match outcome = win or lose
skill difference: lets say radiant's 5 players have a DBR that averages at 1600, and dire has DBR that averages at 1700. Then there's a average DBR skill difference of +100 to the dire. Key thing about how this work is, if radiant wins this game, they gain more DBR as oppose to when dire wins it, because they have defeated opponents which is "stronger" than them. and each player receives a different amount of increase, this increase will depend also on the player's individual DBR.
Below is my hypothesis on how part of the calculation works beside only considering the average DBR difference:
Continueing the scenario from above for example, player 1 in radiant has 1500 DBR, whereas player 2 has 1800 DBR. some kind of formula would work out considering that these players worked together to achieve victory, and give them DBR increase respectively. From player 1(1500) perspective, he defeated oponents that average with more DBR than him, so he will gain more DBR, as oppose to player 2(1800) who faced opponents with lower average DBR than him.
vice versa for the losing team.
since dotabuff currently keeps DBR private I cant illustrate this here so, check this dota1 ranking for example, you can click on individual players, check their match history and see how much "PSR" they gained/losed from matches: http://www.lagabuse.com/stats/
bear in mind that the ladder above has formula for rating calculation is different to dotabuff's.
@victornkc - nice write up. And everyone praises how good those DotA 1 leagues were - obviously the public ranking played a role. Add 1.900.000 spineless creatures to the mix and suddenly it's all about privacy, my data, your data... sad
Privacy comes first, otherwise Valve would have come under criticism for an inferior customer relation
What fucking privacy is everyone harping about. You are a fucking song name with a with a lightning instead of a face and some delusional text over it? Are you ashamed of you anonymous Steam avatar's performance in games, so you would like to hide your anonymous match results? We don't know who you are, where you live, your name, your email or anything even remotely connected to you. How exactly is the match result of game that you participated in with 9 other people, that is already publicly available to see to 3 million other people a private thing?
NOTE: please read the entire thing below that i have stated!
Also, everyone has agreed to the terms below. Privacy does not matter, what information would be given out from a game like dota 2? The amount of games you played so?
Spineless players will always be spinless.. and pussies..... very tight pussies...... so tight that my little brother's 2 inch wiener can't even penetrate... :)
5. THIRD PARTY CONTENT
In regard to all Subscriptions, Software, and related content that are not authored by Valve, Valve acts merely as an intermediary service provider. Valve does not screen such third party content available on Steam or through other sources. Valve does not assume any responsibility or liability for such third party content. Some third party application Software is capable of being used by businesses for business purposes - however, you may only acquire such Software via Steam for private personal use.
6. USER GENERATED INFORMATION
"User Generated Information" means any information made available to other users through your use of multi-user features of Steam or to Valve or its affiliates through your use of the Software or otherwise. User Generated Information may include, but is not limited to, chat, forum posts, screen names, game selections, player performances, usage data, suggestions about Valve products or services, and error notifications. Subject to the Valve privacy policy referenced in Section 1 above, as applicable, and except as otherwise specifically addressed in any Subscription Terms or Rules of Use, you expressly grant Valve and its affiliates the non-exclusive, irrevocable right to use, reproduce, modify, create derivative works from, distribute, transmit, broadcast, and otherwise communicate, and publicly display and perform the User Generated Information and derivative works thereof in any form, anywhere, with or without attribution to you, and without any notice or compensation to you of any kind. For any content for which you take advantage of cloud storage provided by Valve, you grant us a license to store such information for you, and you represent and warrant to us that you have sufficient rights in such content to grant us this license. You acknowledge and agree that we may place limits on the amount of storage you may use.
I just dont know how this system is good for solo players like me(im playing solo 95% of time), im currenty at platinum league with 1597 DBR, but 3 of my friends who usually plays in stack(95%) are in diamond league, and im way over their skill.
I dont know whats the reason but maybe that "skill difference" factor, cause in MM if you play as stack, you will probabbly get stack as enemies and maybe DBR is calculating higher values in that way, so progresion is faster. Cause other stack can have way bigger DBR and if you win against them you get bigger bonus, or vica versa. Did you thought about this?
And also, KDA ratio should be considered in calculation, but should not have big infulence in the final rating, id said KDA ratio should have 15-20% of the overall calculation, specially if you are going to separate solo players and stack players later on :)
"cause in MM if you play as stack, you will probabbly get stack as enemies"
I don't think that's true. Been playing a few games (< 10) as full stack and only got random people (or horribly bad stacks, but I doubt that).
Also I don't think KDA should be considered because it strongly depends on the hero. A good Chen or a good KOTL don't need high KDA's while good carries do. This would most likely lead to having everyone (everyone who cares, however) play carry, killsteal and selfish.
I agree that some support players would try to ks, and some support players really dont have ability to kill, unleas they get lucky last hit, but thats why i said it would have low percentage of overall ration, but ok.
And about stacking players, well most of the time when i do play with friends(3 or more in team), we get enemies as stack 90% of time, maybe you are playing with just 1 friend or i dont know...
Well actually I get matched up vs full stacks quite often while I play solo random. So there's something wrong with Valves MM imho. (That's a different topic though)
I really want to figure out how this works, I have a friend who is higher dbr than me yet my win ratio is higher and I've played a lot more games than him. He usually only plays in a stack where as I often solo queue (about 90% of the time). I'm 1555 and he's 1622, there must be some relation between playing in a group as opposed to playing alone... Also if your DBR doesn't change does that mean you didn't get points because you played against players with lower DBR than you?
კომენტარის დასაწერად გაიარეთ ავტორიზაცია.
"CALCULATION FACTORS
The only factors considered in determining the change in your rating is match outcome and skill difference between players. Other factors such as player KDA are not considered. We feel very strongly that including KDA (or other in-game statistics) would lead to a system that encourages individual play (instead of teamwork), is difficult or impossible to balance, is subject to exploiting and would lead player ratings to drift away from the in-game matchmaking."
Can someone tell me what "match outcome" and "skill difference" is and how they measure it?