The problem I see is people either treating the data as completely invalid, or they take it out of context. For example, people using pick-rate without consideration for win-rate, and vice versa. I also see alot of people use weekly stats in isolation, which is flawed as those stats change daily and you can have some wide jumps from day to day (weekly is perhaps an inaccurate title). For example, I seen Mei go from 50% WR on a monday, to 35% on Wednesday, to 29% on Friday, and back to around 48% on Sunday. This data is certainly useful, but taking just one of those data sets in isolation, without consideration to the rest of the week, simply skewers what it can tell us. Overall, I think itâs a good source but itâs often misinterpreted.
when theres no changes but people change what theyre playing overnight its more than just correlation
That doesnât happen so often, and Iâm sure the opposite trends happen where heroes rise in OWL use yet fall in the live game.
Pretty sure that happened with Mei.
Wow, someone admitting someone else is right on a forum?
Perhaps there are signs of intelligent life out here after all!
You are a one in a million. Treasure yourself.
It does however tell you that these heroâs are being picked significantly more than other healers and offtanks
Helping come to the conclusion that those two characters have something really Good in their kit
that happened when heroes were meta for a long time and people didnt want to play them anymore, like goats and double shield, and it literally happened with hog a week ago, it happened in hero pools when live wasnt even using those hero pools, happens a lot when patches are two weeks after on owl. the only meta i remember originating in ranked is orisa hog
could also mean people like playing ana more than moira, which i dont think would surprise many
It is not invalid. The data is real data. What you have to however do is interpret the data in a way thatâs not invalid. For example looking at a simple pickrate and/or winrate and making conclusions about hero viability or the strength of a hero that way is simply naive. You have to look at all the stats about a hero, then figure out where those statistics come from and then you can make conclusions about them with a grain of salt in the mix
Overbuff has been an accurate tool for meta viability observation ever since it was established and has clearly shown whatâs going on for what 4 years(?) now. Only those who dislike the thought of a huge data sample being used as means to fixate how players play the game or wish that the reality met their opinions instead of the other way, donât believe that it can be used accurately.
Then you can look at gm and masters stats those peopleâs main goal is to win and you can find that Anna and zarya are very popular in high ranks too.
Iâm just saying that those stats can assist and help paint the picture of where a character stands. But it isnât the be all end all
trust this is just something lower ranks think, theres plenty of washed gm players who just play for fun
I think the major issue is itâs basically the only source for data as Blizzard never releases anything. A single source of data is never good for any kind of statistics or data sheet.
Add to that there have been several times where it was blatantly wrong, like when Ashe had over 100% win rate. We begin to wonder about how many times itâs wrong and we donât know?
People claim GM is the smallest pool of players and thus more prone to errors. But people also claim only the high ranks matter⌠so you canât really have it both ways. Overbuff is more information than we get from blizzard but it is still pretty much garbage info.
Itâs like taking a poll of your Neighborhood on which is better Pizza or Hotdogs and claiming itâs true for the country.
Yeah you are right I was one of them for a long time but the majority of the people at least when I was gm were trying to win
But that was my experience. There definitely is a chance that it has changed since then
I was specifically referring to Zaryaâs damage at the higher end of the percentages, where it reaches 20k at around 5% of players. This gets a better estimate for how the best Zaryaâs are playing the game, and I donât really think the average numbers are super accurate (for ana, the avg. un-scoped accuracy is 17%)? I was definitely mistaken on that ana number though, could have sworn it was higher last I checked! Maybe I was looking at a different supports and not hers by mistake? Regardless, it displays a different idea â despite low heal numbers (which Iâm not too keen to take as true) she has a high pickrate due to nade.
it should because if you take basic statistics and see how overbuff messes with dps and support just for having her there then.
Sym being in support instead of dps messes both roles.
because in overbuff they callculate percentages based off the role.
example:
if you have 7 heroes in one role, the â100%â pickrate of supports should be split up between the 7. In overbuff they split it between 8 since sym is a support and the 100% pickrate for dps is being split between 16 not 17
the highest numbers are usually from diamond/masters players, once you hit gm those numbers tend to go down, also sigmas is still about the same
this is because overbuff ignores hitting team mates
why? ana easily has the lowest healing potential besides mercy (who usually outheals her because of the consistency of her heal) and zen for obvious reasons, the other supports all equal her healing but its aoe, ana is actually pretty bad and i think shes only really playable because of the popularity of zarya in ranked, afterall ana was considered nearly a throw pick when dva and sigma were good. i feel theres a lot of recency bias with both ana and zarya after they lucked into a few metas in a row people seemed to forget that they have been pretty terrible for a long time (zarya was played in goats but we try forget that)
There are a few simple things that makes Overbuff data almost completely useless.
-The sample size is TINY. I have seen characters go from bottom to top within a single week. It takes a LOT of effort to put your data here, 99.99% of the population of the game does not bother.
-The way win rate is calculated is completely flawed and twists the value of the data. I still canât even say I fully understand how its calculated. A hero that is âswitched toâ or a hero that is âswitched offâ more often, changes their win rates completely.
-Historically, the win rates shown there have been extremely disconnected with reality. Which is to say actual, solid proof of balance, such as highest level of play data (tournaments) or top 500 data.
Overbuff should not be used at all here. It is basically useless for balance discussion.
it means that it has aspects that are very outdated since sym is a dps hero now and has been for a long time.
The problem is it goes against what they think is the case.
This is pretty much the main reason.
Yes, statistical data isnât always the best thing to look at when it comes to OW. But when looking at pickrates, and winrates (both objective things), using overbuff is one of the most accurate sources we have.
Can we test this?
Like find someone who has no stats on Ana and have them shoot a low HP ally a bunch of times and see if their âun-scoped accuracyâ stat goes down?
She also has the highest healing potential
She may potentially miss all her shots and heal nothing or hit them all with a heal nade and heal a potential of 156HP per second over 4 seconds.
Considered by who?
Do you mean you personally considered her a throw pick?