The last days I checked who bans in UD, which means has the highest MMR in there.
One example:
Guy banning: 4037 games UD, 49,7% WR
Me: 3336 games UD, 51,2%WR
Why is he banning?
Because winrate doesn’t reflect overall MMR.
PLEASE NOTE, this is a very, VERY, basic look at MMR. And is not the actual way it is tracked in HotS (at least, I hope not)
To make it easy to talk about, we’re going to say “the guy” started at 2000 MMR, and you did as well.
I’m assuming these games are all in UD (unranked draft)
So, say a win with even teams is worth 20 MMR.
Say, over his 4037 games, he has won about 2006, and lost 2031. Say on average the other team was favored to win a little, (+21 average MMR, instead of 20). And the losses average out to the “normal” -20 MMR per game.
Over these 4037 games, “the guy” would have gained a lot of net MMR, 1506 to be exact. Bringing his total MMR to 3506.
Now say you’re in a similar situation, except the numbers are reversed. The games you win, on average you were equally favored to win as the enemy was. Gaining the “average” of 20 MMR. But you were favored to win the games you lost on average. So you lost 21 MMR per game.
Despite 1738 wins to only 1598 losses, you would have only gained a total of 1202 MMR. Meaning you have less MMR than “that guy”.
Wrong, thing is the other guy probably has more/better WR in other game types, don’t think MMR is specific per game type despite HotsLogs giving us a different picture
Which probably explains smurf-MMR-boosting for “new” accounts phenomenon that works at start for a while ?
Make no mistake though, MMR (at this point in time) is only and exclusively determined by WR% (or more specifically think it’s more like by winning/losing streaks [my assumption])
But yes = that is the case and hence why everyone hoped/wanted to see PBMM in action as a result…
As far as I’m concerned = they better completely drop off MMR and “redraw/rething/redesign” the whole thing now
1 Like
But it is per game mode.
https://heroesofthestorm.com/en-us/blog/21771817/
Let me break this down to simply text:
You could skip Hero League for a season. This would rely on taking QM/URD MMR as a baseline for reseeding. This was changed. Now it will just take your season old Hero League MMR.
Hence, the MMR values for QM, URD and HL are separate. Presumably TL is also a different MMR from HL.
While my post above is a very simplistic take on MMR, and it certainly not how Blizzard does it… It is not purely based on winrate.
Blizzard’s internal systems likely have a “slightly favored” or even “favored” tag they can apply if they believe one team is better, which would result in that team gaining a lower amount of MMR points from winning a game.
And that means playing against a “favored” enemy team would also reduce the amount of MMR that you lose.
This is a really basic feature of an MMR/ELO/similar systems.
You mean reset it for everyone?
Whenever they reintroduce PBMM I could see a loosening on MMR. But I highly doubt we’ll ever see a reset.
MMR comparison (totals & averages) DOES impact the final rank points gain, but it’s STILL based only on WR%, win game ? = get MMR points, lose game = lose them
So yes, there IS the favoured adjustment number/s but it’s based upon nothing else but MMR total/average on both teams, nothing else
There is PRA as well but it’s a complete mess that I doubt even they themselves know how it functions, but bottom like probably is = even if you lose game, just don’t die is the “message” they try/tend-to share ?, I honestly don’t care, lose by 1 point/death/level/structure is still a loss, very same as if lost by 10 because at the very end again: the MMR impact on MM for next game cares ultimately/only about only WR%
That is why PBMM was a must-have (and sounded like a popular/accepted idea to the community) tbh, people want to know their PERSONAL IMPACT toward the loss/win, and not only that but also have factors/categories that they themselves (the player each of their own) can IMPACT toward the final number of MMR they get
But Blizz changed their minds for some reason, gave up away. Folk tend to say it was bugged/bad, but I tend to think they realized it was a far greater challenge than anticipated (or not profitable enough whilest requiring serious hardwork and lots of commitance) I’d say
That is fine but again = the MMR impact on MM for next game cares ultimately/only about only WR% , it has however an “accelerate/decelerate” to it as well [hence why I said win/loss-streaks, although Blizz never officially confirmed this]. Win more games in a row = gain MMR faster, that’s how they “pump up” smurfs faster toward higher ranks (heck they even bragged about it last Blizzcon or the one before ?)
No, I mean GET RID of MMR completely and make MMR-less ranked [read second part from document, but would vouch try read the whole thing, there’s quite a bit useful info I summarized there]:… https://1drv.ms/w/s!At05MlSwg8Tcf6lfFccrnU1RALU
Ultimately I posted/summarized-thoughts/analysis about the “history” of the problem, some “mild measures” (toward the end of Part 1) that could’ve maybe saved the salvaged/situation a bit, but at this point have to be RADICAL (part 2) to make a change:
And here’s the forum thread: How to Fix?/Save RANKED PLAY in Heroes of the Storm (Documented version)
Bottom-line is: either PBMM re-release, or complete revamp, the game’s in a state that the MM requires RADICAL (i.e. game-definition-founding change)
One-time MMR reset would do no impact what-so-ever and just a matter of time things get bad/worse again, instead = need get rid of the BACKGROUND REASON FIRST that created the problem, not just mild-down the consequence/s
1 Like