How does Overbuff get the correct pickrate numbers when almost every players account is private?
Iâm pretty sure it can see them anyway.
It only see public accounts, that is why itâs accuracy seems to come up in discussions.
2 Likes
It works like a study. There are still enough people with open profiles.
7 Likes
It doesnât. It gets an estimate that can be wildly off reality. For example smurfs and hackers often will not have public accounts so they likely wonât turn up. A lot of people who main less popular characters like Sym often make their account private so again these types of heroes are not accurately represented. Lots of really casual players have no idea what it is and arenât linked to it, again causing a drop in accuracy. Like sometimes it gets so bad if you look at the win/loss percentages they average out to like a 60% winrate, which across the entire game would be impossible.
Overbuff is a useful tool, but do not take it as verbatim, itâs the best we have but itâs still a warped snapshot. Also it does seem to represent higher ranks better, still not perfect but worth mentioning.
They donât Overbuff is poo and we shouldnât be using it, those statistics are not objective as theyâre so full of other variables, this as well as the fact that profiles are on private by default? how anyone can take these as accurate is beyond me.
1 Like
The alternatives are even less objective though.
Iâll agree though, that winrate stats are usually not worth talking about.
But pickrate is a lot more straightforward.
2 Likes
Overbuff certainly should be affected by its opt-in nature. I disagreed with private profiles being implemented mostly on the grounds that the community would lose access to stats to make sense of the game with. That said, despite my reservations when private profiles were put in, the daily stats of hero WR and PR did not vary dramatically before and after the implementation. Which is to say whatever bias is apparent in that data set is similar. Itâs possible the two groups diverged more in the years since, but I doubt itâs by any large amount. So far itâs seemed to fairly accurately track trends in hero usage and overall success rate.
Either you won the game or you didnât, so you look at the avgs of 10000s of games, where all of these âvariablesâ occur at whatever frequency, to discern something useful from the results.
2 Likes
It doesnât.
Some people argue that getting data from unhidden profiles is the same but I disagree. People who go out of their way to unhide profiles are specific type of players and are not a microcosm of the entire player base. Thus, the data overbuff can get is not a âsample size.â
Also, people use pick rates basically as a be all end all stat to determine if a hero is op or too weak but itâs not that simple.
1 Like
How did you determine that?
And how did you determine that the sample is actually random and not biased?
Overbuff is to be taken with a grain of salt.
It is about as accurate as a moderation war wiki article, but it does have a large enough sample size to accurately track general trends.
1 Like
They will take the Overbuff stats seriously if it suits their narratives.
1 Like
Better question is why do people focus on them so muchâŚthey dont really say muchâŚ
1 Like
If someone is trying to prove a point, and overbuff supportâs their argument, you can be damn sure they are going to be focusing on that.
People will then argue overbuff is not reliable.
And they would both be rightâŚ
and wrong.
Stats, they have a massive sample size.
4 Likes
People will also say overbuff âisnât a reliable sourceâ when it DOESNâT fit their narrative though. I have lost count how many times this has happened.
1 Like
I donât, I always take it as reasonably accurate, except maybe GM weekly on low picked heroes.
Like if the stats are stable from day to day the sample size is big enough.
1 Like
It doesnât. It is not correct.
You can see the proper pickrates on the scoreboards.
My favorite is when one section of people use a particular rank (usually GM) to prove that a hero is too good even if theyâre pretty bad elsewhere. Meanwhile, other people will use all ranks to prove that a hero is good even if theyâre pretty bad in GM.
I mean, if youâre willing to use both then itâs literally impossible to balance until every hero has identical skill floors, skill ceilings, and skill scalings; Iâm honestly not even sure if thatâs desirable even if it wasnât a pipe dream.
Then if Overbuff suits neither narative you can just say that Overbuff is inaccurate and throw everything out the window lol. Honestly, I could bring up Overbuff for any hero and cherry pick stuff that would get a LOT of traction on here so long as I pick an opinion thatâs already popular.
1 Like