Thread: 14 matches in a row against higher-ranked players | Forums

  1. #1

    14 matches in a row against higher-ranked players

    Seriously, guys, you have *GOT TO FIX* the matchup system. I am not kidding you, I was just matched against FOURTEEN players in a row with a higher ranking; the last two matches of the previous season, and the first *12* of this season.

    Me -- Villian -- Result
    3115 v 2554 - 0-0 D
    3116 v 3046 - 2-1 W
    New Season
    2941 v 2721 - 3-0 W
    2696 v 2534 - 0-1 L
    2907 v 2818 - 1-2 L
    3865 v 3155 - 3-0 W
    2906 v 2874 - 1-0 W
    2783 v 2644 - 1-1 D
    2593 v 1802 - 0-1 L
    2798 v 2555 - 3-1 W
    2541 v 2341 - 0-1 L
    2792 v 1914 - 3-0 W
    2501 v 2460 - 1-0 W
    2115 v 1983 - 0-1 L

    And we broke the streak with...

    2534 v 2549 - 2-1 W

    Wow. A whopping 15 ranks below me...but that's better than 300, 600, 870 ranks above me.


    Can you please take a look at this? I don't know what's wrong with your matchmaking system...but it's not working.


    And before you argue "this is an anomaly," I tracked the previous 367 matches...I was the underdog 192 times, and the favorite 176 times, all within a fairly narrow Shield Rank range of 1861-3900. [By the way, it is at the low point of 3865 and 3885 that I was matched against #163 and #224 in back-to-back matches.]

    The vast majority of these matches occured when I was ranked in the 2000s; 244 out of 381.

    During the previous season, whether measuring by Rank or by Skill Points, there were far far far far more people below me than there were above...and in a range of, say 5000 Skill Points, again, far more people below than above. And yet, the system found the higher-ranked players with beyond-statistical probability.


    I also don't want to hear "oh, you're outperforming expectations. You had a record of 6-5-1 against opponents where you should have been 0-12. Therefore, you're matched up against higher opponents. Our algorithm takes that into account."


    Just set a Skill Point range that players should play against, not a Shield Rank....then stick to that range. If you're at 12,000 ELO, you should have about an equal chance being matched against a 14,000 ELO player as against an 10,000 ELO player. That would be a fair system.
    Share this post

  2. #2
    Originally Posted by EscapeRoomDave Go to original post
    If you're at 12,000 ELO, you should have about an equal chance being matched against a 14,000 ELO player as against an 10,000 ELO player. That would be a fair system.
    With 50/50% chance you can easily get matched against higher ELO player endless amount of times in a row (don't take it serious I got what you meant).

    Current system is just fine. You are talking about such small percent difference that I cant understand why you even complaining. It makes zero sense to me.

    Im always happy to meet a bit higher rank players than me, because: a) I get more points if I win / lose less if I was beaten (derankers is unpredictible); b) it let me understand if i'm belong to higher rank;
    Share this post

  3. #3
    I tracked the previous 367 matches...I was the underdog 192 times, and the favorite 176 times
    Sounds like things are working as intended, you are seeing a good distribution of over/under matches and even the ones in your long list above show cases where you are right in the ballpark for a competitive match (and your win/loss results fit too!).

    I honestly don't know what you expect when you keep making posts on the issue when your data shows things are looking really good. I think your expectation of having matches in a perfect Gaussian distribution around your rank with a tight standard deviation is not exactly reasonable given the reliance on having the right player online at the right time.

    Actually... can you please post the full data series of these 367 matches? It would be really interesting to drop them into minitab and make a plot showing how good/bad the situation is. I think the results may be surprising to you when you look at the analysis.
    Share this post

  4. #4
    In Asia, the situation is worse.
    We have so called "川Go" time----We face only 川Go or around him.
    Even if I am rank 4000, I face No1-10 continuousely.
    In that case, App can't culcurate ELO, then App stop.
    I have to restart App and I need several minutes to go to next match.
    It is a kind of car accident, right ?
    Today, I lost and lost and lost because of this F*** matchmanking system, and my rank moved from 3000->7000.
    I realised rank sheld and ELO does not any meaning.
    Also so called "deranker" is created by this way!!!
    Share this post

  5. #5
    No, I will not go through the trouble of posting the information from 360+ matches, because you'll simply make the asinine statement that "it's working as intended."

    I only have ranking info, not ELO. And ELO is what should really count.

    Many many many times while ranked 2500+, I was matched against someone in the Top 200-800 players...these are players with 5,000-7,000 more ELO than me.

    Was I ever ranked against someone 5,000 ELO *less* than me? Never. Not once, because I'd have to be facing someone ranked #11,000 or lower.

    The system does *not* give anywhere near a 50/50 distribution using a range of ELO, nor a range of ranking. If it did, I'd be matched against #11,000 just as often as I'm matched against #600.

    I don't want to hear any of your idiotic doubletalk excuses for the system. I want a fair system, not someone defending a broken one.
    Share this post