Yeah, you’re right. Just found a post from the admin stating that the stats for overbuff were never particularly accurate even before private profiles.
Citation needed.
https://www.reddit.com/r/Overwatch/comments/b06ch4/question_about_overbuff/eicef3e/?context=3
That doesn’t state Overbuff wasn’t accurate as a sample. It only states what I stated before.
It states that private profiles haven’t changed the accuracy of the site, meaning the data was already flawed. He never refutes the data being flawed in any of his posts.
In fact, looking through his posts, it seems he doesn’t care about win rates and pick rates, and sees the site only as a tool for self improvement.
He never addresses anything about pick rates and win rates and says it only compares the data of people who use the site to track their own data.
That means for every player who uses overbuff and wins a match, there are up to 11 other players in the same match who do not.
He wasn’t addressing that criticism in that post either. That’s a pretty unfair characterization of that post. All samples are “flawed” in that they aren’t 100% accurate. But some samples are better than others and are fairly accurate in general.
I found an old post from Jeff where someone actually analyzed the pickrate data after he posted it. It doesn’t have pickrate proof because the person didn’t screenshot it and Overbuff doesn’t archive this information.
This post gives more insight into how the site works.
Oh wow, I can post links now.
Reddit is a level 2 permission website.
So that says overbuff is decent for determining the top 10 most played heroes which is expected. Their sample group of most played heroes is likely to reflect the most played heroes for the larger majority. It says nothing about pick rates vs. win rates which we can only assume are wildly innacurate based on what the admin has told us about how the data is collected.
Pickrates are who is most played. That shows them being fairly accurate on pickrate. In order for them to have wildly inaccurate winrate data, their website would have to be a biasing factor on how much you win. I wouldn’t assume that’s the case especially since their pickrate data was fairly accurate.
well regardless of if it was or wasnt before (i dont rememvber but it doesnt matter now)…it sure as heck aint now…so giant grain of salt
Private profile had no impact on their data, which is unsurprising given their collection method. You already had to actively keep being searched to be included. And most of the people who were using it were willing to do what they needed to in order to keep using it.
Well it only collects data for people who use Overbuff. If you have a match and only one player uses overbuff, then it will only collect the data for that one player. That means if you have 2 mirrored teams and the overbuff user is playing d.va and they win, it will count the win for that d.va, but it won’t count the loss for the opposite d.va. Also there could be any number of overbuff users in that match. There could be 3 on one team and 0 on the other, or all 12 use it. That is why that portion of the data cannot be accurate. After seeing how they collect data though, I think the pick rates are reasonably accurate.
While that’s true, devs have made comments about winrates in the past that was reflected on Overbuff. For example, Jeff made a comment during Mass Rez that stated that they don’t balance on statistics alone and that what we thought was unbalanced and what was statistically unbalanced weren’t the same. He mentioned not being happy with Mercy and that Torbjorn and Symmetra had winrates that were too good. According to Overbuff at that time, Mercy’s winrate wasn’t that great and she wasn’t the most picked either. She had average stats. Torb and Symmetra had winrates a bit over 60%.
So his statement there was reflected in OB’s stats.
Maybe in that case, but overall I don’t believe the stats on the site can be used to assess win rates, but can be used to get a rough idea on the popularity of certain heroes.
It happened with Reaper in mid-low ranks too back when he had higher lifesteal. Devs don’t really address winrate as much as pickrate though, so it’s harder to judge how accurate they are on that.
They don’t really analyze matches so much though as just pull profile data. So I think it’s probably roughly correct, at least on the highly picked heroes. I’ve noticed so much variance with sub 1% pickrate heroes that I always take their winrates with a grain of salt.
pharah and bastion were played quite a bit in world cup
pharah has just historically never been relevant in ladder even when she was a strong pick (dive, goats, double shield)
it’s hard to buff pharah, who’s actually very meta atm, just because people are unwilling to play her
Pharah is fine where she is. She’ll naturally get better and better as we get more and more heroes who can fly. I can’t wait to see how Echo works.
Sombra could use some work, but could also use some more counter-play. She’s a hero I think will get better as time goes on and more mechanics are added to the game. What she’s really waiting for is a hero or two who have some sort of “status cleanse” ability that itself (in some cases) cannot be hacked and turned off. Once someone gets something like that, Sombra can see some real improvements since being hacked will suddenly have a fix that’s just a swap away.
Swapping heroes shouldn’t always be the answer, but in some cases it can be an answer.
Symmetra still needs to figure out who she is. Is she a mobility-support or is she an area denial DPS? She’s still trying to be both, and she’s failing pretty hard. The worst part is that she’s close. I feel she just needs a tiny bit more focus. Basically either lose the teleporter (nobody uses it anyway) and give some more area denial somehow or lose the turrets and double down on mobility by giving her an actual movement ability.
Torbjorn either needs the CD on his turret reduced a bit, or he needs his M1 fire rate increased. He’s far slower than McCree in spite of having a much worse gun. Buff his M1 from 0.6 seconds per shot to McCree’s 0.4 with the same 18 shots per reload. People should fear his M1 in the hands of someone who knows how to use it. Instead, he surrenders 50% damage by fire-rate to a hitscan hero. Making his M1 feel good to use (it currently doesn’t) is important considering his turret decreases in value the higher in rank he goes.
Bastion
IMO, Bastion simply doesn’t work the way he’s designed. He’s got the same problem Widowmaker does. Way too much of his kit budget is locked up in sentry mode.
I would give him a partial rework and reverse his ult and his shift. Make the immobile turret form into his ultimate ability where he can afford enough survivability to do his job since the whole thing is on a pretty hard timer, and turn his shift into a fast, evasive, hull-down, fire-support mode.
The moment using his abilities isn’t a binary choice between committing suicide and breaking the game, he’ll see more use at all ranks.
For bastion I would do the following:
[Shift] Tank Mode: Change into a tank. While in tank mode Bastion moves 30% faster, and can use his Main Battle Cannon but cannot self repair. Changing back into recon mode puts tank mode into a six second cooldown.
[M1] Main Battle Cannon: 60 impact damage + 60 explosion damage w/ 3m radius. 1 shot clip with an auto-loader. Bastion’s tank mode’s MBC loads rounds individually and can fire once every second.
[Q] Sentry Mode: Change into a reinforced turret. While in Sentry Mode Bastion cannot move, and can use his Gatling gun. Additionally, while in Sentry Mode Bastion’s armor is increased by 200 points (to 200 health, 300 armor). Additionally, he gains 200 temporary armor that cannot be restored with healing. Sentry Mode lasts for 12 seconds and changing back to recon mode is nearly instant unless he overheats his gun. Sentry mode cannot be ended early. Ironclad is removed from the game because the added armor serves the same purpose.
[M1] Gattling Gun: Just like current sentry mode gun except replace the clip with a heat meter. If his gun overheats his ultimate immediately ends. Sentry Mode bastion otherwise has unlimited ammo for the duration and only needs to manage heat. Overheating his gun plays a 1.5 second “overheated animation” which is punishment for failing the heat management minigame.
TLDR: Pharah is fine, Sombra is fine. Both will get better as the game gets more mature. Sym needs to pick a role and torb needs his M1 buffed. Bastion needs his modes swapped so tank is shift and turret is Q.
It’s not difficult to understand, because it’s an elementary schoolers view of how polls and sample sizes work, all wrapped up in a horrible analogy.
My luck in life, it’s only missing Tracer.