Tonight I decided to switch to a Razakus Priest deck, since based on the outcome of my earlier experiments, it's reasonable to guess that a strategy of picking a more skill-intensive deck might be the best recipe for ladder success if I can do well enough to exceed the average.
I played 18 games (which is of course a small number from which to draw any statistical inferences, so take this with a grain of salt.) Interestingly, of these games, 8 of them (44%) were against other priests. All but one of them were Dragon or Big Priest, which according to Vicious Syndicate's Data Reaper Report, are favored against Razakus Priest. (Dragon Priest, according to them, has about a 55% win rate vs. Razakus, and Big Priest has about a 51% win rate, on average.)
My overall win rate for the 18 games was about 72%. (I went 13-5.) Fortunately for me, against those priest archetypes that were favored by the population numbers, I went 6-2. It's certainly possible to have a good record through just dumb luck, so I won't infer too much about the deck being successful for me just yet.
The important carry-away, though, is that 44% of my opponents were priests, almost all of them archetypes that were supposed to be favored against me, almost twice the representation of priests in the overall rank 5-1 population, By comparison, my second largest group of opponents, warlock, were 22%, not too far off from their 17% representation in the population. The remaining classes were only 1 or 2 games each.
The important carry-away, though, is that 44% of my opponents were priests, almost all of them archetypes that were supposed to be favored against me, almost twice the representation of priests in the overall rank 5-1 population, By comparison, my second largest group of opponents, warlock, were 22%, not too far off from their 17% representation in the population. The remaining classes were only 1 or 2 games each.
I'm guessing your winrate > 50% in general? If 50% WR is the mmr's target, your observation may be spot on. Definitely difficult to have a solid conclusion with a small sample size, but it seems that's a significant design of their algorithm.
I used meme/fun decks yesterday, and I got paired against meta decks and kept losing for 7 or so matches before I encountered a non-meta deck. The one observation that was obvious were the number of gold players I fought. I should probably take notes on this, but I'd say more than 2/3 of my 12 or so matches were against those players. Time spent (and overall wins since starting to play?) is correlated to pairing.
Regarding non-Legend ranked play, Blizzard's most recent statement is that they randomly match players of the same star rank (i.e. no such thing as MMR in non-Legend ranked play). If the idea is that you add Eater of Secrets and - wham! - no more Mages or Hunters, then Blizzard would have to be straight up lying about how they match players. I've known them to be coy or slippery in how they word things, but not too much straight up lying. My trust on this topic could be misplaced, but a lot of players will be very angry if it comes out that they've added an MMR outside of Legend and Casual play.
Regarding non-Legend ranked play, Blizzard's most recent statement is that they randomly match players of the same star rank (i.e. no such thing as MMR in non-Legend ranked play). If the idea is that you add Eater of Secrets and - wham! - no more Mages or Hunters, then Blizzard would have to be straight up lying about how they match players. I've known them to be coy or slippery in how they word things, but not too much straight up lying. My trust on this topic could be misplaced, but a lot of players will be very angry if it comes out that they've added an MMR outside of Legend and Casual play.
To give some defense we do have to remember that it's Activision/Blizzard who also works with Bungie on Destiny 2 which...yeah.. is bad.
The statement is that Activision isn't like EA and thus lets developers work mostly on their own which then leads into Blizzard still being Blizzard and Bungie being a major cause of the Destiny issues.
So the question is whether that's true or whether Activision has more control over the situation over at Destiny 2. If they do then it would stand to reason that they also have more say over at Blizzard and, thus, aren't as trustworthy as they used to be.
The reasoning is there. The problem is there's no evidence to support it yet which is why we don't need tin foil hat conspiracies or outright dismissal. What we need is testing....which OMG this thread is actually doing!
so yeah.. keep it up guys. MOAR tests. Would like to see what happens when you switch to a deck that's dominate against Dragon and Big Priest.
I totally agree with your skepticism about the idea that they'd speak falsely. There's no good reason for them to do so. However, it has been a year since they came out with a few public statements that only ranking is used in sub-legend ranked matchmaking, and there have been more recent statements that both implied a desire to explore new directions and suggest that they've possibly been doing so without talking about it publicly.
Of course, if you are aware of comments they've made much more recently than the beginning of 2017 that reinforce that that's still the case, that would be great to know about in this discussion. I haven't found them, myself.
I totally agree with your skepticism about the idea that they'd speak falsely. There's no good reason for them to do so. However, it has been a year since they came out with a few public statements that only ranking is used in sub-legend ranked matchmaking, and there have been more recent statements that both implied a desire to explore new directions and suggest that they've possibly been doing so without talking about it publicly.
Of course, if you are aware of comments they've made much more recently than the beginning of 2017 that reinforce that that's still the case, that would be great to know about in this discussion. I haven't found them, myself.
My current belief is that they have been running the system on MMR/Star rank matching with no rigging and have been seeing the public stream in a fury because of it. Thus I believe they ARE going to start rigging the system and are working on a new more rigged system soon. They'll tell us about it as they add it in similar to how they've been rigging up the arena system.
Basically what happened with Apple and the 'random' option. Because people don't really want RANDOM, they want Synergy. They don't want a 50% chance to get heads, they want 5 heads and 5 tails in generally regular intervals with just enough random to be interesting. Blizzard is doing just that.
I WOULD follow it with "Then they'll announce it and we'll eat it up and love it." But then Star Wars Battlefront happened and...I'm not sure HOW the community will accept it. It might be best not to bother, at least until other companies stop being stupid and rattling the hornet's nest).
So the question is whether that's true or whether Activision has more control over the situation over at Destiny 2. If they do then it would stand to reason that they also have more say over at Blizzard and, thus, aren't as trustworthy as they used to be.
Players have characterized the Destiny 2 XP scaling thing as dishonest, but I don't in any way agree. Lack of transparency of a system is not the same as offering a comment that misstates how it works.
Same with Hearthstone matchmaking. That I know of, Blizzard hasn't commented on how the matchmaking works since early 2017, and they HAVE been very clear that they want to make some changes to (among other things) make the experience better for newer players. Saying something a year ago that was accurate then is not a promise to leave the system untouched forever after.
As far as having a profit motivation in building game systems: Designers of free-to-play games often talk about the "monetization layer," which are systems that encourage actually pulling one one's wallet and spending money, the "retention layer," which are systems like quests that encourage players to return to the game frequently, and the "base" or "gameplay layer" which is the core game itself. Most players (and game designers for that matter) feel that it's just fine to build the monetization layer to earn money, and that it's pretty scuzzy to tweak the gameplay layer to force spending money (which results in accusations that games are "pay-to-win.")
There's a lot more disagreement about what's ok to do in the retention layer, which (in the case of Hearthstone) includes its matchmaking systems, and (in the case of Destiny 2) includes its XP system. Everything I've seen (including, if true, what's being discussed in this thread) suggests that encouraging spending has not been a major motivation for most of Hearthstone's retention layer design. It's pretty conservative. If anything, a design decision to try to reduce the win rate impact of pure choice of deck would be about keeping the game fun and varied by making more decks viable. (This, incidentally, reduces the necessary spending to keep up with the game if you're happy playing just one or a couple of deck archetypes.)
Regarding the relationship between Blizzard and Activision, remember that the merger took place because Blizzard was able to use the enormous pile of money minted by World of Warcraft to acquire Activision and thus enable their private investors to convert their company's stock to cash. When Blizzard bought Activision in their reverse-merger, made Activision the parent company to get on the public stock markets, and renamed the parent company Activision-Blizzard (which, by the way, is how it went down), the companies placed in writing that Blizzard would retain complete creative control over their own development.
As much as people like to whisper about the impact that Activision as a publisher is having on Blizzard's games, it's very likely close to zero, other than that they both live in the modern game development world that we all do, which has been heavily shaped by the free-to-play phenomenon, and that their shared ownership aligns their goals naturally.
I WOULD follow it with "Then they'll announce it and we'll eat it up and love it." But then Star Wars Battlefront happened and...I'm not sure HOW the community will accept it. It might be best not to bother, at least until other companies stop being stupid and rattling the hornet's nest).
Honestly, we're in All bets are off territory.
Are you thinking they're trying to devise an "anti-tilt" system to keep people playing? It is a known fact that casinos do this (detect a string of losses on your card then suddenly you get a hand delivered "loyal customer" gift card with a little casino credit). A lot of people despise Ice Block, just to give one example, and it would be easy for Blizzard to detect tilt caused directly by this card. Same with Cube Warlock - suddenly people put in double Spellbreaker and you can tell they hate Control Warlock.
I mentioned my overall win rate in the data set that post you quoted was describing. It was 72%. But, and this is important, I do not necessarily believe that matchmaking changes what it's doing based on my own, actual, real-time win rate (though I should be careful to say I don't actually know this.) I believe it makes predictions based on population-wide variances. This is supported by the battle.net server engineer's comments that I quoted on the previous page about feeding business intelligence data (which would be population-wide statistics about the game) into Hearthstone's matchmaking.
The point of my post was that I was being matched against a mix of opponents that ought to hit about a 50% win rate based on population-wide statistics about common matchups.
Are you thinking they're trying to devise an "anti-tilt" system to keep people playing?
I'm not sure what he's thinking, but I don't think you need something as active as what you describe to achieve the same overall result.
What I personally think would be a likely design, based on the battle.net engineer's comments from the last page, would be one that combines automated archetype-matching (like that being done by Vicious Syndicate and HSReplay in their statistics) with periodic static data dumps of how those archetypes fare against other archetypes. The matchmaking system then would use some type of weighting to make certain match-ups proportionally more or less likely so that each base archetype is as close to 50% as possible, population-wide.
It's enough to try to give a typical player a 50% win rate. An active system that notices that player X has won a bunch, let's make her lose, has the problem that it will cancel out the effect of differences in individual player skill, which is very contrary to the basic goal of ranked matchmaking (which is to allow that skill to be recognized.)
Rollback Post to RevisionRollBack
To post a comment, please login or register a new account.
Tonight I decided to switch to a Razakus Priest deck, since based on the outcome of my earlier experiments, it's reasonable to guess that a strategy of picking a more skill-intensive deck might be the best recipe for ladder success if I can do well enough to exceed the average.
I played 18 games (which is of course a small number from which to draw any statistical inferences, so take this with a grain of salt.) Interestingly, of these games, 8 of them (44%) were against other priests. All but one of them were Dragon or Big Priest, which according to Vicious Syndicate's Data Reaper Report, are favored against Razakus Priest. (Dragon Priest, according to them, has about a 55% win rate vs. Razakus, and Big Priest has about a 51% win rate, on average.)
My overall win rate for the 18 games was about 72%. (I went 13-5.) Fortunately for me, against those priest archetypes that were favored by the population numbers, I went 6-2. It's certainly possible to have a good record through just dumb luck, so I won't infer too much about the deck being successful for me just yet.
The important carry-away, though, is that 44% of my opponents were priests, almost all of them archetypes that were supposed to be favored against me, almost twice the representation of priests in the overall rank 5-1 population, By comparison, my second largest group of opponents, warlock, were 22%, not too far off from their 17% representation in the population. The remaining classes were only 1 or 2 games each.
Regarding non-Legend ranked play, Blizzard's most recent statement is that they randomly match players of the same star rank (i.e. no such thing as MMR in non-Legend ranked play). If the idea is that you add Eater of Secrets and - wham! - no more Mages or Hunters, then Blizzard would have to be straight up lying about how they match players. I've known them to be coy or slippery in how they word things, but not too much straight up lying. My trust on this topic could be misplaced, but a lot of players will be very angry if it comes out that they've added an MMR outside of Legend and Casual play.
One does not simply walk into Mordor,
unless they want to be the best they can be.
LardLad:
I posted a pretty long discussion of the history of Blizzard's comments on matchmaking here.
I totally agree with your skepticism about the idea that they'd speak falsely. There's no good reason for them to do so. However, it has been a year since they came out with a few public statements that only ranking is used in sub-legend ranked matchmaking, and there have been more recent statements that both implied a desire to explore new directions and suggest that they've possibly been doing so without talking about it publicly.
Of course, if you are aware of comments they've made much more recently than the beginning of 2017 that reinforce that that's still the case, that would be great to know about in this discussion. I haven't found them, myself.
One does not simply walk into Mordor,
unless they want to be the best they can be.
yambs_tim:
I mentioned my overall win rate in the data set that post you quoted was describing. It was 72%. But, and this is important, I do not necessarily believe that matchmaking changes what it's doing based on my own, actual, real-time win rate (though I should be careful to say I don't actually know this.) I believe it makes predictions based on population-wide variances. This is supported by the battle.net server engineer's comments that I quoted on the previous page about feeding business intelligence data (which would be population-wide statistics about the game) into Hearthstone's matchmaking.
The point of my post was that I was being matched against a mix of opponents that ought to hit about a 50% win rate based on population-wide statistics about common matchups.