Matchmaking is broken and squishes innovation

[RocketZeus]

(Small wall of text)

 

My favorite part of this game is creating a deck that is creative to combat the current meta in the game.  They don't even have to be really creative, I just like decks that go anti-meta.

 

Enter my love for Glaceon - GX

 

While not necessarily anti-meta because it's still a GX pokemon, it puts a big damper on some of the constant GX abilities you see, and some of the most powerful like:

 

Metagross (not as prevelant as a short time ago, but still exists)

Necrozma with invasion ability (This made buzzwole, Hoho-GX and golisopod to name a few, a little tankier and versatile)

Sylvally GX - I've seen used with Buzzwole, fire decks like Hoho-GX or volcannion. 

(FUN FACT:  I actually played and lost to Zak Krekeler at the St. Louis regionals spring 2018, as he was running a celasteela, sylvally GX deck)

But anything that makes an opponent wait a turn to attack again, Sylvally has been thrown in

Gardevoir GX - Again, not as prevelant today, but still exists in the meta

Volcannion Ex - Again not as popular as before, but still has a decent presence

Tapu Lele - Not a pokemon decks are built around, but has a presence in almost every deck

 

But who is in a majority of competitive decks...?!?!

 

Zoroark GX - This guy is everywhere!  He's the draw engine for a lot of decks.  While decks aren't always, but sometimes are, built around him, he makes a lot of decks, even some of the decks mentioned above function. So what if we could shut him down?!  How powerful would you become?  How much fun would it be to see a Zoroark just camping out on someones bench or discarded out of frustration to the discard pile?! 

 

So I do it!!  I had to make some trades to get the Glaceons, but I'm determined!

 

I make my Glaceon-GX deck online, thinking I'm finally going to get my revenge on these powerful GX pokemon! But no, I haven't come up against a single person running a Zoroark.  I'm up to 12-14 matches (I don't think the deck stats for wins/losses are updated to the minute) and I haven't played any of the pokemon mentioned above.  Not once.  

 

But I have received a slew of non-GX decks now that I've created this deck. It's maddening.

 

Pokemon, you're control over the matches you create online is not only frustrating, but really diminishes the fun in the game while quelling creativity.  Obviously you have to try and match people up somehow.  But I don't think aside from the following categories, you need to manipulate the matches any further:

 

Player skill level 

Standard, expanded, legacy...etc

 

The meta will control itself in the types of decks people encounter.  But I'm sure there's a way to see someones skill level, be it wins/losses, events won or another type of criteria.  Controlling my match ups on the level you are now makes this game much less fun.

 

Is this just me?  I've considered compiling data and tracking every match I play to have hard data on this.  Maybe that would garner attention?  What are your thoughts, community?

 

 

[SandaledOtter]

But I don't think aside from the following categories, you need to manipulate the matches any further:

 

Player skill level 

Standard, expanded, legacy...etc

 

The meta will control itself in the types of decks people encounter.  But I'm sure there's a way to see someones skill level, be it wins/losses, events won or another type of criteria.  Controlling my match ups on the level you are now makes this game much less fun.

 

Is this just me?  I've considered compiling data and tracking every match I play to have hard data on this.  Maybe that would garner attention?  What are your thoughts, community?

Go for it, please. No one has presented any evidence that matchmaking is based on anything more than format and Elo, with Elo of reduced importance if finding a match takes more than a few seconds.

[RocketZeus]

On it!  I'm going to play two decks repeatedly, but alternate between the two.  One is with Glaceon GX as my main, the other will be with a deck I've built using no GX's or EX's.  First game I switched to the non-GX deck... Invasion Necrozma/Malamar with Zoarark-GX backing and Lele's.  Irony?

[BowserLuigi]

On it!  I'm going to play two decks repeatedly, but alternate between the two.  One is with Glaceon GX as my main, the other will be with a deck I've built using no GX's or EX's.  First game I switched to the non-GX deck... Invasion Necrozma/Malamar with Zoarark-GX backing and Lele's.  Irony?

Except that Glaceon GX does nothing to stop Ultra Necrozma (apart from stopping Leles and yes the invasion necro but at least the ultras can still charge up but that's general), so you won't have gotten the advantage ether way.

 

Unless the other guy is not using ultra necro then........uh just why?

Edited May 16, 2018 by BowserLuigi

[Princess_Aurora]

Just hit Ultra Necrozma squarely with a Fairy Wind or Infinite Force attack. Or anything that’s fairy. It won’t even know what hit it.

 

Make sure that it dosn’t have time to set up Malamars, though. As soon as two Malamars are on the field, you’re pretty much considered dead.

[Sadra]

Just hit Ultra Necrozma squarely with a Fairy Wind or Infinite Force attack. Or anything that’s fairy. It won’t even know what hit it.

 

Make sure that it dosn’t have time to set up Malamars, though. As soon as two Malamars are on the field, you’re pretty much considered dead.

Thats why always include a single copy of necrozma and mew, not just for consistency since you won't always be able draw that beast or 3 copies of steel every turn but because of gardies. 

 

With 4 inkays and 3 malamars, its almost definite that you get 2 malamars on turn 2. You can't say you lose instantly though, since in expanded every player is taking prizes every turn, one way or another. In the case of standard, yes, if they get that beast set up every turn, you lose. (I haven't played standard for a long time, so I'm just guessing). 

[Nitron2097]

x

Edited May 18, 2018 by Nitron2097

[BowserLuigi]

Zeus isn't specifically talking about meta matchups or cards to play; he's talking about the actual algorithm that decides who you face when you click on "Play" in Wersus mode. And I agree with him; it sucks.

 

I've had exactly the same experience:

After I built Buzzwole-GX, I played an inordinate amount of Psychic decks (weakness).

After I built fighting Garchomp, everyone suddenly started playing Gardevoir.

If I use Leafeon, I see a lot of Volcanion decks (again, weakness).

 

The algorithm needs to simply put you against the very next person who clicks play; manipulating the system "to create better matchups" (which it doesn't) doesn't create a realistic experience. You don't walk into your game store and sit down to play casual games and then the owner says "Wait, hold on...you're playing Buzzwole and they don't have Psychic Pokes so you can't play them."

 

And don't claim that doesn't happen; there are always hundreds of thousand of players online; it shouldn't take fifteen seconds to find an opponent at 7 pm EDT...

 

Event algorithms seem to be even worse. I've tried using Alex Schemanske's Buzz/Lyc list - which is a Top 2 deck - in Standard tournaments, and I brick...every time...and not just no Pokes to start. I mean not even seeing a draw supporter for five draws in a row. Completely unplayable...every time. Why?

I won't deny that it hasn't happened to me before, confirmation bias and all that aside. But really, what's it with the algorithm that specifically targets us forum posters and not...the other guys? Makes you wonder whether we are all facing the fabled super advanced AI that the PTCGO team seemed to have developed not so long ago

 

Btw the game uses an ELO rating system for its matchmaking so it taking fifteen seconds to find an opponent is not a problem worthy of raising your eyebrows over.

 

Bricking? Well...until we see the source code firsthand I doubt there's anything we can do about that.

[Nitron2097]

x

Edited May 18, 2018 by Nitron2097

[SandaledOtter]

So when that little "Advantage" icon pops up for your opponent...what does that actually mean? I've read a lot of stuff from people deducing that it simply goes to the player who's played more games...but I'm not so sure.

Advantage relates to the Elo rating of the players-- in all likelihood the sole component of match-making other than time-in-queue. It isn't about the player who's played more, or won more games, but their wins, losses, and the scores of those they've played against. You can look this up online, but I can't direct you to a source.

 

It's all math, and cares not a bit for what's in your deck.

 

To see this, play using a deck of 59 darkness energy and a single Generations Caterpie. You'll still get advantage on occasion, until you've sufficiently tanked your rating.

[Nitron2097]

x

Edited May 18, 2018 by Nitron2097

[SingingFlyEx]

About Elo: is there anything backing a claim that the ranking algorithm is of the Elo family rather than e.g. home made?

 

About matchmaking: it would be surprising that some advanced matchmaking is deployed for constructed formats but not for theme. Don't tell me "old" theme decks are actually playable now?

 

Explain how a top tier deck dead draws 3 times in a row: sure. Bad luck. 3 is a low number. You didn't compute the badness of your luck, but gut feelings are unreliable. Have you ever played the exact same list for 100 games in a row? Streaks of bad or good luck do occur.

Tournaments are much better than tokens for getting the cards you want. Even chests can be traded. This remains true even if the RNG is bugged, unless perhaps there's a bug consistently linked to your account, in which case you wouldn't be able to playtest on this account.

[BowserLuigi]

^ It's a good assumption (ELO I mean). As someone who concedes very often, I do get matchuped against mostly not-so-great opponents.

Edited May 17, 2018 by BowserLuigi

[SingingFlyEx]

^ My question was: do we now if the hidden ranking algorithm is specifically an Elo rating system, rather than any ranking algorithm. I thought maybe there was an official statement about implementing Elo.

[Nitron2097]

x

Edited May 18, 2018 by Nitron2097

[SandaledOtter]

Some data gathering; I burned my tickets in Theme tournaments...

 

Every time, the other player had "Advantage."

Four out of five, the other player won the coin flip.

Every time, they opened better than me.

Four out of five times I was paired first round against the eventual winner.

 

Take from that what you will.

What I will:

• The Theme format does have a ranking separate from the others, so if you haven't played Theme, that pretty much explains the first point.
• Five coin flips is a very small sample, so there's nothing there, and:
• Since the game isn't targeting you, the opening is just luck.
• A proper single elimination tournament bracket will always pair the highest-ranked players against the lowest. I've no idea if there is any consideration for that in this game, where the tournament is made up of the first eight people to join, regardless of rating or experience. If there were, and you were the lowest ranked there, it would make sense that you faced one of the best players each time.

[BigBearSD]

But who is in a majority of competitive decks...?!?!

 

Zoroark GX - This guy is everywhere!  He's the draw engine for a lot of decks.  While decks aren't always, but sometimes are, built around him, he makes a lot of decks, even some of the decks mentioned above function. So what if we could shut him down?!  How powerful would you become?  How much fun would it be to see a Zoroark just camping out on someones bench or discarded out of frustration to the discard pile?! 

 

So I do it!!  I had to make some trades to get the Glaceons, but I'm determined!

 

I make my Glaceon-GX deck online, thinking I'm finally going to get my revenge on these powerful GX pokemon! But no, I haven't come up against a single person running a Zoroark.  I'm up to 12-14 matches (I don't think the deck stats for wins/losses are updated to the minute) and I haven't played any of the pokemon mentioned above.  Not once.  

 

But I have received a slew of non-GX decks now that I've created this deck. It's maddening.

 

Pokemon, you're control over the matches you create online is not only frustrating, but really diminishes the fun in the game while quelling creativity.  Obviously you have to try and match people up somehow.  But I don't think aside from the following categories, you need to manipulate the matches any further:

 

Player skill level 

Standard, expanded, legacy...etc

 

The meta will control itself in the types of decks people encounter.  But I'm sure there's a way to see someones skill level, be it wins/losses, events won or another type of criteria.  Controlling my match ups on the level you are now makes this game much less fun.

 

Is this just me?  I've considered compiling data and tracking every match I play to have hard data on this.  Maybe that would garner attention?  What are your thoughts, community?

 

So basically this conspiracy theory boils down to you decided you would get many advantageous match-ups on the ladder by using Glaceon-gx. But now that you made the deck the game isn't fun because you are not getting bunches of the much more favorable Zoroark-gx match ups for your new deck.        

 

There can be no doubt this is because the match making system is actively keeping you away from all those Zoroark decks for sure!     

[BowserLuigi]

^ My question was: do we now if the hidden ranking algorithm is specifically an Elo rating system, rather than any ranking algorithm. I thought maybe there was an official statement about implementing Elo.

Don't think the devs actually disclose that. Everyone just assumed it's ELO

 

Btw TC I hope you realize that using Glaceon GX isn't really considered 'creative'

Edited May 18, 2018 by BowserLuigi

[RocketZeus]

I don't think Pokemon is out to get me.  I don't think they are singling me out.  I think they try to create matchups that tend to be kind of equal, all the way down to the type.  So that a deck doesn't just come in and destroy, not sure.  

 

I just finished a small sampling and am working on assembling the data.  I ended up playing 12 matches with two different decks.  I did this by playing matches and alternating between the two decks at the end of every match. My decks were:

 

Glaceon GX Deck:

 

Eevee's w/ 3 Glaceons

2 Rayquaza's to help retrieve energy/power up Poke's

One shining legends kaldeo (small wrinkle)

Oranguru 

1 Prism star Volcanion

 

Goal is to try and use the ability of Volcanion to disrupt opponent, but then also try and set yourself up for critical knockouts and multi-knockout turns.

 

Next deck:

 

Kingdra (from Burning Shadows) 4-1-4 line with 4 Rare Candy's

1 Shining Volcanion 

1 Prism Star Volcanion

Oranguru

2 Kaldeos's (shining legends)

 

This is my more "creative deck" for those keeping track and my more fun deck.  I just like Kingdra and it's attacks!  Goal is to use no EX/GX's.  Try to control who I attack using Guzma's/Prism volcanion.  Get opponent stuck and using Kingdra's sniping ability to set up multi-knockouts while never giving up more then one prize card myself.  This deck was actually more successful than the Glaceon deck.

 

I feel like I could tweak my decks, but I left them alone so I'd have better data.  My intent is to get up to 20 games for each deck.  If the data seems to be showing what I'm hypothesizing, I'm going to redo this again using the same Kingdra deck and a Bulu deck since you can see some parallels in how both my Glaceon and Kingdra decks try to control matchups and set up for multi-knockout rounds.  Using Bulu deck and the Kingdra decks would draw massive contrasts in play style and could yield different results.

 

I copied the data at the end of each match to compile my data.  Being truthful I made three mistakes:

 

I forgot to copy and paste the game data once for each deck.  So I was going to play games.  Then on my tenth game I forgot to paste the data before copying the log of the next game I played.  This led me to playing 12 matches per deck.  I know 24 matches is still a smaller sample size, but could maybe show us something.  

 

Working on getting the data to everyone in an easily consumable way.

Edited May 18, 2018 by RocketZeus