknowledging that I’m worse than the game thinks I am.

The point is that you losing X games in a row is likely to happen if you have a 50% chance of winning a match based on matchmaking, especially with a larger sample space. It has nothing to do with your skill. It’s actually probable that you’ll hit streaks like that.

So the point is, you don’t actually know if you are worse than the game thinks. You just had a bad day. Human beings are really terrible at intuiting probabilities.

Cogbyrn.7283:

The point is that you losing X games in a row is likely to happen if you have a 50% chance of winning a match based on matchmaking, especially with a larger sample space. It has nothing to do with your skill. It’s actually probable that you’ll hit streaks like that.

So the point is, you don’t actually know if you are worse than the game thinks. You just had a bad day. Human beings are really terrible at intuiting probabilities.

When it was just me having a “bad day,” I was in total agreement with what you’ve said, but this losing streak is part of a much longer streak that’s been going on since Oct. 23rd. Unfortunately I don’t have the data beyond the last 10 matches to actually show this. Maybe a Dev can pop in here and supply that info, and just maybe he’ll prove that I’m full of it. But I didn’t make a post out of anger after only 10 games.

Without the data, I have nothing to go by, besides everything in your OP being snarky, including the part where you say you aren’t being snarky. I only have what is in the screenshot, which is not statistically abnormal. Just a bummer.

I’m also used to seeing streaks like this, as I have a friend who has played thousands of DOTA 2 matches, and he would always comment on his losing streaks when they were especially terrible.

Cogbyrn.7283:

Without the data, I have nothing to go by, besides everything in your OP being snarky, including the part where you say you aren’t being snarky. I only have what is in the screenshot, which is not statistically abnormal. Just a bummer.

I’m also used to seeing streaks like this, as I have a friend who has played thousands of DOTA 2 matches, and he would always comment on his losing streaks when they were especially terrible.

With the data provided you have the ability to actually determine how many games have been lost since HoT by comparing the combined current average to the previous average in contrast with matches played. I’m far too lazy to do the math, but you’ll find that the streaks required to reduce 55% win rate to ~51% at 2700 matches is extremely discouraging.

meow one twenty.4376:

Cogbyrn.7283:

Without the data, I have nothing to go by, besides everything in your OP being snarky, including the part where you say you aren’t being snarky. I only have what is in the screenshot, which is not statistically abnormal. Just a bummer.

I’m also used to seeing streaks like this, as I have a friend who has played thousands of DOTA 2 matches, and he would always comment on his losing streaks when they were especially terrible.

With the data provided you have the ability to actually determine how many games have been lost since HoT by comparing the combined current average to the previous average in contrast with matches played. I’m far too lazy to do the math, but you’ll find that the streaks required to reduce 55% win rate to ~51% at 2700 matches is extremely discouraging.

The OP mentions kitten Ranked win-rate, which is what is in the screen shot (since there hasn’t been a ranked season yet). Was Unranked/Total at 55% pre-HoT as well? If not, where was it at?

Great idea, Cogbyrn and Meow. I was fishing around to see what I could find and I stumbled upon this post of mine, dated to Oct. 28th: https://forum-en.gw2archive.eu/forum/game/pvp/Is-this-DH-Nerf-Enough-to-Calm-the-Tears/first. My post links a screenshot, which shows my win/loss ratio as of Oct. 28th.

Here is the data from Oct. 28th: https://dviw3bl0enbyw.cloudfront.net/uploads/forum_attachment/file/208816/dh1.jpg

On Oct. 28th, I had 1787 unranked matches and a 51.32% win-rate. Currently, I have 1992 unranked matches and a 50.80% win-rate. If you do the math, that means between Oct. 28th and Nov. 22nd (a 25 day period), my win-rate (over 205 games) was 46.27%. Because I was exclusively queing ranked up until HoT, that means my win-rate has dropped by 9% since Oct. 28th.

Salamander.2504:

Great idea, Cogbyrn and Meow. I was fishing around to see what I could find and I stumbled upon this post of mine, dated to Oct. 28th: https://forum-en.gw2archive.eu/forum/game/pvp/Is-this-DH-Nerf-Enough-to-Calm-the-Tears/first. My post links a screenshot, which shows my win/loss ratio as of Oct. 28th.

Here is the data from Oct. 28th: https://dviw3bl0enbyw.cloudfront.net/uploads/forum_attachment/file/208816/dh1.jpg

On Oct. 28th, I had 1787 unranked matches and a 51.32% win-rate. Currently, I have 1992 unranked matches and a 50.80% win-rate. If you do the math, that means between Oct. 28th and Nov. 22nd (a 25 day period), my win-rate (over 205 games) was 46.27%. Because I was exclusively queing ranked up until HoT, that means my win-rate has dropped by 9% since Oct. 28th.

Ah, so you were queuing Ranked in your pre-hot experience at kitten win-rate, and your 205+ games since HoT has a 46.27% win rate. So in short, instead of 11 games won out of every 20, you’re winning approximately 9.

What’s really interesting is that your perception is of a much, much more frustrating experience, when it’s just a couple extra losses that you aren’t used to every 20 matches (on average, streaks aside). Given 205 games since HoT, you are also running a 52.5% win-rate only including your Ranked matches pre-HoT.

Of course, you can try to keep applying filters and segments to the data in order to make it look bad, but even a 46.27% win rate isn’t that far below 50%. Considering randomness in unranked, I would imagine it could be within a reasonable deviation from “par”. I could be wrong, but I still don’t think it’s anything to ask for an artificially reduced MMR for.

EDIT – This is also the second time I’ve been censored because of the number 5. Once was “as” followed by 5, this is “a” followed by 55.

Thanks, swear filter. You’re the real MVP.

Cogbyrn.7283:

Thanks, swear filter. You’re the real MVP.

Swear filter, always fighting the good fight!

It looks like this has been pretty much resolved? One of my favorite aspects of statistics is how you can make bad things look good and good things look bad and still be “correct.” I consider every interpretation a sham until I see the data myself. :/

Hi Evan! Thanks for taking the time to respond. Nope, this is the account I play on. I’m not sure why my match log would stop on 11/21, as I’ve been playing consistently. As mentioned in my above post, I’ve had 205 unranked games on this account between Oct. 28 and Nov. 22. Are you seeing that the matches are absent before 11/21, or after?

EDIT – Pay no attention to the man behind the curtain.

Dear Evan Lesh can you do same for me as well? I main thief and face premades/teams made of top players quite frequently. There is no way i can compete even vs mediocre players on thief atm. Thank you.

Just throwing an idea out there (sometimes it helps software developers) could the absence of matches in the log be related to an absence of matches in the system, thus, making salamander mmr to be permanently stuck on a high level?

In case of doubt you could ask a dev to play with salamander in pvp and then proceed to check the log to see if it shows the new game.

Will we ever see that mmr? Games like dota share that bad boy freely. Maybe gw2 can start)

Pretty sure Everyone,s MMR will reset for when leaderboards open soon , well i hope so anyways . Q for Evan , is MMR bound to both unranked/ranked or separate ? , I do hope its separate so new players n vets get separated from unranked/ranked players

Evan Lesh.3295:

Also since noone will have played ranked in a while, everyone’s ratings will be more volatile than normal. This will help shake things up a bit.

This concerns me

Evan Lesh.3295:

Talyn Sneider.1825:

Just throwing an idea out there (sometimes it helps software developers) could the absence of matches in the log be related to an absence of matches in the system, thus, making salamander mmr to be permanently stuck on a high level?

In case of doubt you could ask a dev to play with salamander in pvp and then proceed to check the log to see if it shows the new game.

This is the first time I’ve seen missing matches. There are plenty of match logs up until 11-21, and it could just be a delay in our metrics system, which is separate from what the API pulls from (which is right from the source).

Confirmed we’ve got some backed up data so I can’t investigate right now. Your MMR looked pretty stable, though.

Thanks for the reply, hope this helps a bit (even if it’s only to remove possibilities off the table).

One more idea, could the delay of the data account for this situation? Imagine this generic example:

win – 10 mmr
win – 20 mmr
win – 30 mmr calculation is working with this info
lose – 30 mmr ( should be 20 but as data is delayed so mmr algorithm couldn’t take this into account)
lose – 30 mmr ( should be 10 but as data is delayed so mmr algorithm couldn’t take this into account)
lose – at this point the player can start to perceive his games as unfair…he lost several games and is still facing teams based on the 30 mmr

If this holds up, you can also factor games with low pop on the equation that can inflate/deflate mmr, imagine I as a player play at an hour where there are not many players that can match my mmr, i could increase or decrease my mmr creating a deviation due to the inability (not on the algorithm but on the player pool) of finding a match for my mmr.

Now on the following day, I’ll go play in prime time and get paired with players that are on my new mmr level, that would adjust my previous deviation by playing at low hours, but if data gets backed up I’ll remain in the inflated/deflated mmr score, leading to either a seried of wins/ loses depending on how much my mmr has been inflated/deflated.

I know my example is too small but this could happen in a day to day basis (example: during the week pvp at late hours at the weekend pvp during prime time)

Again hope this helps, and as a software developer I must give you props for the amount of efort and research youre doing to make an algorithm that is fair for us all.

Uhm… noob question? Where do I see those stats for myself o.o?

Nevermind found the site ;3

Evan Lesh.3295:

Decrypter.1785:

Pretty sure Everyone,s MMR will reset for when leaderboards open soon , well i hope so anyways . Q for Evan , is MMR bound to both unranked/ranked or separate ? , I do hope its separate so new players n vets get separated from unranked/ranked players

It is separate. Also since noone will have played ranked in a while, everyone’s ratings will be more volatile than normal. This will help shake things up a bit.

evan, can you try to convince whoever decided turning ranked off was a good idea to never turn it off again? being limited to unranked has completely killed my interest in PvP. the quality of the matches tanked, and there’s a not-insignificant chance of ending up on spirit watch or skyhammer, two maps i loathe.

and the worst part is that i don’t even get why ranked should be turned off just because the leagues are off.

Evan Lesh.3295:

Decrypter.1785:

Pretty sure Everyone,s MMR will reset for when leaderboards open soon , well i hope so anyways . Q for Evan , is MMR bound to both unranked/ranked or separate ? , I do hope its separate so new players n vets get separated from unranked/ranked players

It is separate. Also since noone will have played ranked in a while, everyone’s ratings will be more volatile than normal. This will help shake things up a bit.

Hi evan, i dont get this? How does absence of play (in ranked) lead to ratings being more volatile? Thanks in advance.

Thanx for the answer Evan this is good news

Can we please get back to topic and talk about Salamander’s loosing streak? If you now have access to the data, can you please take a look at it?

How did they look from an MMR-POV? Was he just in a possible fluctuation or was there a bug involved?

Evan Lesh.3295:

I have all the data now. You played everyone one of those games in a 3-man party, which will bump up your effective MMR for matchmaking. Prior to that day you played solo or duo mostly. The performance of your team must not be as good as the MMR bump is trying to account for in premades, so your MMR is adjusting downward to accommodate.

All this mmr talk! Just let us know our mmr

Or just tell me if my mmr is higher than my normal queue buddies

Honestly, I don’t see why we can’t see our own MMR. Just make it that you can’t see other people MMR if you worry about bad behavior. What’s the concern on now showing MMR in PvP panel?

Sunshine.5014:

Honestly, I don’t see why we can’t see our own MMR. Just make it that you can’t see other people MMR if you worry about bad behavior. What’s the concern on now showing MMR in PvP panel?

Im guessing without the 10’s of thousands of concurrent players, the mmr’s might look really silly with who is matching with who, so it’s best to keep it hidden to save face.

Evan Lesh.3295:

I have all the data now. You played everyone one of those games in a 3-man party, which will bump up your effective MMR for matchmaking. Prior to that day you played solo or duo mostly. The performance of your team must not be as good as the MMR bump is trying to account for in premades, so your MMR is adjusting downward to accommodate.

Thanks for looking into this. It’s a relief to know that my MMR is adjusting downwards at least. What still confuses me is, if I remember correctly, we were matched against a lot of 2-3-4-5 man teams that day, who should also get the MMR bump you mentioned. So for whatever reason, the team I was running with just didn’t warrant the bump?

Evan Lesh.3295:

Salamander.2504:

Evan Lesh.3295:

I have all the data now. You played everyone one of those games in a 3-man party, which will bump up your effective MMR for matchmaking. Prior to that day you played solo or duo mostly. The performance of your team must not be as good as the MMR bump is trying to account for in premades, so your MMR is adjusting downward to accommodate.

Thanks for looking into this. It’s a relief to know that my MMR is adjusting downwards at least. What still confuses me is, if I remember correctly, we were matched against a lot of 2-3-4-5 man teams that day, who should also get the MMR bump you mentioned. So for whatever reason, the team I was running with just didn’t warrant the bump?

You’re right that other teams you were matched against would have the same bump. It’s possible they were coordinating more or had more practice than your team.

Okay…although I must confess I’m still not clear on how this all works. I just left this match, where I was cannon fodder for a few high-level players who were in a 3-man while I was solo queing (though we had a 2-man on my team). Shouldn’t the Abjured 3-man have an MMR bump that puts them well out of my range? Is this just an anomalous example? I’ll sthu if so but I’m really just trying to understand what’s going on with my metrics. Thanks for the continued responses.

Salamander.2504:

Evan Lesh.3295:

Salamander.2504:

Evan Lesh.3295:

I have all the data now. You played everyone one of those games in a 3-man party, which will bump up your effective MMR for matchmaking. Prior to that day you played solo or duo mostly. The performance of your team must not be as good as the MMR bump is trying to account for in premades, so your MMR is adjusting downward to accommodate.

Thanks for looking into this. It’s a relief to know that my MMR is adjusting downwards at least. What still confuses me is, if I remember correctly, we were matched against a lot of 2-3-4-5 man teams that day, who should also get the MMR bump you mentioned. So for whatever reason, the team I was running with just didn’t warrant the bump?

You’re right that other teams you were matched against would have the same bump. It’s possible they were coordinating more or had more practice than your team.

Okay…although I must confess I’m still not clear on how this all works. I just left this match, where I was cannon fodder for a few high-level players who were in a 3-man while I was solo queing (though we had a 2-man on my team). Shouldn’t the Abjured 3-man have an MMR bump that puts them well out of my range? Is this just an anomalous example? I’ll sthu if so but I’m really just trying to understand what’s going on with my metrics. Thanks for the continued responses.

Not to make light of your situation as this is never fun. But, you have to appreciate the names. Like, BoB from accounting. How is that guy anyhow?

If anything, the problem is that ranked queue is off.

If you’re high MMR and you’re stuck in unranked, you get low MMR players on your team. —> Lose

If you’re low MMR and ranked queue is down, you get high MMR players on the opposing team. —> Lose

I made this observation after ~5 unranked matches, so it’s probably accurate.

Personally, after the crappy unranked conquest matches, I switched to stronghold and I farmed the hell out of it last night. Only losses are 4v5s and premades.

We should all be talking about the lack of a solo queue and the lack of temporary pvp bans for leavers. Dishonor in the game is not even noticeable. The 1% of players who are leavers ruin the game for the other 99% and they need to get shut down.

One d/c? 0-5 minute pvp lockout. Two d/c? 30-45 min lockout. Three d/c? 60-120 min lockout. A history of d/c? Double the lockout duration. If you don’t want to be there—don’t queue.

Gav.1425:

If anything, the problem is that ranked queue is off.

If you’re high MMR and you’re stuck in unranked, you get low MMR players on your team. —> Lose

If you’re low MMR and ranked queue is down, you get high MMR players on the opposing team. —> Lose

I made this observation after ~5 unranked matches, so it’s probably accurate.

Personally, after the crappy unranked conquest matches, I switched to stronghold and I farmed the hell out of it last night. Only losses are 4v5s and premades.

We should all be talking about the lack of a solo queue and the lack of temporary pvp bans for leavers. Dishonor in the game is not even noticeable. The 1% of players who are leavers ruin the game for the other 99% and they need to get shut down.

One d/c? 0-5 minute pvp lockout. Two d/c? 30-45 min lockout. Three d/c? 60-120 min lockout. A history of d/c? Double the lockout duration. If you don’t want to be there—don’t queue.

I don’t really have a problem as a high mmr player in soloqs. Still win 70% ish which is expected. I think the reason you won so many stronghold matches is because the players in that are extremely bad. Like worse than hotjoiners.