Whats up with people saying Razakus and Cubelock are the best decks in the game, and yet VS shows statistics from thousands of matches that show otherwise? I'm not saying they are right, but this really grinds my gears
Whats up with people saying Razakus and Cubelock are the best decks in the game, and yet VS shows statistics from thousands of matches that show otherwise? I'm not saying they are right, but this really grinds my gears
You know what grinds my gears? People that use statistics as evidence of something without using their own original thoughts to comprehend them. Why post at all if you won't try to add your own input? You could just make a thread and post the link without saying anything.
Did you play when Patron Warrior was the best deck in the game? Did you know it had a sub-50% win rate?
Look at how many people in the world championship are bringing warlock and priest. Are they all idiots?
Just try thinking about it for a second. Maybe it will come to you.
I personally believe that highlander priest, murloc paladin, control warlock, and tempo rogue add up to the best decks in the game and the win rate differences between them are very small at the top level.
Whats up with people saying Razakus and Cubelock are the best decks in the game, and yet VS shows statistics from thousands of matches that show otherwise? I'm not saying they are right, but this really grinds my gears
You know what grinds my gears? People that use statistics as evidence of something without using their own original thoughts to comprehend them. Why post at all if you won't try to add your own input? You could just make a thread and post the link without saying anything.
Did you play when Patron Warrior was the best deck in the game? Did you know it had a sub-50% win rate?
Look at how many people in the world championship are bringing warlock and priest. Are they all idiots?
Just try thinking about it for a second. Maybe it will come to you.
I personally believe that highlander priest, murloc paladin, control warlock, and tempo rogue add up to the best decks in the game and the win rate differences between them are very small at the top level.
Wow, calm down fella, I just wanted to have a tighter comprehension of the statistics and whats really going on in the meta, no need to be rude. Have a nice evening pal
Whats up with people saying Razakus and Cubelock are the best decks in the game, and yet VS shows statistics from thousands of matches that show otherwise? I'm not saying they are right, but this really grinds my gears
You know what grinds my gears? People that use statistics as evidence of something without using their own original thoughts to comprehend them.
That is a really specific gear grinder scenario. I think it might be time for a vacation. Someone showing some skepticism over real data is a breath of fresh air compared to people using fake data and completely believing it. The OP looks pretty smart bringing his confusion to the forum compared to people that follow the latter example.
A couple of things can lead to why some of the “best” decks in game have some low-ish winrates. These can include, but are not limited to:
The prevalence of a deck leads to targeting and tech choices, as well as an opponent mulliganing for that deck.
The extremely high skill cap of decks. Skill cap is more or less how hard it is to play the deck 100% optimally. Every play you make maximizes your probabilities of winning the most. Ofc, even pros, much less regular players, can’t reach this. However, pros can get close, and the difference between a pro and a regular player gets pronounced the more intricate and hard to pilot a deck is. Using the patron warrior example, it had a sub 50% winrate on ladder, but was one of the most broken decks in the game if you look at how pros played it. If only the best pros played all the decks on ladder, the winrate of cubelock and razakus priest would probably go up, and some of the aggro winrates would probably go down
Hm I see your point, that's the explanation I usually hear; but it really makes me wonder if piloting skills in HS really affect the winrates so much on such a large range of matches
Whats up with people saying Razakus and Cubelock are the best decks in the game, and yet VS shows statistics from thousands of matches that show otherwise? I'm not saying they are right, but this really grinds my gears
You can just let yourself get influenced only by stats that organize the decks by win rate that doesn't completely represent th reality, specially when probably the 2 strongest decks are also the only 2 decks that require a lot of experience and knowledge in the game, Highlander priest and cubelock, as much as salty kids cry about those being auto-pilot bs, if you have played the deck and you can be an objective person, you see the reality, both decks are very difficult to pilot properly and efficently and yes you can blow any opponent if you get the dream hand but that can be said for any deck, the real skill test is how do you play them when you don't or even when your hand is complete garbage, this type of slow and combo oriented decks can't be compared with Murloc pally, which is at the top at all ranks, I'm not saying the deck is not good but with a curvestone fast deck is much easier to not fuck up, you play a lot less turns on average and there are many "optimal" starting hands that can win you the games easily, you can pretty much mulligan 1-drop into 2-drop murlocs and you'll be on a nice position, even with whatever 1-drop into 2 is jut very good tempo, both priest and warlock require you to mulligan for specific match-ups you can't mulligan the same for aggro paladin and then jade druid, you would lose a lot and that's the problem., people play a looooot of priest and warlock but or sure they make a lot of mistakes, it's harder to make a lot of mistakes with fast decks like both popular paladin archetypes, the games are shorter and the decks are super intuitive and it has nothing to do with ranks, there are bad players at rank 15 but there are also terrible legend players that have no idea about how to pilot one of this skill intensive decks.
Just ask any streamer and/or Pro player and they will answer that the absolute best decks in the meta rn are priest and warlock, specifically Highlander and cube, under the right hands those are borderline unstoppable , from there they'll probably say Tempo rogue, fast paladins or Aggro druid and they'll also confirm that both of those op decks are very skill intesive, win rates and stats are not everything, you always have to know how to read into them or you'll just be lying to yourself thinking those decks are not as good as they actually are.
Hm I see your point, that's the explanation I usually hear; but it really makes me wonder if piloting skills in HS really affect the winrates so much on such a large range of matches
It does actually.
We love to bash on hearthstone for it's 'fun, interactive gameplay' and how 'skillful' it is. However, you then have to realize most of the people who say this never show up in high end tournaments, nor are fighting for the top 10 spots in legend. In fact, despite all of these decks being highly used by the populous you still end up with similar people on the same spots in the rankings. That is, we fuss about Tier 1 decks tearing up newbies at rank 20 without really declaring "wait, someone is using the same deck that hits #1 legend and is STILL in rank 20". Then you go up in the ranks and you STILL see the same deck with folks still stuck at the lower ranks.
Win rates are determined by the deck you are running and the person running the deck. A newbie with a tier 1 deck isn't going to hit Legend, while a top performing player isn't going to hit #1 legend without a deck suitable to topping the meta.
So yes, the player behind the deck matters A LOT in hearthstone. Don't believe the "I play well but RNG/meta/blizzardriggingthegame keeps stopping me." hype.
As far as the vicious sindicate stats, they judge deck win rates while ignoring player skill. Thus that rank 15 player using cubelock and losing to my divine shield Bolvar Paladin is as important as that guy who beat the #1 legend player with it. And there's a LOT more rank 15 players losing to bad decks than #1 legend players.
That doesn't even GET to metas where a deck is just hard to use (patron couldn't break 50% for the first 5 months due to no one knowing how to use it, or even build it properly), metas where the deck great but keeps getting countered in ladder (Rogue most of the time) or decks that are too slow for proper ladder climbing (any control deck), you get the idea. MANY factors determine the win rate of a deck.
Statistics don't lie. It's just that we keep putting words in its mouth. When a deck is shown to have a 40% win rate then that means... it has a 40% win rate. Saying ANYTHING, any one word beyond that without more evidence or data is poorly analyzing the data at best and being a scummy liar at..well pretty typical. You need information outside of the stats if you want to know WHY something happens.
Note that none of this is flat out saying that Raza Priest/Cubelock are the strongest decks or not. But we're going to have to get into analysis and other information beyond raw win rates to explain WHY Murloc Paladin wins more than Raza before we jump into conclusions, and there's a LOT of possible reasons to go over.
A lot of people answered this question well, so I won't rehash what they have to say.
I will add that this is why the Tempostorm meta report isn't just a load of bullshit like people think, despite the fact that it isn't backed up by data in the same way as VS or HSreplay, etc.
A lot of people answered this question well, so I won't rehash what they have to say.
I will add that this is why the Tempostorm meta report isn't just a load of bullshit like people think, despite the fact that it isn't backed up by data in the same way as VS or HSreplay, etc.
WELLLLL, it has the same issue, except that instead of giving us the data for us to stuff bad analysis into, they give you pre-built analysis. Instead of trusting ourselves we're trusting the folks of Tempostorm. There's also the fact that many people just take the raw number they spit out and don't even look into their reasoning to see if it's at all accurate. Blindly trusting the "#1 deck in tempostorm" is about as useful as looking at VS' power score and saying "This is the best deck."
VS, Tempostorm, HSreplay, Hearthpwn, your own data, or just what you casually gleamed, it's all the same. Are you actually THINKING about what you are looking at and asking what it's actually saying and if it's true, or are you just grabbing some random bit of text and going "SEE!? I'M RIGHT!?"
If the former, ALL of them are useful in their own way. If the latter, they are all just a load of bullshit. And a lot of people think in the latter.
A lot of people answered this question well, so I won't rehash what they have to say.
I will add that this is why the Tempostorm meta report isn't just a load of bullshit like people think, despite the fact that it isn't backed up by data in the same way as VS or HSreplay, etc.
WELLLLL, it has the same issue, except that instead of giving us the data for us to stuff bad analysis into, they give you pre-built analysis. Instead of trusting ourselves we're trusting the folks of Tempostorm. There's also the fact that many people just take the raw number they spit out and don't even look into their reasoning to see if it's at all accurate. Blindly trusting the "#1 deck in tempostorm" is about as useful as looking at VS' power score and saying "This is the best deck."
VS, Tempostorm, HSreplay, Hearthpwn, your own data, or just what you casually gleamed, it's all the same. Are you actually THINKING about what you are looking at and asking what it's actually saying and if it's true, or are you just grabbing some random bit of text and going "SEE!? I'M RIGHT!?"
If the former, ALL of them are useful in their own way. If the latter, they are all just a load of bullshit. And a lot of people think in the latter.
Oh I agree that you need to take it all in and reach your own conclusions. But I was mostly speaking to the people who are like "lul Tempostorm, no data lul."
Those types of people don't know what they're talking about.
The subset of players that each of the sites like VS, TS, Metastats etc takes from is a lot lower than people think, so while their stats have "some" accuracy, they are often not always entirely reflective of the wider player base.
That said, they are better than nothing - they also tend to use Bayesian statistical analysis rather than Actual percentage results. Some think this is a better method, but the jury is still out on that in fairness since it always has to start with a biased result schema.
The subset of players that each of the sites like VS, TS, Metastats etc takes from is a lot lower than people think, so while their stats have "some" accuracy, they are often not always entirely reflective of the wider player base.
That said, they are better than nothing - they also tend to use Bayesian statistical analysis rather than Actual percentage results. Some think this is a better method, but the jury is still out on that in fairness since it always has to start with a biased result schema.
Well that's statistics. You'll never get the 100% full population, so you get a random sample that tries to simulate the full population.
The stat sites do a good enough job of getting that random sample, so honestly, that it's not close to the full population is a moot point. It's a sample, and one good enough to pull data from with a reasonable range of error. Unless this is leading to a debate over whether their collection methods ARE getting a proper random sample and not a biased one.
As far as the analysis, I'm far away from stats studies so it's all muddy, but I've always held that it's less an issue of the method used as it is the transparency provided. So long as we know what they are doing and how it affects the data we can figure out how useful it is in various situations.
The subset of players that each of the sites like VS, TS, Metastats etc takes from is a lot lower than people think, so while their stats have "some" accuracy, they are often not always entirely reflective of the wider player base.
That said, they are better than nothing - they also tend to use Bayesian statistical analysis rather than Actual percentage results. Some think this is a better method, but the jury is still out on that in fairness since it always has to start with a biased result schema.
Well that's statistics. You'll never get the 100% full population, so you get a random sample that tries to simulate the full population.
The stat sites do a good enough job of getting that random sample, so honestly, that it's not close to the full population is a moot point. It's a sample, and one good enough to pull data from with a reasonable range of error. Unless this is leading to a debate over whether their collection methods ARE getting a proper random sample and not a biased one.
As far as the analysis, I'm far away from stats studies so it's all muddy, but I've always held that it's less an issue of the method used as it is the transparency provided. So long as we know what they are doing and how it affects the data we can figure out how useful it is in various situations.
I think the point I was trying to make about the sample size vs the population (real), was that although it is a sample and is often a good indicator, even samples can sometimes be way off / out of the ordinary.
For example, we might see Big Priest dominating for 4 weeks on the trot, then suddenly they drop off the top tier. At which point people claim Big Priest "is done for" or "is no longer viable" until it magically reappears the next week. (This is just an example, not something I am claiming happened) Hence you get random discrepancies to the real data - like perhaps what the OP is showing in his stats.
I'm not saying that is the "facts" - just an example of a possibility. :-)
RE: the stats studies etc - yeah, theres some good debate going on about the best way to examine meta data and how reliable it is. It's partly why we see so much diversity in the stats. And also why I often like to record my own stats for review - the sample size may be smaller, but it's often better that way as I get a more realistic view of my place in the rankings etc (compared to the stat sites that are usually very skewed by the lower ranks playing every deck they dream is the next big thing! :-P )
The subset of players that each of the sites like VS, TS, Metastats etc takes from is a lot lower than people think, so while their stats have "some" accuracy, they are often not always entirely reflective of the wider player base.
That said, they are better than nothing - they also tend to use Bayesian statistical analysis rather than Actual percentage results. Some think this is a better method, but the jury is still out on that in fairness since it always has to start with a biased result schema.
Well that's statistics. You'll never get the 100% full population, so you get a random sample that tries to simulate the full population.
The stat sites do a good enough job of getting that random sample, so honestly, that it's not close to the full population is a moot point. It's a sample, and one good enough to pull data from with a reasonable range of error. Unless this is leading to a debate over whether their collection methods ARE getting a proper random sample and not a biased one.
As far as the analysis, I'm far away from stats studies so it's all muddy, but I've always held that it's less an issue of the method used as it is the transparency provided. So long as we know what they are doing and how it affects the data we can figure out how useful it is in various situations.
Fair enough, but the sample is not random. Every single player that's recorded by data aggregators has a fairly high stake in the game as they are using tools to better themselves. Thus it's more likely they are more skillful then the general populace. Though it could be said that many higher skill players do not opt into this tracking (for obvious reasons).
The chart is accurate from what i have seen. You will see a ton of cubelock and tempo rogue frequency. The druid and paladin variants have been insanely strong, yet you barely see them. Perhaps some type of shift where people choose not to play aggro despite it being stronger. Highlander priest on the other hand... it's the most represented deck in HCT. Hard to argue with that. I would personally love to see murlocadin destroy everyone that chose not to play it. Might actually make the tournament entertaining.
The sample is not random and you are right in assuming that only players that are highly engaged will use the software. However there are two aspects that you are neglecting that have an impact. The first is that these are ladder games so the skill level is expected to be close especially in the late weeks of a season regardless of the use of software or not. Second, vS does take that into account by reporting only the decks of the opponents, not the players that run the software.
This is true. I'd strongly suggest that folks read the VS FAQ before posting on threads which are critical of their data collection and analysis. I've not noticed a single criticism in this thread that isn't preemptively addressed and answered in the FAQ.
With respect to the OP's post - VS explains in the text accompanying virtually every meta-report why the best decks often under-perform on ladder. If the OP is interested, he'll take the time to read a few dozen words of text, and he'll never be confused again. If he's not interested, he'll instead take the time to bitch about data which he hasn't bothered making any attempt to understand, and he'll always be confused.
https://www.vicioussyndicate.com/drr/vs-power-rankings-data-reaper-report/
Whats up with people saying Razakus and Cubelock are the best decks in the game, and yet VS shows statistics from thousands of matches that show otherwise? I'm not saying they are right, but this really grinds my gears
A couple of things can lead to why some of the “best” decks in game have some low-ish winrates. These can include, but are not limited to:
The prevalence of a deck leads to targeting and tech choices, as well as an opponent mulliganing for that deck.
The extremely high skill cap of decks. Skill cap is more or less how hard it is to play the deck 100% optimally. Every play you make maximizes your probabilities of winning the most. Ofc, even pros, much less regular players, can’t reach this. However, pros can get close, and the difference between a pro and a regular player gets pronounced the more intricate and hard to pilot a deck is. Using the patron warrior example, it had a sub 50% winrate on ladder, but was one of the most broken decks in the game if you look at how pros played it. If only the best pros played all the decks on ladder, the winrate of cubelock and razakus priest would probably go up, and some of the aggro winrates would probably go down
Hm I see your point, that's the explanation I usually hear; but it really makes me wonder if piloting skills in HS really affect the winrates so much on such a large range of matches
I always try to look at different websites to see if there's a general consensus. Here are the ones I know:
http://metastats.net/
https://www.vicioussyndicate.com/vs-data-reaper-report-76/
https://hsreplay.net/meta/
https://tempostorm.com/hearthstone/meta-snapshot/standard/2018-01-08
Edit: of course, as other people said, you need to realize what this data means and how realistic it is
One does not simply walk into Mordor,
unless they want to be the best they can be.
A lot of people answered this question well, so I won't rehash what they have to say.
I will add that this is why the Tempostorm meta report isn't just a load of bullshit like people think, despite the fact that it isn't backed up by data in the same way as VS or HSreplay, etc.
One does not simply walk into Mordor,
unless they want to be the best they can be.
The subset of players that each of the sites like VS, TS, Metastats etc takes from is a lot lower than people think, so while their stats have "some" accuracy, they are often not always entirely reflective of the wider player base.
That said, they are better than nothing - they also tend to use Bayesian statistical analysis rather than Actual percentage results. Some think this is a better method, but the jury is still out on that in fairness since it always has to start with a biased result schema.
One does not simply walk into Mordor,
unless they want to be the best they can be.
Hence you get random discrepancies to the real data - like perhaps what the OP is showing in his stats.
And also why I often like to record my own stats for review - the sample size may be smaller, but it's often better that way as I get a more realistic view of my place in the rankings etc (compared to the stat sites that are usually very skewed by the lower ranks playing every deck they dream is the next big thing! :-P )
The chart is accurate from what i have seen. You will see a ton of cubelock and tempo rogue frequency. The druid and paladin variants have been insanely strong, yet you barely see them. Perhaps some type of shift where people choose not to play aggro despite it being stronger. Highlander priest on the other hand... it's the most represented deck in HCT. Hard to argue with that. I would personally love to see murlocadin destroy everyone that chose not to play it. Might actually make the tournament entertaining.