[WBMM] Discussion & Suggestions

[Golden.]

Tbf whenever you lose you only lose like 10-20 mmr if you don't have an extremely terrible score. Whenever you win, you can gain like 20-30 in one game even if you're around the bottom of the scoreboard. You simply gain more for winning and playing mediocre, than you lose for losing and playing mediocre. If you play well you can gain 60-70 mmr in one game, and you only lose like 5-10 mmr if you lose badly.

I often gain MMR when I lose as well

[maskmanmarks]

Tbf whenever you lose you only lose like 10-20 mmr if you don't have an extremely terrible score. Whenever you win, you can gain like 20-30 in one game even if you're around the bottom of the scoreboard. You simply gain more for winning and playing mediocre, than you lose for losing and playing mediocre. If you play well you can gain 60-70 mmr in one game, and you only lose like 5-10 mmr if you lose with a decent score.

Edit: this can serve as an example:

played like shit and lost (-15)

[close]

played like shit and won (+21)

[close]

If getting 650 damage constitutes as playing like shit then you must speak to my good friend Fenton who got 39 damage in a 19 round game...

[Vegi.]

Tbf whenever you lose you only lose like 10-20 mmr if you don't have an extremely terrible score. Whenever you win, you can gain like 20-30 in one game even if you're around the bottom of the scoreboard. You simply gain more for winning and playing mediocre, than you lose for losing and playing mediocre. If you play well you can gain 60-70 mmr in one game, and you only lose like 5-10 mmr if you lose with a decent score.

Edit: this can serve as an example:

played like shit and lost (-15)

[close]

played like shit and won (+21)

[close]

If getting 650 damage constitutes as playing like shit then you must speak to my good friend Fenton who got 39 damage in a 19 round game...

You call Fenton a legend now

[DayBoul]

Tbf whenever you lose you only lose like 10-20 mmr if you don't have an extremely terrible score. Whenever you win, you can gain like 20-30 in one game even if you're around the bottom of the scoreboard. You simply gain more for winning and playing mediocre, than you lose for losing and playing mediocre. If you play well you can gain 60-70 mmr in one game, and you only lose like 5-10 mmr if you lose with a decent score.

which proves that their formulas are unbalanced and probably gave a winrate value equal if not higher to that of KD, while also ignoring damage. That allows low to average players to still grind just by spamming games with a consistent negative KD and having better players carry them.
I dont think ive ever seen anyone get more than -30/40 even with a particularly awful score while getting +60/70 is much easier

both tks and KD should be separate values, individually lower than winrate but higher if combined. Damage should also be included to prevent killsteal machines from getting easy elo e.g Faisan Wursti Zeyden otherwise that would mean there is literally no assists system

P1 and P3 with same score but 200 dmg disparity

[close]

P1 and P3 with same score but almost 300 dmg disparity

[close]

[Tardet]

The sole reason why the complete formula is not displayed publicly (like any proper MM system) is because LP and his team are perfectly aware that any given cretin will spend their days and nights complaining about it and offer so-called 'better' systems to justify their loss of MMR or bad rank.

Even if in your case, there was to be a well-thought and argument reasoning that could lead to the improvement of said system, most people will just disregard it without even using their brains, just to be a pain in the ass. That's almost a guarantee.

Obviously, you could make the argument for LP to simply ignore people who clearly can't comprehend how MRR should theoretically function but if you know how many messages he receives on a daily basis, you will probably understand he is not particularly looking to fuel a topic which will inevitably lead to a large increase of his voluntary workload.

Coco or anyone else can be Lord on WBMM and still be absolutely dogshit at the game, the rank displayed there doesn't mean crap and even if someone playing bad still manage to climb the rank (which highlights a possible issue in the system at the very least, I will give you that), it's probably not worth to go through the trouble of starting a discussion which inevitably isn't going anywhere. We already had arguments such as this one when WBMM was released on native and I don't think they contributed to changing anything in the MMR system in a drastic fashion.

[Janne]

i wish i only lost 8 mmr but im dog shit

[DayBoul]

@tardet the formulas are already live, most of the work is done. All you need is tweaking the values of winrate and KD, even without including damage that would work. Wouldnt be perfect but still very much OK
It also looks like they did tweak them a bit in the opposite way before the recent opening since there were many more jokers the first time, including some who are now lord, knight, master knight (coco, vegi, maskman ect) ; i have a hard time believing all of them just happened to collectively improve their game. Assuming they did change it, surely reverting them wouldnt be that hard

To fully fix it you could copy the chess elo ranking system and implement the relative strength of the players to gain more or less MMR depending on who's in the team you beat and who you kill. That would nullify the excessive mmr shit players get from getting carried or duoing aswell as deny noobslayers and ks machines from underserved mmr
The chess elo system is pretty much without flaws so again, all you'd need is adapt it to a team game.
You see, that i'd totally understand how it wouldnt be worth the hassle but what i mentionned about winrate/KD values really isnt a massive amount of work

Obviously if lapache clarified that changing it isnt an option this discussion wouldnt be fueled at all but he never clarifies anythign so yeeeee 

[Rikkert]

The system is flawed regardless. In a close game, someone holding a 2v1 for 10-20 seconds can win your team a round and possibly the game. This is more valuable than someone getting 2 kills at the end of the round when you're already up 5v2. However the person holding the 2v1 won't get any points for blocking and holding the 2v1, but the person getting the easy kills in the 5v2 will get points for stealing two meaningless kills. You also won't get any points if you're the one stabbing first and hitting your opponent's block in 2v1's, but your teammate last-hitting the enemies will. Nor will you get any points if someone lunges in at you and you block it to stun the enemy, providing your teammates with an easy kill. All of these are things that are not really quantifiable for a points system such at this. You can't really start giving points for the amount of stabs and/or blocks people hit. You can't really give people points for "proximity to the kill", even if they might be the reason the enemy dropped their block. The system will forever be flawed, and that is fine. However that does not mean that any and every criticism should simply be ignored. The system could definitely be improved upon in some areas. Currently, a first-pick and a follow-up kill that might win you the round is worth as much as the guy that manages to get two last hits in the 5v2 at the end. Similarly, clutching a 1v3 will get you the same amount of points as running between 2v1s and getting 3 easy kills. Currently, a kill on Python is worth as much as a kill on forever_joker. These things could definitely be improved upon, although I don't know how practically feasible they are, given that I know fuck all about scripts.

[Janne]

Currently, a kill on Python is worth as much as a kill on forever_joker.

where problem ?

[Tardet]

Spoiler

@tardet the formulas are already live, most of the work is done. All you need is tweaking the values of winrate and KD, even without including damage that would work. Wouldnt be perfect but still very much OK
It also looks like they did tweak them a bit in the opposite way before the recent opening since there were many more jokers the first time, including some who are now lord, knight, master knight (coco, vegi, maskman ect) ; i have a hard time believing all of them just happened to collectively improve their game. Assuming they did change it, surely reverting them wouldnt be that hard

To fully fix it you could copy the chess elo ranking system and implement the relative strength of the players to gain more or less MMR depending on who's in the team you beat and who you kill. That would nullify the excessive mmr shit players get from getting carried or duoing aswell as deny noobslayers and ks machines from underserved mmr
The chess elo system is pretty much without flaws so again, all you'd need is adapt it to a team game.
You see, that i'd totally understand how it wouldnt be worth the hassle but what i mentionned about winrate/KD values really isnt a massive amount of work

Obviously if lapache clarified that changing it isnt an option this discussion wouldnt be fueled at all but he never clarifies anythign so yeeeee 

[close]

I went off the information that (based on your previous messages and what I knew) the formula was at least partially hidden. If it's mostly public, then scratch what I said about it, obviously its no longer relevant.

The elo chess system is indeed a good example of a valid system that has made its proofs but then again as you mentioned, I am not sure the WBMM admins would be arsed to implement this in a way that makes it work. At the end of the day, it's mostly In LP's hands and I don't want to talk about what could or couldn't be done as I don't know shit. I guess I will let you await for his answer, whenever that is ^^

[AccursedGull]

the point system is not fair right now, and will never be 100% fair and representing the impact and skill of an individual. It is simply not possible because there are so many criterias which can impact the game to your teams favour. The current stage of the point system is good tho, probably wasn't easy to do

[DayBoul]

rikkert and gull its obvious the system will never be perfect but there is a bare mininum and the bare mininum is what im asking for

easily half of what you said is a fantasy and probably impossible to implement due to NW code limitation. Tho i dont see why implementing a system of value for kills wouldnt be possible, formulas for the chess elo ratings are literally free to use by anyone on internet and they have a relative player strength system but since chess is a 1v1 game the challenge will be to adapt it to a 6v6 game. Other games proly have already done it but as tardet said, lets not forget this is some random 10yo 200players game and its not worth the hassle

again its not about pushing the formulas to their limits and shit, just following common sense. Nothing justifies giving winrates such a high value

Also @tardet by "live" i meant "in use". They're not public

[Vegi.]

Dayboul joker confirmed?

[Arch4ngel]

uf community hates u then u will be in weak team

[Scottish Unicorn]

Wish I only lost 20 MMR

[close]