Originally Posted by
willtron
Which is why a Tracers stats will be compared to other Tracers within a given ranking range, and not other heroes. That way it doesn't matter if you flanker vs a direct damage DPS dealer. Same with objective time, it doesn't matter in terms of an absolute value, just as a relative value to other Tracers within your comparable bracket.
So just as a super simple visual example - say as a Tracer in a game you have 30 kills and 1 minute objective time. Stats show on average at your current rank of x, on that map Tracer should be getting around 23 kills and have 1:30 objective time. So you're above on kills but below on time on the objective.
The rating calculator will use a coefficient that is favourable in kills but slightly penalising in objective time. So if 23 kills and an objective time of 1 minute are assigned a value of 1 each, then 30 kills could be 1.1 and the objective time 0.9. Also factored in are other tracer specific factors - times blinked, health recalled, metres covered etc.
Again, hero specific calculations, not a global system - looking at average stats will eliminate the problem of x hero vs y hero, so yes, Roadhog, sure kills and hooks would be a major consideration rather than DVa which would consider her strengths and abilities, Reinhardt's shield uptime and damage blocked will be more of a factor than Winstons. Why is it unrealistic and almost impossible to calculate? It just seems to me you can't wrap your head around it so you declare it as impossible.
So players would find a way to play their hero well and be rewarded for it? That's sort of the point I'm getting at. Any system is abusable if someone tries hard enough, as long as the ranking equation is balanced then it shouldn't be all too abusable, if at all. How do you envisage something like this being abused? Play Genji and just farming kills while flanking and your team loses? Great, you'll be recognised for the kills, but objective time and the loss will work against you, and the stats compared to other Genji's at that level will dictate how rewarded you are for the kills and how penalised you are for the objective time.
Then just account for team play by assigning a value. So furthering my last example, because 25% of Genji's kills were team based and the Genji average for that map is 40%, the team play factor will be slightly below the average, working against him.
I disagree and I think penalising players for the matchmaking system grouping you with random players and losing even if you play incredibly well yourself is a halfhearted design.
What I think isn't feasible is your personal matchmaking rating being so heavily tied to random people. I'm proposing the team aspect of that takes a backseat and individual skill based on the meta-statistics for that hero are used.
- - - Updated - - -
Which is part of my point, if you were good enough to get 3 gold medals then you were either incorrectly matched with people on your team, a failing of the matchmaking and ranking system - one of my greivences.
So yes, it does say something about your team as a whole, if you as a poor 76 player could get 3 golds, then either you were above them, or they were below you as a standard.
Under my proposed system, depending on how you did vs other 76s within your ranking range it would calculate how well rewarded you were.
I don't know if you won or loss that game but for the purposes of my example I'm going to assume you lost. Under the current system, you'd straight up lose rating. As Endus said, you might not lose as much based on your relatively better performance than your team mate but you still lose rating.
Again, another assumption - I'm going to say you're rank 2500. So here's a super bastardised version of it all:
The first thing looked at is your stats versus 76's between a rank range of say 2400 and 2600 for example. On the map you played, you scored below average on accuracy, below accuracy on kills, above on healing done, above on objective time, neutral on helix kills.
So from the off, you'll be unfavourable in accuracy and kills, but favourable in healing and objective and neutral on helix kills - how much is dependent on how far from the regional average you are.
You'd get a small bonus multiplactive coefficient for the gold medals because you're better relative to your team. so think 1.00 for nothing, 1.25 for 3 golds.
And teamplay would be look at, so lets say as 76 because you did a lot of healing to team mates that is looked on favourably, as is your assist kills and kills within the objective.
Then finally the loss coefficient is put in.
Ultimately because you did well on a losing team, you'll still lose rating, but on ~5. Because your team mates were much more below the average in that game, they lose more rating.