*disclaimer, I am happy with my rank. I only to seek better match quality and more accurate skill assessment of players.
When analyzing this game certain things always hold true - the value of winning and value of losing.
The issue I see are a lot of problems like boosted accounts, potato people, smurfs, stacks vs solo q. A lot of it is because we over value the end result vs the journey to get there.
Here I go again with crazy ideas - sometimes in brain storming to fix an issue you have to throw away even the basic concepts and question them.
Two things are always immutable :
The gain you get with a win.
And this is a âteam gameâ.
But those are not mutually exclusive. And some things get overlooked by only looking at one immutable variable over the other. For example, What is the value of a win if you did nothing? What is the value of a loss if you did everything?
My challenge and debate is, how would this ranking system look if stats sometimes would even overrule the âwin/lossâ value we hold dear.
Think about it, There are some undeniable stats that showed you did absolutely nothing to contribute but if the team wins you still get that sr. Would people play more selfishly? Maybe, but would they play MORE than whatâs happening now?
This is a bigger issue the bigger the team is. I think weâre going to see a major shift in tank skill come OW2 because it will be glaringly apparent where a tank is placed. An underperforming DPS can be carried and enabled by a solid team, or vice versa a good player can be dragged down by a bad team.
I think the shift to 5v5 will solve a lot of this issue, but we should still retain PBSR.
Honestly there is a reason personal performance stops mattering once you hit diamond. Itâs because no-one cares what you did in that game you lost. You lost and thatâs it. If you play better than your conterpart then you will win more games than lose.
Stats very often donât tell the whole story in a game as complex as overwatch. For example an echo being enabled by 4 other people will have the best stats in the game but does that mean they should receive more sr than the rest of their team. They played in a way that got them a win so donât they deserve equal credit? Many things that influence a win simply cannot be detected in stats, for example comms (shot calling can easily win games).
This system would basically encourage these people who are good at shot calling and enabling teammates to mute their mics and focus on playing well individually since that will reward them in the long run.
Thatâs the idea. The game should place more emphasis on accurate stat recording, as is evident with the frequency of Play of the Games where DPS dominate the majority of them, with the occasional heavy-hitting tank. Supports are virtually unseen in more than 90% of PoGs, because their support for the team isnât counted towards them.
Nano a Deathblossom or Genji Blade, while sleeping a diving Winston and nading the entire enemy team, allowing your DPS to secure those kills? PoG will go towards whoever got Nanoed instead. Or pocket-heal a squishy or tank on the verge of death and turn the battle around in your favor? Itâs as if you did nothing.
I think much of the issue is how stats are recorded, and the stats that players can see donât tell much of the story anyway. In a sense, the blue card stats at the end of the game do a much better job at demonstrating a playerâs performance, and if stats like those were visible instead of what we currently have in OW1, this might promote better team cohesion and individual awareness.
For instance, if a player is able to realize they only have 10% kill participation (or otherwise) and can adjust their playstyle (if they can help it).
Otherwise, in discussions like these, the pendulum tends to swing too far in one way or the other. A more generalized approach, imho, is to provide both players and the matchmaker higher quality information from a playerâs performance, so that the player is better represented on the ladder and is able to level independently of their team.
Make stats reflect better team play, quality over quantity, and you solve both problems with a single solution. Farm stats by actually helping your team.
Otherwise youâre just ignoring a perfectly valid solution, for some reason.
Better stats are needed. If gold medals mean little to nothing then show us the stats that do actually matter in a game. Probably the AI is not advanced enough to track such stats like Nano alluded to. So in plat and below they are basing how much SR we win and lose on some probably irrelevant stats that we donât even know.
I personally donât really care about that part of it anyway. In fact, I would rather gain a smaller amount of SR for a win, since that means my MMR likely barely changed at all. Which means my next game will be easy. Losing on the other hand, sucks when you put up good stats because your MMR doesnât lower enough, so you get equally difficult games even after youâve already lost 10 in a row.
The issue with this is some stats are âintangibleâ such as zoning which can be done by many heroes:
tanks can âmove forwardâ (Rein holding shield a good example) to pressure squishies backwards
spam heroes (Junk, Pharah, etc) can attack a choke, hit nothing (lowers accuracy stat) but make enemies play more defensively
all heroes can deal less poke damage (lowers hero damage stat) but deny enemy support charge
a flanker spawn-camping a support, forces them to hide in spawn but it hurts all stats for the flanker (as under the current stat system theyâre âdoing nothingâ) but provides a huge benefit for their team
Itâs easy to say âmake stats reflect better team playâ but I donât think youâve considered how that translates to a statisticâŚ
final blows, ult shutdowns, less deaths, ult elimsâŚit isnt hard
think about the games where you had massive sr gains (that didnt result from a brand new acct) and you will notice that not only did you die fewer times than your normally do, you had higher ult elims, more final blows (ie: 60%+ of elims are from final blows) shutdown an enemy ult (many many ways to do this)
do these things and on WINS you will see larger sr gains. stats do matter, its just the ones no one ever looks at
I remember in one of my games when i took this game seriously, i was playing anubis temple; i slept a pushing bastion who just used teleporter then nanod our rein when he had maybe 1/4 health , saving us from them getting the point . I thought it was a hell of a play , not to mention point saving . The pog was the rein getting nanod , holding primary fire and getting a quad kill while i pumped him with heals. Ive had plenty of great plays on support , but for some reason , i was hella salty about that specific play of the game .
There are a few cool stats that you canât get anywhere except requesting your data via customer support. They have useful things, like map specific W/L ratios, which cannot be tracked anywhere else.
I think a win/loss per map per hero would be incredibly useful for people, and probably a bit enlightening. As well as losses vs a given hero.
Ok then replace the medals with those stats instead. Although I think it would still be a lot easier for certain heroes to get better stats in those categories as well. Moira with final blows, or any character with cc will be easier to shut down ults than a Symmetra for example. Then you have the other issue of if you shut down an ult with a Sym or Mei wall, will the game even count that as a shutdown the same way as a McCree stun?
Regardless, I want everyone to be able to see the stats that matter so itâs obvious who is over/under performing and you donât get mocked for saying âI have 4 goldsâ when trying to defend yourself from throwing accusers.
I donât get âmassiveâ SR gains unless I was playing in bronze or something. No other rank than bronze or GM would I get more than say +30 SR for a win or lose more than -30 for a loss. (I have played in them all except GM so Iâll give the benefit of the doubt on that one.)
Even then, these stats are both qualitative and quantifiable.
Easy metric. Units move forward into âenemy territoryâ and the âlineâ held by the enemy team is moved backwards as they advance. It might take a big brain to work out all the math, but thatâs why Blizzard devs get paid the big bucks.
Oh, and Splatoon already does this metric exceptionally well.
Goes back into the âdefensive lineâ metric I mentioned earlier. What a team does in response to player activity can be quantified, and in ways that arenât immediately understandable to the individual.
What I am saying, optimistically here, is that developers can implement tools that are capable of extracting this kind of data from the match metadata. The easiest of such implementations is â% kill participationâ or â% damage takenâ (healer), as two examples of qualitative data.
Thus, a âfeedâ mechanic would be useful. More reasons to improve the game!
This ties into the previous solution. Giving this a modicum of additional though, perhaps a stat bonus can be recorded whenever a player is behind enemy lines and provides value of some kind.
These kinds of solutions can be implemented, but they are a very high level concept.
Guys thanks for taking this topic seriously. Iâm always the guy who asks the âdumbâ question but realize later that itâs not dumb itâs just the harder road and more work.
To add to âspaceâ I was thinking⌠birds eye View, and grouping mechanics of the teams. But donât think about run it thru ML of games where a tank too space and where a tank did not. Iâm SURE ML could figure out on new cases definitely yes or no.
Because what I think what weâll fined, the maps, maybe beautiful but just like a chess board there are only a limited ways to go. So we may think OW has âinfiniteâ outcomes and such but itâs the micro algorithms maybe not that many.
It would honestly be a bit better if actual developers participated in conversations like these.
In many ways, the community might not properly address problems or issues that a developer might have a clear image of. Some issues canât actually be addressed. Other issues might not fit the design of the game.
And some issues raised by the community are valid and just need someone on the developer team to vouch for it.
Well, generally, as a coder myself, usually algorithms arenât necessarily a one-size-fits-all solution lol. Thereâs typically a methodology for problem solving.
For example, building tools that can record âspace makingâ stats for tanks or âdeaths preventedâ stats for supports. This could even extend to âflankingâ stats for dive and flank characters.
certainly, smaller chips can generate more data points. Totally agree. More data points fed into ML can make a world of difference.
Honestly thats why I loved the scoreboard - give me more data so we can figure something out.
What I dislike, is when people throw up their hands and say âits impossibleâ. Sometimes we play these games and we think the human is so complicated and this game most be super complicated. But like removing a dimension, birds eye view, this game gets dumb simple really fast.
