Matchmaking question

In the announcement, @joel said that EMH would be based on roster power and league. Whose roster power is he referring to? The attacker's or defender's roster power?

I would have thought it should be the defender's roster power but I'm hearing otherwise now. If I'm in legend 1 I would think that I would see as an easy opponent as someone who has a low roster power based on their algorithm. My own roster power I wouldn't think should affect who I'm seeing as opponents. So for example, if two people are both in legend 1 and one has a 400k roster power and the other had 200k roster power, shouldn't they both be seeing similar opponents?
«1

Comments

  • edited May 2
    Foozle wrote: »
    In the announcement, @joel said that EMH would be based on roster power and league. Whose roster power is he referring to? The attacker's or defender's roster power?

    I would have thought it should be the defender's roster power but I'm hearing otherwise now. If I'm in legend 1 I would think that I would see as an easy opponent as someone who has a low roster power based on their algorithm. My own roster power I wouldn't think should affect who I'm seeing as opponents. So for example, if two people are both in legend 1 and one has a 400k roster power and the other had 200k roster power, shouldn't they both be seeing similar opponents?

    It is based off the defender's roster power. Your example is correct. You would both see similar opponents.

    Last night while raiding I saw 2 single hero defenses in my hard slot XD Looked like they were from accounts that had put up the defense during organized pvp token event and never changed back.
  • This is not accurate. Matchmaking is based on "Calculated Roster Power" (CRP). CRP is a "secret" formula use to give every player an arbitrary value to be used to match you up against a similar opponent. CRP is derived from 10-20 heroes on your roster. From my experiments it appears to include Icebloom, Nitpick, Goblins, Undead, and many more. A players CRP is used to find opponents with a similar CRP.

    This is why players who are level 65 with no epics are able to get so high on the leaderboards. Their (Hard) opponents are laughably easy and can perform very long streaks with no effort other than gems and time.
  • edited May 2
    This is not accurate. Matchmaking is based on "Calculated Roster Power" (CRP). CRP is a "secret" formula use to give every player an arbitrary value to be used to match you up against a similar opponent. CRP is derived from 10-20 heroes on your roster. From my experiments it appears to include Icebloom, Nitpick, Goblins, Undead, and many more. A players CRP is used to find opponents with a similar CRP.

    This is why players who are level 65 with no epics are able to get so high on the leaderboards. Their (Hard) opponents are laughably easy and can perform very long streaks with no effort other than gems and time.

    Everything Eej has said in the summit goes against this. I'm quite sure I'm correct on this, and don't really feel like going through the massive summit log to find the quotes. I'll explain how matchmaking works though.

    A pool of available opponents is taken based purely on your level (And potentially now ranking as some tweaks have been made).

    That pool of available opponents is split into bottom third, middle third, top third based off their "calculated roster power". A random opponent from bottom third is easy, random opponent from middle is medium, random oppoent from top is hard.

    They are looking at tweaking this further to tighten hard matchmaking for Legend I.

    Edit: Level 65s can get so high on the leaderboards because they are seeing matchmaking based on their level alone.

    @Eej I try not to tag you often, but it would be great to have some confirmation here. If there's confusion with Foozle/Jax/Myself then there's probably confusion with almost everyone.
  • Humor me. Unequip runes from Icebloom, your undead, Nitpick and your goblins. Fight 1 match. Now go and fight 10 matches and tell me you don't notice a HUGE drop in your opponents difficulty.
  • edited May 2
    I really hope this isn't true : /

    Using PvE heroes to calculate roster power for PvP seems beyond illogical. Not to mention, giving people different opponents because their calculated roster power is weaker also makes zero sense from a competitive stand point.

    Hopefully Eej will step in and clarify. If what you are saying it true... well it's completely absurd that it would work that way. Undead/Goblins have very little correlation to someones pvp roster strength.

    Edit: There is of course a chance patterns are being read where they don't exist. Ie: Icebloom runes unequipped followed by a series of easy opponents without any actual correlation. Sometimes I'll get a slew of easy opponents and then a series of difficult ones. (I often get the most difficult opponents early in the season and I think this has to do with the tweaking they did to matchmaking where your current tier affects your available opponents. Or more your current tier affects you can be offered to as an opponent.

    Either way, I know your statistics are pretty meticulous. I'll try to look into this further if we don't get clarification from Eej.

  • Matt I don't think from reading Summit that Eej did answer this as you suggest - I felt that someone needed to push more specifically on the attacker or defender part (I wasn't in summit at this point).

    Obviously we all pretty much agree that basing it on the attackers runes seems unfair but based on my own opinion experience at 290k roster power and comments from the likes of Nomadz on how much harder his choices are I believe Jax is correct.

    EEJ did however bring up at the last summit a tightening of raid choices on hard option at legend 1 which for top end pvp is a very good thing - hopefully this will help a lot.
  • Humor me. Unequip runes from Icebloom, your undead, Nitpick and your goblins. Fight 1 match. Now go and fight 10 matches and tell me you don't notice a HUGE drop in your opponents difficulty.

    Or just pvp on my account - same difference! @290k rp but essentially with a pvp roster broadly as strong as anyone I can attest to pvp being a little on the easy side!
  • edited May 2
    ace_1650cp wrote: »
    Matt I don't think from reading Summit that Eej did answer this as you suggest - I felt that someone needed to push more specifically on the attacker or defender part (I wasn't in summit at this point).

    Obviously we all pretty much agree that basing it on the attackers runes seems unfair but based on my own opinion experience at 290k roster power and comments from the likes of Nomadz on how much harder his choices are I believe Jax is correct.

    EEJ did however bring up at the last summit a tightening of raid choices on hard option at legend 1 which for top end pvp is a very good thing - hopefully this will help a lot.

    I'm also taking some references from numerous blue posts that were made when matchmaking initially came out that were buried in later threads, and using some basic logic on what would be completely impractical. Of course DB has done things that I considered completely illogical before (And if what Jax is saying is true... this would be another case).

    Edit: The low/middle/top third was specifically based on a blue post made on a random thread.
  • Yeah I think you are right from what has been said and I read all the same stuff Matt.
  • I don't know. My roster power isn't insignificant (nowhere near @NomadzReaper), and for the most part, my choices haven't been terribly hard. I haven't stripped any runes. So, I am torn on what to think.
    nunya
  • My level 68 account (way down in lower leagues due to playing it only for crowns) sees a similar selection of opponent power in e/m/h as my top 100 account. The actual matchups are different--I believe due to different leagues--but the overall range of power is close enough that I doubt I would be able to tell if you switched the matchups between the two accounts for a day. My 68 is not even close to maxed out for its level, and as far as I can tell it is not getting a break with easier matchmaking.
  • D_Rodd wrote: »
    My level 68 account (way down in lower leagues due to playing it only for crowns) sees a similar selection of opponent power in e/m/h as my top 100 account. The actual matchups are different--I believe due to different leagues--but the overall range of power is close enough that I doubt I would be able to tell if you switched the matchups between the two accounts for a day. My 68 is not even close to maxed out for its level, and as far as I can tell it is not getting a break with easier matchmaking.

    Do you have any epics on your 68 account?
  • Yikes, I'm working on epics for my undead team and they're runed, but my two favorite pvp teams, heck, they just made 70 and I have yet to rune them, let alone epics, which may take forever.

    Bottom line; the team I choose for pvp in a first-round-to-defender environment is entirely different than the team I use for PVE (sustain/outlast). I sure hope that's not why my opponents keep walloping me.

    I'm also disappointed that the so-called E choice is worth so little. I'm an avg lvl 70 player and I have no expectation to be up in the top 100, I only want to earn enough honor to buy things in the honor shop, and some of the things I want require champion league.

    Pvp should be fun for player like me, not full of anxiety...picking either the E choice and progressing slower than a snail or picking a harder choice and banging my head against the wall.

    Thank GOODNESS honor is kept over seasons now, at least I can get what I want in 2 more seasons from now.


    (I haven't begun to rune or epic a goblin team yet - its on my list of todo's.)
  • Foozle wrote: »
    Do you have any epics on your 68 account?

    Only 4 - Torchy, Igo, MK, and Hansuke. None above 7. The whole roster is 274k RP, mostly from stars.

    To me the biggest difference in matchmaking difficulty seems to be between the beginning of the season and the end. For whatever reason, the matchups seem to get easier for both accounts. If I were to hypothesize:

    - on my level 68, matchups may get easier as the week goes on because the pool of potential matches excludes more powerful players who have already advanced to higher leagues

    - on my top 100, matchups may get easier because the pool of potential matches includes less powerful players who have moved up to become available

    I agree the exact details of the matchmaking mechanisms aren't clear, so I would also be interested in further clarification.

  • Down at level 42, I see some way easy teams in the hard slot, but most other times I see teams that have such powerful runes I don't even try. I can rarely beat the Hard slot.

    My roster is pretty strong for my level but, my runes are only average. I have not runed any Undead, or Icebloom, but my goblins have mostly lesser and few improved runes. No greater Runes. I do have nitpick runed average. I do have a pretty good MK for my level but there are better.

    I thought I read that they would be looking at 4-8 heroes to determine your match up? But I am thinking that is not how it works.
    I tend to think that they look at many more than that. I think I am "penalized" do to my deep bench rather than having just a few super strong heroes.

    Not sure. But I regularly see opponents in the hard slot that have teams with 25% more team power than I can bring. if I can bring 10.5K team power I am seeing hard teams in the 13k-14k power range. I can't beat them. They are sometimes 5 levels higher than me with all greater runes. No contest. Most matches down here are a contest of who has the strongest MK. Whoever's MK survives the first round wins.

    When my defense gets beat, I look at the revenge option and usually these guys use the same 4 heroes on defense as offense and after reviewing their profile, they have no good heroes after the 4 prime PVP ones. It's as if they have a roster of all 2-3 star heroes and 4 heroic heroes with all greater runes. This keep their compared roster power down and they have 4 that can wipe the map with anything. Hardly seems like a good system. I guess any system can be gamed and taken advantage of when you know how it works.

    Just my thoughts. Still collecting mental data from match ups and wins and losses.
    litespeed
  • edited May 3
    Jax has an interesting point.
    I may unrune all my useless heroes a season and see what happens.
    Haven't used undead or goblins in awhile and I'm sure I can handle losing my icebloom.
    I just a hit 300kRP and I've been seeing my own guild members as well as other top players with around 400K RP as my middle and right options since the PvP change.

    What's the worst that can happen I don't use those heroes as much anyway.

    Kinda sad if this turns out to be true.
    As I equally level/rune/star my entire roster since day 1
    Follow me on Twitch @Auraofsmiles
  • As we get closer to the end of the season a new matchmaking mechanic kicks in. In order to keep everything "balanced" everyone will be attacked the same number of times per league. So for those of us that rush out to Legend we'll see the number of attacks we receive steadily increase for the first few days. Once we get to around 30 defenses ... all attacks stop. Friday, Saturday, Sunday we typically get zero attacks. The reason for this is that new players are entering Legend and having to play "catch-up" and get their 30 defenses so that everything remains "balanced." (Extremely dumb imo because doing anything creative or strong on defense is a complete waste of time and effort.)

    Anyway ... as we get to Monday and Tuesday it becomes crunch-time for those who do not PvP as much as the hardcore folks. This results in the less skilled "Easier" teams now finally reaching Legend. For the PvP'ers who rushed out to Legend early and PvP a lot, this results in Monday and Tuesday being very easy days since our matchups are significantly easier. All the less hardcore PvP'ers who are finally reaching Legend need to get their 30 defenses so everything stays balanced resulting in easier matchups for the hardcore PvP'ers.
  • Yep, that sounds accurate according to my experience, Jax. Thanks for adding some extra details to support my hypothesis above.

    However, it does not support your assertion at the top of this thread that matchmaking is overall easier for the sub-70s. On what evidence do you base that conclusion?
  • edited May 3
    Based on EEJ's comments in the summit about Calculated Roster Power being used for matchmaking purposes and that it was a subset of heroes on your roster. From that knowledge I proceeded to start testing un-runing different heroes attempting to guess which heroes had an impact. I concluded that Icebloom, Undead, Nitpick and Goblins included at least a few of the heroes that add up your Calculated Roster Power.

    After making this recommendation to 3 people and seeing that it DID have significant impact on their (H) PvP choice I felt this was a solid hypothesis. Then this previous week we had "Aether Weekend" where I needed to heavily rune Yasmin, Nubnub, Astrid, Dagrund and Leo. I totally forgot I had done this, but when I went to PvP I was shocked at how hard my (H) PvP matches were. Solid 30k+ Power teams back to back. I went back and un-runed these heroes and quickly my (H) PvP queue went back to normal. Normal being random opponents ranging from 20k up to 30k. Huge difference having 1/3 opponents be extremely challenging vs. every opponent being extremely challenging.

    And before anyone starts calling me a cheater ... realize the first thing I did with this knowledge was share it with everyone so that we can make the system better.
  • Thank you for all your research @JaxBoomstick
    nunya
  • Based on EEJ's comments in the summit about Calculated Roster Power being used for matchmaking purposes and that it was a subset of heroes on your roster. From that knowledge I proceeded to start testing un-runing different heroes attempting to guess which heroes had an impact. I concluded that Icebloom, Undead, Nitpick and Goblins included at least a few of the heroes that add up your Calculated Roster Power.

    After making this recommendation to 3 people and seeing that it DID have significant impact on their (H) PvP choice I felt this was a solid hypothesis. Then this previous week we had "Aether Weekend" where I needed to heavily rune Yasmin, Nubnub, Astrid, Dagrund and Leo. I totally forgot I had done this, but when I went to PvP I was shocked at how hard my (H) PvP matches were. Solid 3k+ Power teams back to back. I went back and un-runed these heroes and quickly my (H) PvP queue went back to normal. Normal being random opponents ranging from 2k up to 3k. Huge difference having 1/3 opponents be extremely challenging vs. every opponent being extremely challenging.

    And before anyone starts calling me a cheater ... realize the first thing I did with this knowledge was share it with everyone so that we can make the system better.

    If what you are saying is correct, the system is inherently broken and this would be yet another sad example of something failing that seems so intuitively obvious.

    Keep hoping for some blue clarification!
  • Thanks for the details, @JaxBoomstick. If it does turn out to be the case that unruning heroes would result in easier matchups being offered, that could pose a significant issue with the matchmaking imo. I do not believe the game should penalize increasing the power of one's roster.

    And I applaud you for sharing what could be considered a potential significant advantage. Thank you for your honesty and positive contribution.
  • Personally I wouldn't call the system inherently broken. I'd say its a solid compromise between folks who want PvP to be easy and folks who want it challenging. The flaw in the system is with the Top 100. The Top 100 should not follow the same mechanics as league play. The Top 100 should not be "balanced" It should be winner take all, battle royale.

    Upon entering Top 100 your queue should be no (E), no (M) and only (H) choices who include anyone in the Top 100. No balance against who has been attacked more vs. less ... if you have a weak defense you should get hammered.
  • edited May 3
    Based on EEJ's comments in the summit about Calculated Roster Power being used for matchmaking purposes and that it was a subset of heroes on your roster. From that knowledge I proceeded to start testing un-runing different heroes attempting to guess which heroes had an impact. I concluded that Icebloom, Undead, Nitpick and Goblins included at least a few of the heroes that add up your Calculated Roster Power.

    PVP suggestion for DB (@Joel @Eej) so would it be better if they just took the total power of ALL your heroes (calculated when going into PVP at that exact moment to account for all currrent equipped runes) and used that to get your Calculated Roster Power, which then matches to get your opponents? Or would that be a disaster?

    What do my fellow gamers think?

    Curious here...
    Officer - Level 67 - "The Bastion"
    http://bastion-db.boards.net/
  • So, basically PvE Heroes are used to determine PvP strength...smh lol. I love seeing the 30k UD teams though aka free win.
    IGN & Line: N00BST4R - Final Boss
  • edited May 3
    sirolk99 wrote: »
    Based on EEJ's comments in the summit about Calculated Roster Power being used for matchmaking purposes and that it was a subset of heroes on your roster. From that knowledge I proceeded to start testing un-runing different heroes attempting to guess which heroes had an impact. I concluded that Icebloom, Undead, Nitpick and Goblins included at least a few of the heroes that add up your Calculated Roster Power.

    PVP suggestion for DB (@Joel @Eej) so would it be better if they just took the total power of ALL your heroes (calculated when going into PVP at that exact moment to account for all currrent equipped runes) and used that to get your Calculated Roster Power, which then matches to get your opponents? Or would that be a disaster?

    What you my fellow gamers think?
    Curious here...

    You're asking for a significant amount more server side work for considerably less accurate matchmaking.
  • You're asking for a significant amount more server side work for considerably less accurate matchmaking.
    Ahh...
    Officer - Level 67 - "The Bastion"
    http://bastion-db.boards.net/
  • Obviously a sample size of 1 isn't hard data, but my experience supports Jax's point.

    I'll preface this with the fact that I suck at PVP, so even my 'after' results will not impress most of you. I hate rune crafting, preferring the certainty of epic upgrades, so I was really outclassed with the new matchmaking. I only won about 85% of matches against my easy opponents, and had won maybe 2 hard matches total in the first few seasons.

    My prior record long win streak was 12.

    Immediately after spending a few million gold removing runes from heroes I never use, I went on a win streak of 35, only stopping because I got bored. I now have another streak going of 16.

    I won more trophies in the last 24 hours, than in all prior seasons combined.

    I'm now really questioning whether I should ever upgrade epics for non essential heroes, which is probably not what the devs were going for.
  • So basically remove runes off all the heroes you don't use, and we'll get way easier opponents in matchmaking and can climb the leaderboard easier. Sweet! Lmao! ;)
    Officer - Level 67 - "The Bastion"
    http://bastion-db.boards.net/
  • edited May 8
    I agree with Jax the Top 100 should be a special league on its own that is extremely cutthroat. There shouldn't be any "dumbing down" once you reach this tier. Top 100 should all be fighting each other and maybe expand it to the top 200 just for some flavor.

    I have been hearing for a while that you could lower your roster power to increase your winrate by having easier opponents, but could never practice this because (1) it would diminish my personal pvp experience and (2) I wouldn't want to hurt my guild's roster power score just for my personal gain. If this is actually being done it should be quickly fixed as no one should be able to manually reduce the difficulty of their pvp matchups, unless somehow it correlated with the amount of trophies earned as well.
«1
Sign In or Register to comment.

© 2015 Big Fish Games. Inc., Big Fish, the Big Fish logo, and Dungeon Boss are
trademarks of Big Fish Games, Inc., used with permission www.bigfishgames.com