Page 13 of 34 FirstFirst ... 3 11 12 13 14 15 23 ... LastLast
Results 121 to 130 of 334

Thread: feedback on matchmaking after the lastest patch

  1. #121
    Quote Originally Posted by Miedoduck View Post
    Hey,

    I'm not a dev but looking at the match statistics I can tell you what probably went on there from a matchmaking perspective, VH, High and Normal aren't brackets, they are just the skill level of all 10 players combined. In this case, I would say queue times probably had a bit to do with it too.

    You are probably the most skilled on your team, however the Razor on their team seems equally skilled (maybe slightly less then you). If you assigned letter grades I would say

    Morph -- A
    Lion -- F
    Tidehunter -- D
    Jugg and Leoric -- C

    Their team
    Razor -- A
    Rest of his team -- lower end of C

    So you could claim that the MM was uneven in your favor, the queue time was getting pretty long so MM didn't give you the most optimal solution, but a good enough one. With that though, Lion may have just been having a terrible game, everyone has them, when shit just doesn't seem to stick, you make a mistake you otherwise would never have. His item build of your entire team makes the most sense and looking at it, the Dire had better builds, if MM was completely even I would say they should have won.
    The razor was pretty bad. The only reason we won is because Razor did absolutely nothing in the mid game teamfights. He had Deso and Yasha as his items for the teamfights. He had been fed top but only had 950 hp and didn't understand how to maneuver in teamfights. We were able to comeback because he was squishy enough for me to zone him out. I went back to check the replay for the game and discovered that Jugg and Lion had been constantly diving the top lane, leading to a 1-14 score between them at 9-10 minutes, both at 7 deaths. The Crystal Maiden was actually ahead of our entire team in networth for a decent amount of time. Jugg claimed they had been dived on, but the replay said otherwise. Endgame item possession doesn't really tell the entire story. The Razor has gold because he had Jugg and Lion throwing themselves at his tower. He didn't play well in teamfights nor did he spend his gold wisely. I actually rallied my team by pointing out that Doom had no teamfight presence and Razor had 0 survivability items, just Phase/Yasha/Deso. SK was decent, but Lion and Jugg both suicided a ton and the other team clearly had no idea how to utilize the gold advantage. They were 15k gold ahead at 30 minutes and then got throned at 39 minutes. The other team was 7k+ gold ahead the entire game and yet they milled around doing nothing and losing any team fights. I think you are basing too much of your rating on K/D and farm amount. In HoN, I've seen games of 10 1400s (1535 is 50%) have 1 of them go a score like 25-2. K/D is relative.

    I am aware that the bracket is an average of skill level. The question is just how much a skill difference does the matchmaking allow in a given game. When 85% of my games are very high, it feels imbalanced when I'm put in a normal level game. It ends up with a game that is just not enjoyable to most participants.
    Last edited by AtheistGod; 07-19-2013 at 09:55 AM.

  2. #122
    Basic Member
    Join Date
    Nov 2012
    Posts
    1
    need statistics like this

    Attached Images Attached Images
    • File Type: jpg 1.jpg (73.6 KB, 7 views)

  3. #123
    Basic Member
    Join Date
    Nov 2011
    Posts
    549
    The game is called Defense of the Ancients...

    ...not Inflation of the Numbers.

  4. #124
    Basic Member
    Join Date
    Nov 2011
    Posts
    154
    Quote Originally Posted by eldorquO View Post
    I play a lot of dota1 on a chinese platform (I live in China). On this platform, it forms your overall ranking using a combination of your won/loss and your ranking with each hero. The hero ranking is based on your stats (kills/deaths/assists/creeps/notsurewhatelse) compared to what most players do WITH THAT HERO. So if you're on a hero that doesn't generally get good k/d, it doesn't matter because you are compared to the average on that hero, not the average overall. and it's adjusted heavily in favor of whether you won or lost (so it's hard to be rated high with a hero if you always lose when you pick them, even if your stats are good). If you are ranked as being far above the average with a hero, that gives you a bigger boost to your overall ranking the farther you get from average, but only for people ABOVE the average. It doesn't penalize you for sucking with a hero except that you are more likely to lose the game. There are definitely some issues with how they create the hero rankings, and it's far from perfect, but one thing that is nice is that it's hard to be rated very very highly without being at least decent with a good percentage of the heroes in the game. To raise your ranking, you have to play a variety of heroes, and show that you aren't terrible with them, and if you want to raise your ranking a lot, you have to show that you are exceptional with a few heroes. I like this system, because a good player should be able to do well with most heroes compared to a bad player, and a system like this would prevent issues where a new player only really knows how to use 5 heroes, but picks them every game and does well with them, causing his overall rank to get too high (so that when he picks a hero he's not intimately familiar with, he is awful and gets flamed). Some sort of MM using a system similar to this I think might be more accurate, and it generally separates "experienced" players from new players in a reasonable fashion. I think valve could do it better than the chinese platform does. it's a good idea but like I said the implementation has some issues.

    Before the match, the platform gives the average rating of each team (which is usually pretty close). So all you know is the average and what your rank is relative to that (unless you want to pay them money, in which case you can get all sorts of random info about everyone in game; I have never paid them). After the game is over, however, they show every players rank PLUS what their rating was with the hero they picked that game. Which explains a lot of onesided stomps, when you realize that everyone on your team picked a hero they suck with, and everyone on the other team picked heroes they are good with. Even though the overall rank is the same, there's not much you can do about a situation like that. Since the game is over, giving this info doesn't result in any flaming or anything, and there's no practical way to save it (although on this platform it is actually publicly available, although it wouldn't need to be). Maybe only I like this, but I think it's a good idea worth considering.

    Edit: just for the record, I have had no issues with the matchmaking in dota2 so far (albeit in only a few games). Generally when I do well I win, and when I do bad I lose, and there's usually a good reason for me doing bad (like playing like an idiot, or bad hero matchup, or something like that)
    This approach is very interesting. I especially like how it considers more detailed data when determining your skill. I think the more data the matchmaking considers, the more accurate it can potentially be. I agree that people tend to play better or worse when playing certain heroes.

    Showing each player's skill on heroes they picked might be crossing the line a bit. You will get people blaming others for picking heroes you're bad with. I don't think players should be bashed for picking heroes they want to try out. How else will you get good at that hero other than playing it?

    I hope Valve looks at this idea carefully, and think of possible implementations.
    Last edited by 100kV; 07-19-2013 at 10:07 AM.

  5. #125
    Basic Member
    Join Date
    Jan 2012
    Posts
    82
    Quote Originally Posted by fletcher View Post
    Yes, we use a Elo-type system. We have data that shows that it correlates well with a number of quality metrics.

    Showing players their Elo has proven to cause a lot of unwanted behaviour. (The same problems exist in Chess.)

    We know that there really is value in knowing how the match was formed, and even in knowing which players were the best, and also who was partied up with whom. It helps to analyze your own skill and try to understand how to get better. We are actively working to try to figure out what data we CAN provide to players (perhaps after the match) to help them do this analysis, without causing too many of the undesirable effects I mentioned. Right now, if there's a steamroll, people look at the only data available to them, which is totally understandable. Also, I believe that there have been problems in the past where it was not doing a good enough job segregating truly inexperienced players from experienced players. We've improved on that. However, our definition of "new player" probably doesn't match exactly with many players. (Their definition is often: "the other guy has fewer wins that I do.") And we have to deal with smurfs and players coming in from other games in the genre, where win count is definitely totally unrelated to the skill level of the player.
    Stats are great, but players don't need to know anything about MMR (either theirs or anyone else's). They also don't need to know anyone else's stats outside of that particular game. Each game is an isolated event, and even if the game was matched poorly, you have all the information you need to analyze what went wrong. A player plays well or they play poorly. There are different aspects to play (how well they farm in certain points in the game, map awareness, mana/cooldown management, coordination) and those are visible in replays. Players have everything they need in viewing a replay. All we need is a better digest (WC3 end screen?) for each match so we can see the really glaring problems at a glance. Replays will always be the way to truly find out why a game went the way it did. Players asking for information on MMR or how the matchmaking system works are just looking for a scapegoat. Don't feed this idea.

  6. #126
    Basic Member
    Join Date
    May 2012
    Posts
    5
    fletcher: Match 246699658

    I would really like to know what the hell happened to that tinker and juggernaut. They were some of the worst players I've seen in very high. I'd appreciate it if you could give me a bit of info. The jugg and tinker fed kill after kill in such a manner that, if it wasn't on purpose, I can only attribute to being absolutely horrible at this game.

  7. #127
    Banned
    Join Date
    Nov 2012
    Posts
    531
    match ID: 247166654

    Description: Every game that i play is HIGHLY favourable in either one or the other side. This is Solo queue and for the Nth the MM is match making 1250 wins player with absolute noobs ( relatively speaking )
    Bloodseeker - 871 wins - 0-10
    Leshrac - 391 wins - 1-10
    Alchemist - 734 wins - 1-8
    Invoker - 811 wins - 1-6

    I dont find dota2 enjoyable anymore. The Match Making algorithm is clearly inadequate and the developers are CLEARLY incapable of producing a working one.
    I haven't played a balanced game in over 3 weeks and i play a lot.

    Please hire an expert ouside your company to help you create an adequate MM algorithm. You guys should be ashamed of your work regarding the MM.

    ABSOLUTELY EVERY game that i play is worthy of being reported in that thread. It's not like "sometimes it works , someties it doesnt" - it's 100% inadequate.

    If I quit it will be, cuz the MM is shit .. seriously .. the game is good, but the MM is absolute shit. You have bigger player base than ( for example HoN ) and still somehow the MM is producing infinitely worse matches.

  8. #128
    Basic Member
    Join Date
    Jul 2013
    Posts
    1
    Quote Originally Posted by NB. View Post
    Ok im the guy who has 2000+ wins here. My question is: If you guys improved matchmaking, fixed the bug that occured in the last couple weeks, did you anyhow restore, roll back the MMR for affected users? Im not complaining about the loses that I have right now, in fact, i do win aquite a bit beside losing with this new MMR or w/e. But my problem is that the game has never ended at a close fashion like what it used to be when i played about a month ago, winning or losing, games are just not enjoyable for someone who seek high competition like me. Back then most of the people i get matched with are people who played as much games as I do OR pros players or stacked party of decent players. Now im getting newer players in my game where their decision making made no sense, not even counting some are lack of basic knowledge of the game.

    Also, how EXACTLY did you improve search time? At what cost that make the match making now search faster than before? Was that a simple algorithm improvement or you had to sacrifice skill variance in order to achieve such?

    Back around a year ago, Gabe did have a talk at a university where he briefly touch the match making issue in DotA2 and how Valve was not satisfied with what it was(I think he used the word 'loudsy implementation'?). Since then, hardly anything changed, is the a set plan to improve/replace the current system and if there is, whats the ETA?

    Edit: including my dotabuff profile http://dotabuff.com/players/72468268
    I played a game with you yesterday (this one: http://dotabuff.com/matches/246564422). I was on your team. You said exactly ZERO words the entire game. The first half of the game, you seemed fine and played alright. The second half of the game... You sat bottom lane, in the trees around the secret shop, and came out when the creep wave pushed far enough towards the shop for you to safely last hit (a few creeps every couple minutes). If you want to win games, STAY WITH YOUR TEAM. We were playing 4v5 (4v6?) for basically the entire second half of the game, and didn't get a single word out of you when we complained. If you don't want to play with a team, play a single player game or something like TF2 where your team isn't relying on you so heavily. You completely threw away a win, and made us play for 65 minutes and lose in what should have been a 40 minute win. You are the problem, not other people.

  9. #129
    Basic Member
    Join Date
    Dec 2011
    Posts
    20
    A possible solution:

    Make a Team Elo visible to the members. Much like wow arenas.

  10. #130
    Basic Member
    Join Date
    Dec 2011
    Posts
    11,187
    Quote Originally Posted by eldorquO View Post
    I like this system, because a good player should be able to do well with most heroes compared to a bad player, and a system like this would prevent issues where a new player only really knows how to use 5 heroes, but picks them every game and does well with them, causing his overall rank to get too high (so that when he picks a hero he's not intimately familiar with, he is awful and gets flamed).
    this system sounds nice, but it seems to be based on the assumption that every player tries to establish the highest rank possible with his skillset.
    what about abusers who only play one hero (let's assume they multiaccount for every hero) even though they are exceptionally skilled. wouldnt those players be paired with "newbies" and be able to stomp them again and again?

    this doesnt necessarily have to be intentional abuse. i know a fair share of players who have a very narrow skillset. a friend of mine only plays furion and clinkz and is able to compete in front page games with those heroes. true, he is alot more average with other heroes *cough support *cough*, but where would he be matched on the chinese server in mid tier?
    Make sure to read the Forum Rules as well as the stickied Threads of the Forum Section you are posting in.

    Contributions i'd like to highlight:
    My Suggestion: Coaching System
    My Sticky: Intended Changes List
    My Challenge: Completely Fixed Hero Challenge: Skywrath Mage

Posting Permissions

  • You may not post new threads
  • You may not post replies
  • You may not post attachments
  • You may not edit your posts
  •