New Feature - Champions 2.0 (Live with R287)

entrailbucket

@meadowsweet said:

@entrailbucket said:
Matchmaking is based on top 3, not top 1, so buying and levelling up that Iron Man won't immediately throw them to the wolves.
But what is the alternative here? How should it work?

It's probably not realistic, but I keep wondering how different would be if the game's PVP was based on an Elo or chess-style rating system: every player has a score / ranking. When you win, it goes up. When you lose, it goes down. The amount it goes up or down is based on your opponent's ranking (so an underdog upset matters more than a coin-toss competition between equals.)

Levels of top characters isn't a terrible approximation, but in reality someone who has Level 550 Chasm, Thor, Beta Ray Bill, Okoye, Hulk, etc. is not the same as someone whose only Level 550's are Wasp, Rescue, Loki, Archangel, Jeffrey, etc. The first player would have a much higher Elo (because they almost always win) while the second player's Elo would go down (because they often lose against teams with a better meta / synergy.)

So your match-making would be based on your actual performance, skill, or win-loss records. Not just based on the fact that you impulsively decided to level up Mysterio, Northstar, and Talos prematurely...

This is how it worked 10 years ago, when the game first launched!

We broke it IMMEDIATELY. What we'd do was join an event (at 0 points) then retreat 100 or so times straight (can't lose points when you don't have any!) until our win/loss record went into the red, and the game started showing us 1* teams. Another fun one was joining lightning rounds just to retreat over and over.

LavaManLee

They've boxed themselves into a corner, unfortunately. Some games don't calculate the MMR until, essentially, you have chosen your team. Meaning that depending on who I throw out there, that is the team I see. So if I want to just fight with my 3*s at 266, that's fine. I won't see 550s but my rewards might not be as great. But it's too late for anything like that. That would break the system overall.

Again, I think a lot of this would be easier for noobs if they would just publish something about what to expect. It doesn't exist. Everything is patched together by tribal knowledge and previous posts. The communication gap, IMHO, is the biggest reason why new people quit. Not that they can't play their 2* Wolverine or they can't auto-fire. It's that the game makes no sense without lots and lots of extraneous work. Why make it so hard?

Heck, we can't even agree on what Ascension does and we've all been here 7+ years.

entrailbucket

The best way to solve the problem is relative balance, even though nobody seems to want that for whatever reason.

The game communicates to players that Wasp and Chasm are equal, because they both have 5 stars. How does a new player know which guys are better, without resorting to discord or the forum or whatever? If they have more stars, they're better!

Matchmaking assumes that Wasp and Chasm are roughly equivalent, that the Wasp player can generally beat the Chasm player, and vice versa.

What's breaking the system is the fact that there are these MASSIVE discrepancies in power within a tier. Fix that, and nobody has to worry about levelling up the "wrong" guy.

Scofie

@entrailbucket said:

@meadowsweet said:

@entrailbucket said:
Matchmaking is based on top 3, not top 1, so buying and levelling up that Iron Man won't immediately throw them to the wolves.
But what is the alternative here? How should it work?

It's probably not realistic, but I keep wondering how different would be if the game's PVP was based on an Elo or chess-style rating system: every player has a score / ranking. When you win, it goes up. When you lose, it goes down. The amount it goes up or down is based on your opponent's ranking (so an underdog upset matters more than a coin-toss competition between equals.)

Levels of top characters isn't a terrible approximation, but in reality someone who has Level 550 Chasm, Thor, Beta Ray Bill, Okoye, Hulk, etc. is not the same as someone whose only Level 550's are Wasp, Rescue, Loki, Archangel, Jeffrey, etc. The first player would have a much higher Elo (because they almost always win) while the second player's Elo would go down (because they often lose against teams with a better meta / synergy.)

So your match-making would be based on your actual performance, skill, or win-loss records. Not just based on the fact that you impulsively decided to level up Mysterio, Northstar, and Talos prematurely...

This is how it worked 10 years ago, when the game first launched!

We broke it IMMEDIATELY. What we'd do was join an event (at 0 points) then retreat 100 or so times straight (can't lose points when you don't have any!) until our win/loss record went into the red, and the game started showing us 1* teams. Another fun one was joining lightning rounds just to retreat over and over.

But things are different now. What If your MMR was set a an average of the previous season and the current one so far. So, yeah you can try and game it, but in order to get the PvP rewards, you actually have to win something, and that puts it back level. You could miss almost an entire season to skew it but by trying to earn those "lost" rewards, the current season performance puts you back where you should be. Consistent players see the same. Try harder, punch up and win, you'll see tougher opponents until you get back to your natural level.

Phumade

@meadowsweet said:

@entrailbucket said:
Matchmaking is based on top 3, not top 1, so buying and levelling up that Iron Man won't immediately throw them to the wolves.
But what is the alternative here? How should it work?

It's probably not realistic, but I keep wondering how different would be if the game's PVP was based on an Elo or chess-style rating system: every player has a score / ranking. When you win, it goes up. When you lose, it goes down. The amount it goes up or down is based on your opponent's ranking (so an underdog upset matters more than a coin-toss competition between equals.)

Levels of top characters isn't a terrible approximation, but in reality someone who has Level 550 Chasm, Thor, Beta Ray Bill, Okoye, Hulk, etc. is not the same as someone whose only Level 550's are Wasp, Rescue, Loki, Archangel, Jeffrey, etc. The first player would have a much higher Elo (because they almost always win) while the second player's Elo would go down (because they often lose against teams with a better meta / synergy.)

So your match-making would be based on your actual performance, skill, or win-loss records. Not just based on the fact that you impulsively decided to level up Mysterio, Northstar, and Talos prematurely...

ROFLCOPTR

Its worth mentioning that the game already employs elo scoring in matching. Your event points is your elo rating and the game establishes the matches point value based on an elo style formula.

Ultimately, who do you think MMR should be priotrizing? The biggest rosters are typically the most experienced and "skillful" players. Certainly, by move count or time per match, they are generally the fastest players in the game. Independent of how people view their playstlye/aggressiveness these are experienced and knowledgeable players.

FWIW, The top 25 leaderboard on both pvp and pve are a good measure of the skill, experience and dedication of those players. I really don't see names that shock me into thinking how they get that score?

Lets be realistic in assessing how a new player grows. a roster of 5* 450 champions is only viable to the 400-600 score range. Past 700, your going to be facing 480+ players with extensive game experience and knowledge.

entrailbucket

@Scofie said:
@entrailbucket said:

@meadowsweet said:

@entrailbucket said:
Matchmaking is based on top 3, not top 1, so buying and levelling up that Iron Man won't immediately throw them to the wolves.
But what is the alternative here? How should it work?

It's probably not realistic, but I keep wondering how different would be if the game's PVP was based on an Elo or chess-style rating system: every player has a score / ranking. When you win, it goes up. When you lose, it goes down. The amount it goes up or down is based on your opponent's ranking (so an underdog upset matters more than a coin-toss competition between equals.)

Levels of top characters isn't a terrible approximation, but in reality someone who has Level 550 Chasm, Thor, Beta Ray Bill, Okoye, Hulk, etc. is not the same as someone whose only Level 550's are Wasp, Rescue, Loki, Archangel, Jeffrey, etc. The first player would have a much higher Elo (because they almost always win) while the second player's Elo would go down (because they often lose against teams with a better meta / synergy.)

So your match-making would be based on your actual performance, skill, or win-loss records. Not just based on the fact that you impulsively decided to level up Mysterio, Northstar, and Talos prematurely...

This is how it worked 10 years ago, when the game first launched!

We broke it IMMEDIATELY. What we'd do was join an event (at 0 points) then retreat 100 or so times straight (can't lose points when you don't have any!) until our win/loss record went into the red, and the game started showing us 1* teams. Another fun one was joining lightning rounds just to retreat over and over.

But things are different now. What If your MMR was set a an average of the previous season and the current one so far. So, yeah you can try and game it, but in order to get the PvP rewards, you actually have to win something, and that puts it back level. You could miss almost an entire season to skew it but by trying to earn those "lost" rewards, the current season performance puts you back where you should be. Consistent players see the same. Try harder, punch up and win, you'll see tougher opponents until you get back to your natural level.

I don't understand what you're describing here, sorry. When you say "an average of the current season and the previous one" are you talking about points scored, or win/loss record, or something else?

If it's points scored, that gets messy pretty fast, in a whole bunch of different ways.

meadowsweet

@Scofie said:
@entrailbucket said:

@meadowsweet said:

@entrailbucket said:
Matchmaking is based on top 3, not top 1, so buying and levelling up that Iron Man won't immediately throw them to the wolves.
But what is the alternative here? How should it work?

It's probably not realistic, but I keep wondering how different would be if the game's PVP was based on an Elo or chess-style rating system: every player has a score / ranking. When you win, it goes up. When you lose, it goes down. The amount it goes up or down is based on your opponent's ranking (so an underdog upset matters more than a coin-toss competition between equals.)

Levels of top characters isn't a terrible approximation, but in reality someone who has Level 550 Chasm, Thor, Beta Ray Bill, Okoye, Hulk, etc. is not the same as someone whose only Level 550's are Wasp, Rescue, Loki, Archangel, Jeffrey, etc. The first player would have a much higher Elo (because they almost always win) while the second player's Elo would go down (because they often lose against teams with a better meta / synergy.)

So your match-making would be based on your actual performance, skill, or win-loss records. Not just based on the fact that you impulsively decided to level up Mysterio, Northstar, and Talos prematurely...

This is how it worked 10 years ago, when the game first launched!

We broke it IMMEDIATELY. What we'd do was join an event (at 0 points) then retreat 100 or so times straight (can't lose points when you don't have any!) until our win/loss record went into the red, and the game started showing us 1* teams. Another fun one was joining lightning rounds just to retreat over and over.

But things are different now. What If your MMR was set a an average of the previous season and the current one so far. So, yeah you can try and game it, but in order to get the PvP rewards, you actually have to win something, and that puts it back level. You could miss almost an entire season to skew it but by trying to earn those "lost" rewards, the current season performance puts you back where you should be. Consistent players see the same. Try harder, punch up and win, you'll see tougher opponents until you get back to your natural level.

Yeah, it would seem like there'd be some simple ways to identify the people who are "rope-a-doping" opponents by alternating between weak teams and strong teams and exclude that. Only use wins in the calculation, or longer-term averages, or median scores, whatever.

You could also use health as a proxy: if the winning team in a match is down to 10% of original team health with 2/3 characters KO'ed, that was a close match. If the 5★ team still has 99% of its health when it KO's the 2★ team... that was not a close match.

entrailbucket

@meadowsweet said:

@Scofie said:
@entrailbucket said:

@meadowsweet said:

@entrailbucket said:
Matchmaking is based on top 3, not top 1, so buying and levelling up that Iron Man won't immediately throw them to the wolves.
But what is the alternative here? How should it work?

It's probably not realistic, but I keep wondering how different would be if the game's PVP was based on an Elo or chess-style rating system: every player has a score / ranking. When you win, it goes up. When you lose, it goes down. The amount it goes up or down is based on your opponent's ranking (so an underdog upset matters more than a coin-toss competition between equals.)

Levels of top characters isn't a terrible approximation, but in reality someone who has Level 550 Chasm, Thor, Beta Ray Bill, Okoye, Hulk, etc. is not the same as someone whose only Level 550's are Wasp, Rescue, Loki, Archangel, Jeffrey, etc. The first player would have a much higher Elo (because they almost always win) while the second player's Elo would go down (because they often lose against teams with a better meta / synergy.)

So your match-making would be based on your actual performance, skill, or win-loss records. Not just based on the fact that you impulsively decided to level up Mysterio, Northstar, and Talos prematurely...

This is how it worked 10 years ago, when the game first launched!

We broke it IMMEDIATELY. What we'd do was join an event (at 0 points) then retreat 100 or so times straight (can't lose points when you don't have any!) until our win/loss record went into the red, and the game started showing us 1* teams. Another fun one was joining lightning rounds just to retreat over and over.

But things are different now. What If your MMR was set a an average of the previous season and the current one so far. So, yeah you can try and game it, but in order to get the PvP rewards, you actually have to win something, and that puts it back level. You could miss almost an entire season to skew it but by trying to earn those "lost" rewards, the current season performance puts you back where you should be. Consistent players see the same. Try harder, punch up and win, you'll see tougher opponents until you get back to your natural level.

Yeah, it would seem like there'd be some simple ways to identify the people who are "rope-a-doping" opponents by alternating between weak teams and strong teams and exclude that. Only use wins in the calculation, or longer-term averages, or median scores, whatever.

You could also use health as a proxy: if the winning team in a match is down to 10% of original team health with 2/3 characters KO'ed, that was a close match. If the 5★ team still has 99% of its health when it KO's the 2★ team... that was not a close match.

This is easy to say but very hard to do in practice. It's also going to be pretty trivial to game the system. For example, if health lost matters, why can't I just bring guys who hurt themselves or each other? What if I just drag out every fight purposely so I get hurt more?

Bowgentle

Basically, this community is EXTREMELY good at breaking PVP.
Which is how we ended up with layers upon layers upon layers of fixes for every exploit we ever found.
Which is also why we have zero confidence in the MMR rework they've promised for a year now.

Scofie

@entrailbucket said:

@Scofie said:
@entrailbucket said:

@meadowsweet said:

@entrailbucket said:
Matchmaking is based on top 3, not top 1, so buying and levelling up that Iron Man won't immediately throw them to the wolves.
But what is the alternative here? How should it work?

It's probably not realistic, but I keep wondering how different would be if the game's PVP was based on an Elo or chess-style rating system: every player has a score / ranking. When you win, it goes up. When you lose, it goes down. The amount it goes up or down is based on your opponent's ranking (so an underdog upset matters more than a coin-toss competition between equals.)

Levels of top characters isn't a terrible approximation, but in reality someone who has Level 550 Chasm, Thor, Beta Ray Bill, Okoye, Hulk, etc. is not the same as someone whose only Level 550's are Wasp, Rescue, Loki, Archangel, Jeffrey, etc. The first player would have a much higher Elo (because they almost always win) while the second player's Elo would go down (because they often lose against teams with a better meta / synergy.)

So your match-making would be based on your actual performance, skill, or win-loss records. Not just based on the fact that you impulsively decided to level up Mysterio, Northstar, and Talos prematurely...

This is how it worked 10 years ago, when the game first launched!

We broke it IMMEDIATELY. What we'd do was join an event (at 0 points) then retreat 100 or so times straight (can't lose points when you don't have any!) until our win/loss record went into the red, and the game started showing us 1* teams. Another fun one was joining lightning rounds just to retreat over and over.

But things are different now. What If your MMR was set a an average of the previous season and the current one so far. So, yeah you can try and game it, but in order to get the PvP rewards, you actually have to win something, and that puts it back level. You could miss almost an entire season to skew it but by trying to earn those "lost" rewards, the current season performance puts you back where you should be. Consistent players see the same. Try harder, punch up and win, you'll see tougher opponents until you get back to your natural level.

I don't understand what you're describing here, sorry. When you say "an average of the current season and the previous one" are you talking about points scored, or win/loss record, or something else?

If it's points scored, that gets messy pretty fast, in a whole bunch of different ways.

I'll admit I didn't put much thought into it. 🙂

It could be a combination of different things: win/loss ratio (but points average gained/lost, rather than actual losses?), placement from previous season, average strength of the top 10, rather than 3, Shield Rank etc. Each one of those factors could have a weighting, and you can see the 100 players above and below you in each slice etc.

My basic point was: I think it needs to consider longer term player data - you can break one event but sabotaging your whole season is really unlikely. There is so much player data you could analyse now that a dynamic system for calculating suitable opponents can be made with safeguards in. I get that players will try and break it, and at this stage of the game, I'd think the top players would probably enjoy trying to break an MMR system that constantly responded to their workarounds.

Scofie

@meadowsweet

You could also use health as a proxy: if the winning team in a match is down to 10% of original team health with 2/3 characters KO'ed, that was a close match. If the 5★ team still has 99% of its health when it KO's the 2★ team... that was not a close match.

I agree with this in principle but in the Class of 2023 event, I've finished with anywhere between 50% and 95% average health for similar levelled opponents. A lucky or unlucky cascade shouldn't be able to impact who you face next.

Ptahhotep

The devs have stats for the performance of each character across the entire player base. These could be used to give each character a rating when determining mmr, so Wasp would be lower than Chasm, for example . The only way to exploit it would be with a combination of weaker characters that are strong together. But that is arguably skilful play that should be rewarded anyway.

entrailbucket

@Scofie said:

@entrailbucket said:

@Scofie said:
@entrailbucket said:

@meadowsweet said:

@entrailbucket said:
Matchmaking is based on top 3, not top 1, so buying and levelling up that Iron Man won't immediately throw them to the wolves.
But what is the alternative here? How should it work?

It's probably not realistic, but I keep wondering how different would be if the game's PVP was based on an Elo or chess-style rating system: every player has a score / ranking. When you win, it goes up. When you lose, it goes down. The amount it goes up or down is based on your opponent's ranking (so an underdog upset matters more than a coin-toss competition between equals.)

Levels of top characters isn't a terrible approximation, but in reality someone who has Level 550 Chasm, Thor, Beta Ray Bill, Okoye, Hulk, etc. is not the same as someone whose only Level 550's are Wasp, Rescue, Loki, Archangel, Jeffrey, etc. The first player would have a much higher Elo (because they almost always win) while the second player's Elo would go down (because they often lose against teams with a better meta / synergy.)

So your match-making would be based on your actual performance, skill, or win-loss records. Not just based on the fact that you impulsively decided to level up Mysterio, Northstar, and Talos prematurely...

This is how it worked 10 years ago, when the game first launched!

We broke it IMMEDIATELY. What we'd do was join an event (at 0 points) then retreat 100 or so times straight (can't lose points when you don't have any!) until our win/loss record went into the red, and the game started showing us 1* teams. Another fun one was joining lightning rounds just to retreat over and over.

But things are different now. What If your MMR was set a an average of the previous season and the current one so far. So, yeah you can try and game it, but in order to get the PvP rewards, you actually have to win something, and that puts it back level. You could miss almost an entire season to skew it but by trying to earn those "lost" rewards, the current season performance puts you back where you should be. Consistent players see the same. Try harder, punch up and win, you'll see tougher opponents until you get back to your natural level.

I don't understand what you're describing here, sorry. When you say "an average of the current season and the previous one" are you talking about points scored, or win/loss record, or something else?

If it's points scored, that gets messy pretty fast, in a whole bunch of different ways.

I'll admit I didn't put much thought into it. 🙂

It could be a combination of different things: win/loss ratio (but points average gained/lost, rather than actual losses?), placement from previous season, average strength of the top 10, rather than 3, Shield Rank etc. Each one of those factors could have a weighting, and you can see the 100 players above and below you in each slice etc.

My basic point was: I think it needs to consider longer term player data - you can break one event but sabotaging your whole season is really unlikely. There is so much player data you could analyse now that a dynamic system for calculating suitable opponents can be made with safeguards in. I get that players will try and break it, and at this stage of the game, I'd think the top players would probably enjoy trying to break an MMR system that constantly responded to their workarounds.

Again, I think this might be easy to describe, but actually implementing it -- working out the specifics of it all -- would be quite difficult. I have a huge roster with 22 550s, but I don't really care about scoring points. Should I get easier matches than someone who scores 3000 points per event?

And again I'd ask, what problem (that currently exists) would this solve?

If the issue is that players level up the "wrong" character and are suddenly given matches they can't win, then the best fix for that is to make absolutely sure there are no "wrong" characters.

If the issue is that players want their PvP opponents to get easier as they level up their rosters...well, that's just not how Player vs Player works in competitive games.

Phumade

@Ptahhotep said:
The devs have stats for the performance of each character across the entire player base. These could be used to give each character a rating when determining mmr, so Wasp would be lower than Chasm, for example . The only way to exploit it would be with a combination of weaker characters that are strong together. But that is arguably skilful play that should be rewarded anyway.

This is an interesting concept, but I don't really think it address the fundamental complaint.

Giving MMR to individual chars is basically just ranking them by how well they work together within a specific pvp. From a transparency perspective, this is wonderful and is everything that forum posters have EVER wanted. A definitive and objective way to catalog how well a chars performs with a mandatory featured, or oddball ruleset)

But thats not really the fundamental complaint. To paraphrase, people don't want to be matched up against more experienced and better resourced players. Maybe using the words "experienced" and "better resourced" seems perjorative, but its simply reflecting experienced players know when, who and what teams to run etc.

FWIW, Once your score is over 1200, mmr is absolutely matching up the most experienced and best resourced players. Placement is fully decided by people making strategic choices on team composition, who they hit and how long they stay unshielded.

However, I suspect the experience between 500-900 is where the tension is most evident. this is the breakover point where the placement players vs 900 or (1200 and done) begin to separate.

My question for the forum is "what do you think MMR should prioritize?" In a perfect world, what does the T5 Leaderboard look like? What types of ques should you be seeing from 500-900?

Bad

I don't understand the concept of people leveling the wrong character, or some 5s being worse than others.
Nobody told me which characters were good or bad, I only had to play pvp consistently in order to figure it out.

Uninspired

From the original post, it sounds like it is more expensive to use shards to level an ascended character vs. a normal character, based on the second step leveling up Psylocke:

Leveling Up Ascended Characters & Ascending Again

...

How to complete a second Ascension for an Ascended 4-Star Psylocke to give her 5-Star rarity:

-On Ascended 4-Star Psylocke’s character information screen, select “Level Up”
-Use 3 Psylocke covers, or 1200 Psylocke shards, to level her up
-Repeat this process until she reaches max 4-Star Champion level, level 370

It costs 300 shards to add a level to a normal 3*. The example says you need either 3 covers or 1200 shards to add a level to a 3* ascended to 4* status. This is 400 shards per level (1200 / 3 = 400), meaning an extra 100 shards.

This is consistent with earlier posts saying the shard cost would be based on the stars of the ascended character and not the original basis, so it seems correct. However, this does suggest that you should never spend shards on an ascended character, but instead save them for the inevitable 3* dupes you will build. Am I missing something?

Bowgentle

@Uninspired said:
From the original post, it sounds like it is more expensive to use shards to level an ascended character vs. a normal character, based on the second step leveling up Psylocke:

Leveling Up Ascended Characters & Ascending Again

...

How to complete a second Ascension for an Ascended 4-Star Psylocke to give her 5-Star rarity:

-On Ascended 4-Star Psylocke’s character information screen, select “Level Up”
-Use 3 Psylocke covers, or 1200 Psylocke shards, to level her up
-Repeat this process until she reaches max 4-Star Champion level, level 370

It costs 300 shards to add a level to a normal 3*. The example says you need either 3 covers or 1200 shards to add a level to a 3* ascended to 4* status. This is 400 shards per level (1200 / 3 = 400), meaning an extra 100 shards.

This is consistent with earlier posts saying the shard cost would be based on the stars of the ascended character and not the original basis, so it seems correct. However, this does suggest that you should never spend shards on an ascended character, but instead save them for the inevitable 3* dupes you will build. Am I missing something?

It doesn't really matter.
Eventually you'll need to level the ascended chars and spend covers and shards there.

DAZ0273

If anybody does get matched up with a level 550 Iron Man, let us know if the AI ever manages to fire his blue power. I would put Spider Woman red odds on that one.

Pantera236

@Uninspired said:
From the original post, it sounds like it is more expensive to use shards to level an ascended character vs. a normal character, based on the second step leveling up Psylocke:

Leveling Up Ascended Characters & Ascending Again

...

How to complete a second Ascension for an Ascended 4-Star Psylocke to give her 5-Star rarity:

-On Ascended 4-Star Psylocke’s character information screen, select “Level Up”
-Use 3 Psylocke covers, or 1200 Psylocke shards, to level her up
-Repeat this process until she reaches max 4-Star Champion level, level 370

It costs 300 shards to add a level to a normal 3*. The example says you need either 3 covers or 1200 shards to add a level to a 3* ascended to 4* status. This is 400 shards per level (1200 / 3 = 400), meaning an extra 100 shards.

This is consistent with earlier posts saying the shard cost would be based on the stars of the ascended character and not the original basis, so it seems correct. However, this does suggest that you should never spend shards on an ascended character, but instead save them for the inevitable 3* dupes you will build. Am I missing something?

They misspoke, you only need 900 shards to level up a 3Psylocke4.

bbigler

Is that true? They said shard costs would be the same in each tier regardless if they’re ascended or not.