Addressing some misconceptions regarding balancing in VG
So apart from being a moderator here and the coach of a pro team (Renegades) I am also part of the core testing group, one of two captains in the PBE and a long standing member of Phase 1 PBE (I’ve been playtesting regularly for over 2 years now). I have also visited the Lair and met up with Nivmett and the other folk involved in balance, and was lucky enough to be there the day that a patch dropped and that a hotfix was deemed necessary (Reza - on release was significantly underpowered) so I have seen how that decision was made, and was invited as a playtester to contribute my thoughts on the proposed buffs to make Reza viable.
So broadly speaking, outside of people at SEMC themselves, I think I have one of the better understandings of how balance decisions are made, what the process is, who is involved and so on.
There are actually two balance teams at SEMC - the live balance team and the new heroes/items team. The PBE serves as the intersection of these two groups - with Nivmett at the pointy end of balancing (along with Nightwalker59 who also runs the PBE more directly than Niv now does).
It is worth noting when balancing you have to take into consideration 3 things - the overall game environment (ie if we change the health of turrets what are the impacts of that?), changes to items, and changes both directly to a hero and to heroes who interact with them. So its like a web where changing one node of that web has flow on effects on the rest of the web. Its not as simple as “buff Skye to counter X” because if you do so you will find out that making one of the best mobility heroes in the game strong is very very oppressive…
It is also worth noting that the changes that can be made by the balance team are limited - changes that are basically tweaks to numbers are much preferred to kit changes because these can be tweaked directly by the balance team whereas kit changes cost programming time from elsewhere, require approval from higher up and typically take several patches before they can be implemented. Enviromental changes and item changes are similar - number tweaking is much easier than changing how something works, both are easier than changing how something looks.
So how is hero balancing done each patch?
Firstly the live balance team identifies heroes that may need tweaking - they do this by looking at the data on winrates at all tiers both seperately and together. The broad aim is to have all heroes within 5% of a 50% winrate - with up to 10% for heroes who are primarily counter pick heroes such as Reim or Petal. This is passed onto the PBE team who consults with the new hero, items and gameplay team to get changes from them and the potential flow through of those changes onto other heroes and other aspects of game balance, along with any kit changes which are going to be tested and so on. From this an initial set of changes are developed and then put into the PBE. The PBE members get informed what the changes are and why they are being made and then we playtest with balanced parties to see whether the changes are too much, too little or just right. Based on the winrates, player feedback and observation, the PBE team tweaks the numbers and tries again the following day. This usually starts about a week after a patch drops and continues until a week before the new patch drops, with a new build and playtesting typically 6 days a week.
How about new heroes - there will have been months of tweaking, redesigning and internal playtesting before a new hero hits the PBE. When they do they usually reach Phase 1 testing first - Phase 1 tests not just probably changes to the game, but also things that are being experimented with such as new items etc - so most of the new stuff you see in Phase 1 won’t make it into the game. Phase 1 consists of around 100 players - almost all very high level in terms of their skill level, understanding of the game and dedication to playtesting - many playtest for several hours every day. Its not uncommon for a new hero to hit Phase one without a character model yet (usually they use Rona or Catherine as a substitute) and with a defined kit but with a couple of options on how that kit might be implemented (Flicker for example could have had a much weaker slow + root when it ended on his B and a pool of silence on his A or the kit we all now know). And the initial testing is regarding which kit is “more fun” - so its about how it feels, what roles we can make it work for etc. Then the kit is locked and it becomes about tweaking the numbers to make the hero balanced, working out what they interact with and so on. Phase 2 becomes involved at this stage, bringing in about 500 more players for testing.
It is fundamentally harder to balance new heroes than existing heroes however because you don’t have the data yet to figure out exactly where they sit - take on release Reza as an example - the general feeling in the PBE prerelease was that Reza was a balanced hero - on release 35% winrate… Why? Because he is a high risk high reward hero, so high skill players are much better placed to utilize his kit fully. He needed some buffs to make him a bit more forgiving for lower skill players, whilst also not being too good at higher level of play. Typically balance changes will be made over the next few patches to bring a new hero in line with the target winrate.
Finally balance decisions are made within a nexus of pressures - SEMC don’t want to release a new hero who is deeply UP because it supresses their sales for months even if you hotfix them, likewise they don’t want to release an OP hero because it makes the game less fun and more work for the balance team down the track. However at the same time the hero designer wants to protect their beautiful new creation and keep them to their vision, so changes have to be in line with that.
Balance in VG is actually impressively good - easily on par with Dota2 and LoL and historically has typically been better than them (back in 3v3 only days) probably because of the smaller set of heroes. This can be determined by looking at winrate variance between heroes. Yes sometimes heroes are too strong or too weak - but if thats the case they will usually be addressed within a patch or two.