Nobody knows how accurate overbuff is any more and any discussion that quotes it gets pushed back on with “it’s not accurate” or some such argument.
Blizzard has the real stats and a lot more that they could easily release in the form of an improved overbuff. The amount of detail we could get is probably ridiculous (down to percentage hits and misses, average moira orb damage, torb turret damage, health pack stats, hacks per game, etc.). What would be the harm in sharing this in an up-to-date portal? It would allow real discussion with real metrics. And it would allow the community to actually have meaningful conversations about the effects of buffs/nerfs instead of the mess we have now.
It could even make the forum less toxic if we had some real information to use to validate points being made. Currently for every “X is OP” thread we have as many “X is UP” threads with only overbuff and opinions to back it up. It is making the forums much worse than they need to be.
34 Likes
Where do you think Overbuff gets it’s stats from? Do you think that some players just manually report it to the Overbuff people and they put it on the site?
4 Likes
It depends on who’s profile is private or public.
1 Like
Yeah but enough players are public to make the stats accurate within an acceptable margin of error. So for the most part it’s pretty accurate
17 Likes
I would point out that there have been multiple occasions in Overwatch where a hero has had over a 17% pick rate on Overbuff which is literally impossible and has been for about 4 years now. It clearly tries to take the data that is available to it and adjust for what it does not know.
2 Likes
Yeah, shut down Overbuff. It provides nothing accurate but people keep using it as a reference. Ugh.
Maybe you were filtering by catagory of heroes. Ana has over a 30% pickrate for supports, so does Rein for tanks. DPS is still split into offense and defense over there so that also can inflate pick rates.
2 Likes
Do you have data on what “enough” is? Or are you just assuming?
At least hundreds of thousands, maybe millions. Statistics doesn’t require the entire pool, merely a sample to study. We just have to assume, but obviously full data is preferred.
7 Likes
As much as I want public access to statistics and an official Overwatch api, I can’t say I think it would make anything less toxic. I bet it would further validate people’s opinions based on incorrect usage of statistics.
Exactly. Look at how many people wake up every day and use this to fuel their agendas before they even have breakfast. It is pathetic.
From what the devs have told us, it usually lines up to what overbuff says. So I think that argument is fairly unfair for accuracy sake.
That said, I think real stats won’t help. No stats would likely be better for ladder.
8 Likes
Just look up sample size.
You need only few public profiles to have a quite accurate numbers
2 Likes
Overbuff is fine with me.
I work with data and if it was my job, Overbuff would be a fine source.
12 Likes
Yes, actually. They very well should report their own stats as it would be more extensive and accurate information and we might even understand their balance decisions a little better. My guess is they keep it to themselves so they can, in those rare cases, show up in a thread and say “well actually our stats say…”
Uhh, what hero what date?
1 Like
Even with data provided by Blizz. it doesn’t change anything.
Data showing their playerbase is mostly in silver, gold and plat did nothing to change why or how they balance things or make things fun.
The problem with using Overbuff stats in discussions isn’t so much the accuracy of the data as the lack of interpretation, which would affect first-party data as well. It’s the same problem that pretty much all competitive games have had since the dawn of stat trackers: People look at a number and make sweeping assumptions about what that number means without context.
6 Likes
I don’t think we need to discuss the numbers as we’re not the developers.
I don’t think it could be accurate in that way because accurate statistics are supposed to be more randomized than that. I don’t know if people choosing to private or public their profile is proper randomization and introduces bias.