Announcement

Collapse
No announcement yet.

Discuss: WN8 and other Ratings and Rating Formulas

Collapse
X
  • Filter
  • Time
  • Show
Clear All
new posts

  • Discuss: WN8 and other Ratings and Rating Formulas

    So this is something I've argued ad nausaem with the public/random players in WoT. So far, none of believe me when I say that WN8 does not factor in cap points into its equation and have its own fallicies when it comes to "ranking" how good a player is. There are some human vs. machine logic inherent in this discussion as well, but most of the players just blow me off. It seems they take WN8 for granted as a "god rule" or the sorts. I find fault in that logic and assumption.

    There's also the issue I find that victory percentage is a heavy factor in the equation and sadly for me, I can't expect to win by carrying a losing team most of the time. Nor can I expect to have fun and win if most of these random tankers don't know how to play their tanks (especially heavies) and can't form a decent strategy to win (ie: not capping when they should and deciding to chase down a tank only to have all the chasers die to a single target, etc).

    Sadly, these discussion with randoms usually degenerates quickly into them name calling me and/or saying "you suck", "you're horrible" all according to WN8. Sadly, most of these randoms are about as good as I am according to WN8 or worst. Pisses me off to no end, but I only wished some of them were less ignorant.

    Some relevant articles/threads out there:

    The MYTH of WN8 (and Eff ratings in general): http://tazilon2003.wordpress.com/201...gs-in-general/

    * Very in-depth look at the equation/formula that makes up WN8 and goes into a nice discussion as to why efficient ratings don't encapsulate enough to be of any real meaning.

    Capping, tier, and WN8: http://forum.worldoftanks.com/index....-tier-and-wn8/

    * Just a reiteration (in one of the posts) that WN8 does not take into account cap points amongst other variables. I skimmed this thread, so there is maybe a decent discussion here.
    |TG-18th| Acreo Aeneas
    TG World of Tanks Clan Executive Officer
    Former 9th & 13th

    Pronounciation: Eh-Cree-Oh Ah-Nay-Ess
    Still can't say it? Call me Acorn then. -.-





    SSDs I Own: Kingston HyperX 3K (240 GB), Samsung 840 Pro (256 GB), Samsung 840 EVO (250 GB), Samsung 840 x 2 (120 GB), Plextor M5S (120 GB), OCZ Vertex (30 GB)

    TG Primer and Rules

  • #2
    Re: Discuss: WN8 and other Ratings and Rating Formulas

    Doesn't matter in any case. Win rate is luck. It's a team game, so if you get very bad teams for 90% of your games, you will lose no matter what.

    Also, if you do an average of 300 damage per battle, you're still an awesome asset to the team because you're doing the *right* 300 damage, whereas someone doing 1500 damage is just stat-whoring.

    Or so you'll hear people say, right after they announce that they're not playing to win, they're playing "to have fun." (and presumably, winning isn't fun for them)

    Let me ask you a different question: Suppose WN8 can be gamed. Who cares?

    Any metric, when applied across large enough a population, can be gamed to some degree. Either a few people will be able to produce wildly incorrect results, or many (or most) people will have stats that aren't precise reflections of their skill. So WN8 doesn't accurately reflect scouts' accomplishments. A metric on its own is just so much e-peen measuring. The correct question to be asking is, how are you attempting to apply what the metric tells you?



    My purpose is to be able to quickly evaluate the relative effectiveness of my and the enemy's team members. For this, neither WN8 nor the old efficiency metric are good enough. Both amalgamate performance across dozens of wildly different vehicles and play styles and pretend that you can draw some highly specific conclusions from this mess of generality.

    When I'm looking at my teammates in a random battle, I want something extremely specific: I want to know how good they are *in that particular tank*. For this purpose, I'm relying on a combination of win rate per tank, total battles played in the tank, and WN8. WN8 is probably second-lowest on the hierarchy of significance for me, followed only by overall win rate.

    If after 50+ battles someone has a 43% win rate, I don't much care what their overall WN8 is -- they aren't going to be able to pull their weight in this particular tank.

    Now, if you're trying to use WN8 in some other fashion, that's a different story.

    And capping? In 90% of all cases, capping with the express purpose of winning by capout is a terrible idea.




    Who needs a life when you can have a heavy bolter?
    --BlackMirror
    <23:03:38> "|TG|Smachin<BF Admin>" was kicked from the server by "|TG-70th| Zhohar" (UNDERAGE ban.)
    Anything over $600, and it would be pointless to try and reason with Grandma
    --Blackraven93

    Comment


    • #3
      Re: Discuss: WN8 and other Ratings and Rating Formulas

      I don't know everything about how WN8 is calculated, but it seems to be a fairly good rough-gauge of player effectiveness. I've taken a poke around WN8's methods and it really seems to be an excellent normalizing tool. Close to ideal, I'd even say, given what information it has access to.

      Personally, I still think that nothing is a better player skill gauge than win rate, evaluated on a per-tank basis. (granted, this is a large part of what WN8 does) Yes, you sometimes get put on teams that are too flaky to carry and yes, sometimes you get enemy teams too skilled to beat, but you get it the other way too. Statistics don't cheat. If you've got a sample size as large as 1000 battles or more, you will have been on the "better" team 50% of the time. Any variation from 49%, (accounting for draws) above or below IS directly indicative of how well YOU perform.
      Teamwork and Tactics are OP


      Strait /strāt/ (Noun) A narrow passage of water connecting two seas or two large areas of water: "the Northumberland Strait".

      Comment


      • #4
        Re: Discuss: WN8 and other Ratings and Rating Formulas

        I had a guy a few days ago tell me that I was under 1000 WN8 and therefore did not know how to play the game. He had a smidge over 1000. People really do take the numbers for more than what they're worth.

        I do agree with Cheb that WN8 is too generalized across the game. Stats per tank would be way more effective. I usually only look at their winrate in that tank, and then overall. It is obvious that most of the 1000+ players generally know what they're doing while watching them though.

        I like the number and pretty colors. Ever since I started tracking my numbers, I've been improving. That's really the main benefit for me. I saw how badly I sucked, and decided to do something about it. Now, according to my overall number, I suck slightly less.

        One thing I do find utterly ridiculous is when people use the numbers as an insult. I played a match in the past day or two where people were making fun of all of the "radishes" and "tomatoes" in the match. I find it sad when somebody puts that much into a few numbers. I guess if that's the only place they feel superior, so be it.

        Comment


        • #5
          Re: Discuss: WN8 and other Ratings and Rating Formulas

          Its ok Pirate. I had a guy the other day tell me that because I had less than 1k battles I should uninstall WoT.

          [unit][squadl][command2]

          KnyghtMare ~You could always tell the person holding the gun to your head you would like to play on a different server...

          Comment


          • #6
            Re: Discuss: WN8 and other Ratings and Rating Formulas

            I like Cheb's idea of a "how well are you doing in that tank" measurement of the player. I wish I could see more stats on how much damage they've done in their tanks, etc instead of just winrate in their tanks during a battle.
            |TG-18th| Acreo Aeneas
            TG World of Tanks Clan Executive Officer
            Former 9th & 13th

            Pronounciation: Eh-Cree-Oh Ah-Nay-Ess
            Still can't say it? Call me Acorn then. -.-





            SSDs I Own: Kingston HyperX 3K (240 GB), Samsung 840 Pro (256 GB), Samsung 840 EVO (250 GB), Samsung 840 x 2 (120 GB), Plextor M5S (120 GB), OCZ Vertex (30 GB)

            TG Primer and Rules

            Comment


            • #7
              Re: Discuss: WN8 and other Ratings and Rating Formulas

              Accuracy percent (for damaging shots only) is a more accurate indicator of player skill than win rate. Experienced players don't take shots they can't make while new or bad players will blast away at so much as a square millimeter of exposed hull. Sometimes those random shots get lucky but overall players with more skill will have a higher accuracy. The rub is that WoT tracks nonpenetrating/non critical impacts as hits, meaning someone who routinely blasts away on a KV1 glacis with a Pz1C 7.92 could have a 90% accuracy rating but a very, very low effectiveness rating.

              WN8 is extremely easy to pad if you have high level tanks since it places a heavy weight on damage inflicted. You can see at http://www.wnefficiency.net/wnexpected/ the various stats expected of tanks to achieve good rates. A good example is the T30 which has a 105, 120, and 155 cannon available. The 120 averages 400ish damage if you penetrate. The 155 averages 750 and will penetrate far more often. My personal WN8 jumped nearly 10% in 25 games after getting the 155, meaning that the same number of average hits with double damage so tremendously outweighed all previous data that it was able to drag more than 1k games upwards retroactively.

              On the other hand, WN7 was easy to cheat if you played a lot of mixed tiers (due to penalties) and picked a tank that helped you do that last 30 points of damage for the kill, which was weighted more heavily than damage. You were rewarded almost 1.5:1 for dealing 100 damage to a Maus for the kill than the other 1900 HP to drag it down.

              Comment


              • #8
                Re: Discuss: WN8 and other Ratings and Rating Formulas

                Sort of off topic, but I was looking at noobmeter and apparently yesterday I played 35 battles and had a WN8 of 4211. Whoa! Today was a totally different story. Couldn't get a decent game going to save my life. I seriously lost probably 90% of games today. They're all landslide loses too since 9.0.

                Comment


                • #9
                  Re: Discuss: WN8 and other Ratings and Rating Formulas

                  Originally posted by pirate golf View Post
                  Sort of off topic, but I was looking at noobmeter and apparently yesterday I played 35 battles and had a WN8 of 4211. Whoa! Today was a totally different story. Couldn't get a decent game going to save my life. I seriously lost probably 90% of games today. They're all landslide loses too since 9.0.
                  Well at least Bronco won't be saying that my climb was slow the last few months. XD
                  |TG-18th| Acreo Aeneas
                  TG World of Tanks Clan Executive Officer
                  Former 9th & 13th

                  Pronounciation: Eh-Cree-Oh Ah-Nay-Ess
                  Still can't say it? Call me Acorn then. -.-





                  SSDs I Own: Kingston HyperX 3K (240 GB), Samsung 840 Pro (256 GB), Samsung 840 EVO (250 GB), Samsung 840 x 2 (120 GB), Plextor M5S (120 GB), OCZ Vertex (30 GB)

                  TG Primer and Rules

                  Comment


                  • #10
                    Re: Discuss: WN8 and other Ratings and Rating Formulas

                    My 2 cents on the subject. I tried to improve my 60 day rating and then took a few days off. My ratings went down. I quit trying for improvement and just play for fun again. Well what are the ratings good for. They are what a lot of clans look for if you try to join.
                    The Old Guy
                    kin3
                    sigpic


                    Comment


                    • #11
                      Re: Discuss: WN8 and other Ratings and Rating Formulas

                      kin, I'm curious about this. What were you doing in order to improve your rating? And do you mean that your ratings dropped because you weren't playing as well after the break, or do you mean that taking a break itself caused them to drop?

                      I highly recommend WOT Statistics for those trying to monitor/improve their Win8. It lets you monitor your performance for each battle, as well as the performance for a given play session and changes over time.

                      But then, I'm a metrics nerd like you wouldn't believe.




                      Who needs a life when you can have a heavy bolter?
                      --BlackMirror
                      <23:03:38> "|TG|Smachin<BF Admin>" was kicked from the server by "|TG-70th| Zhohar" (UNDERAGE ban.)
                      Anything over $600, and it would be pointless to try and reason with Grandma
                      --Blackraven93

                      Comment

                      Connect

                      Collapse

                      TeamSpeak 3 Server

                      Collapse

                      Advertisement

                      Collapse

                      Twitter Feed

                      Collapse

                      Working...
                      X