LogFAQs > #973113989

Dude foolmo you just don't get it and you're going to pretend you're smart just like usual

Here's the last post I'm giving you on this. I'm going to give you a completely absurd example so you can (hopefully) get why you can find anomalies in data by anecdote in extreme circumstances.

Let's say Cosmo's overall play rate was something minuscule like 0.01%. Say people never see Cosmo in games. Say you've got hundreds of reports from the playerbase where people swap to decks with Wong and claim they see him 50% of the time.

Now you can either say duh huh frequency illusion duh huh confirmation bias or you can realize that maybe some these people are reporting what's actually happening and that if Cosmo's play rate was actually 0.01% the odds of this happening in widespread numbers would be so small that you wouldn't be seeing it in random samples even if people were being biased because they would have to outright be making things up to see him that often.

Now obviously we have snap.fan's data to compare to and it does give us a decent starting point. For example, valkyrie's play rate is supposedly 6% so it does become a bit odd if I see her play rate increase to 50% if I use cards she counters-- but it's not foolproof because we don't have the data broken down well enough and there could be a higher rate of Valkyries at my matchmaking tier despite not seeing her much at all when I play other decks. It's not a low enough percentage to be ironclad.

Now if it was not Valkyrie and instead Shadow King at his 0.53% play rate I was seeing the argument is stronger. Considerably so. Because the odds of seeing multiple Shadow Kings in a short time frame is very small.

That's why if we had more confirmed data in general this becomes way more testable even with the small sample sets the playerbase has, but doesn't until then. Otherwise you have very smart people who even when presented with tallies will always call it RNG and that we should say the earth is flat because the people who know better said so until we have something that clashes as hard as a Shadow King. Valkyrie at 6% (keep in mind she was higher rarity then so even lower play rate) won't be enough to appease anyone to dive into it further, especially since the explanation then can be hand waved with "well shes in more decks but people are holding her more often when you're not using your destroy deck with big stompies"

Unfortunately because the game is reasonably well balanced you don't have a lot of cards that are basically never used, and if you do it's often because there are better options to do what you want like Shadow King vs Valkyrie. That's why we need more data. If we get breakdowns by Tier and Collector Level we could probably find a lot more low percentage counterpicks to investigate.