Well, the sad news is: most likely no, you don't understand them (well enough). At least, not how they are computed, or what they really mean.
And - before you take offense - neither do I, mind ;)
But there's a professor, with a PhD specializing in statistical theory, who wrote some stuff shedding light on the subject. More precisely, about reading numbers from sites like HS Replay or VS, and understanding the biases they introduce.
IMO it's a wonderful read, providing useful insight, and it also spawned a great discussion. You can find it here.
It really isn't all that complicated, but also not all that problematic.
Anyone, who gives HSreplay a visist can see, that the site gives out confusing data without any explanation and lacks transparency. VS is doing a better job mirroring the metagame, but also with limitations. For some reason, the person writing all this has not mentioned that any site giving out winrates etc. are using data from people who care enough to record their games in the first place, likely people who play a lot and better than the "normal" player. This is a problem to Hearthstone in general: Most discussions about balance or the meta in general are twisted, when they are led by a vocal minority that is (apparently) only found on rank 5 and higher, although Team 5 has stated multiple times that the majority (>50%) of all players are on rank 20 and lower. Sadly, even some developers fall for that narrative that the game is in a good state, when the top 5% of all players feel the game is "balanced". But that's a discussion for another day.
This discussion about winrates is somewhat irrelevant. aside from Team 5 making card changes based on data that are, assumably, gathered in similar ways. But that's their problem. Any statistical analysis about a deck's winrate can only give you an idea of how well you will perform with or against that deck, nothing more. A winrate is no clear cut fact, because the reported winrate comes from a player with more experience, or from a rank environment that is more favorable to his deck, or uses a limited number of games, or the deck was changed halfway through recording and the deck you copied is an outdated version, or for some other reason. So what? The few people who actually rely on statistical reports for some reason hopefully know, that they have to see these data more critically. Everyone else can move on with their lives, if they can't reproduce the winrate of the deck they copied. They try out new decks, until they run out of dust.
The most relevant statistic on VS is how frequently a deck is played, and how high the relative winrates are (for specific matchups). The popularity of a deck is more important than its supposed strength, as it will determine what else you can play. The winrates don't really matter, especially when they are not differentiated.
The most relevant statistic on VS is how frequently a deck is played, and how high the relative winrates are (for specific matchups). The popularity of a deck is more important than its supposed strength, as it will determine what else you can play. The winrates don't really matter, especially when they are not differentiated.
This can't be stated enough. The frequency of play of a deck can drastically affect its winrate. Unpopular decks arent teched against like popular ones. And deck tracker software can't handle that all the time as an unpopular deck might use commonly used cards on their blowout losses and unpopular and expensive decks are really only heavily played by pros, streamers and hardcore players with better than average skill.
Also using tracker's winrate data in large amounts is sketchy as the Hardcore Community is both aggro haters and bandwagon jumpers. So the pool of people who would use trackers and sumbit it is a subculture of its own and skews data more.
Rollback Post to RevisionRollBack
To post a comment, please login or register a new account.
Well, the sad news is: most likely no, you don't understand them (well enough). At least, not how they are computed, or what they really mean.
And - before you take offense - neither do I, mind ;)
But there's a professor, with a PhD specializing in statistical theory, who wrote some stuff shedding light on the subject. More precisely, about reading numbers from sites like HS Replay or VS, and understanding the biases they introduce.
IMO it's a wonderful read, providing useful insight, and it also spawned a great discussion. You can find it here.
HTH!
It really isn't all that complicated, but also not all that problematic.
Anyone, who gives HSreplay a visist can see, that the site gives out confusing data without any explanation and lacks transparency. VS is doing a better job mirroring the metagame, but also with limitations. For some reason, the person writing all this has not mentioned that any site giving out winrates etc. are using data from people who care enough to record their games in the first place, likely people who play a lot and better than the "normal" player. This is a problem to Hearthstone in general: Most discussions about balance or the meta in general are twisted, when they are led by a vocal minority that is (apparently) only found on rank 5 and higher, although Team 5 has stated multiple times that the majority (>50%) of all players are on rank 20 and lower. Sadly, even some developers fall for that narrative that the game is in a good state, when the top 5% of all players feel the game is "balanced". But that's a discussion for another day.
This discussion about winrates is somewhat irrelevant. aside from Team 5 making card changes based on data that are, assumably, gathered in similar ways. But that's their problem. Any statistical analysis about a deck's winrate can only give you an idea of how well you will perform with or against that deck, nothing more. A winrate is no clear cut fact, because the reported winrate comes from a player with more experience, or from a rank environment that is more favorable to his deck, or uses a limited number of games, or the deck was changed halfway through recording and the deck you copied is an outdated version, or for some other reason. So what? The few people who actually rely on statistical reports for some reason hopefully know, that they have to see these data more critically. Everyone else can move on with their lives, if they can't reproduce the winrate of the deck they copied. They try out new decks, until they run out of dust.
The most relevant statistic on VS is how frequently a deck is played, and how high the relative winrates are (for specific matchups). The popularity of a deck is more important than its supposed strength, as it will determine what else you can play. The winrates don't really matter, especially when they are not differentiated.