Heroes of Overwatch, let's talk data!
Why is no one saying anything?
Maybe people still thinking on what of two topics replay to.
Data rules. This is a lot of rope for people to hang Blizzard with though, because even though itās clearly well-reasoned and explained, people will still find a way to complain about OWL, pro players, streamers, and Symmetra.
28 Likes
So, I wonder how are they going to address the problems with and against certain heroes (i.e. Reaper, Junkrat, Symmetra, Mei, etc.)?
Or, for clarity, how will they maintain their design while correcting their issuesā¦?
11 Likes
When the Baptiste buffs came through everyone was like Waaahh itās so OP but then he has the lowest win rate of any support by far
17 Likes
as a quantatitive analyst Iād quite like to know the sample sizes, as that data looks hella-noisy
For the most part, all supports who received a balance adjustment, including the bugfix to Mercyās Valkyrie, saw an instant change in unmirrored win rateāexcept for Baptiste, who received changes in a later beta patch. This means that by and large, those changes had their intended effects!
this bit raised an eyebrow
23 Likes
A very interesting look underneath the hood of balancing OW. Interesting to see the variety of data tools available. Makes me wish a few more were available to us!
I wonder about the granularity of some of the usage data. They mentioned being about to determine the percentage of a characterās usuage on a map, and their contribution to win or loss. Do their data tools allow them to identify where the hero was used? Like using Symmetra on first point only, then swapping after a point is lost? Or attack versus defense? The auditor in me is salivating at the data prospects. Lol
3 Likes
They literally said they have a thing that says how long a hero was played during any given match, and they know that Sym is mainly used for defending the first points of maps or whatever
1 Like
Yes, but the Symmetra first point usage conclusion could be based on the data, or it could be a conclusion based on the data and other observations. Is the data for when and where characters are used built in, or did the high win rate data prompt addition analysis with other tools?
3 Likes
Have you guys considered one hero per match and draft system during the match? That could avoid queue times and reduce mirrors.
The game works based on counters, mirror matches are things that reflects issues on current design. Removing mirrors will also broader the hero usage and flag often the overperforming heroes and the maps they overperform
8 Likes
D.Va is unsurprisingly showing terrible stats, I donāt understand why they didnāt give her a buff in either patch.
28 Likes
Can you guys make these stats public, please? It would be nice to be able to see them ourselves from time to time.
30 Likes
I would like to also see a similar dataset but around the top % of players. I wanna know what they figured out.
5 Likes
It would be nice to see the charts of each rank.
Also hope they revert the soldier nerfs and give mercy some love outside of an unintended infinite flight bug.
5 Likes
Hero Balance: How data & feedback drive change
Our team had a focused goal for the PvP beta: we wanted all our heroes to be fun to play without being unfair to play against. Our usage rate analyses suggested that we were on track with the former, and thus our focus turned to game balance.
[Looking at Mei, Symmetra, Junkrat and Reaper, but especially Mei]
Are you certain about that?
16 Likes
thats because its a hero that has a much higher barrier of entry in terms of how much skill allows u to carry, and the majority of the community is bad at playing support and thinking its a weak role when it never was to begin with
1 Like
Nice to have reassurance how win rates are taken into account and itās not just about popularity equals overperforming.
Wonder how the popularity kids will move the goal post now.
Edit: lol like clockwork
13 Likes
However, due to the nature of Overwatchā namely the ability to switch heroesā basic map win rates are not representative of a heroās actual performance
The devs debunking the oversimplicity of the winrate argument with 1 developer blog. You love to see it.
To arrive at a better metric, we approach win rates fractionally by looking at how long a certain hero is played within a map.
We address this by applying the same win fraction calculation to the periods of time where only one team is fielding that heroā we call this an āunmirroredā state. By looking at unmirrored win rate, we can see how a hero like Sojournā who reached over 50% mirrored rateā performs when there isnāt another Sojourn on the opposing team and separate her win rate further from 50%.
As predicted: Weighted and unmirrored winrate is what matters. Not raw winrate.
Robotwizard predicted this years ago. The man deserves a medal.
23 Likes
So what Iām seeing here is that Ana sucks to play against and is in every game just about
Sounds normal
25 Likes