Results 1 to 9 of 9

Thread: [Suggestion] A Performance Rating system that rates players for each skill category

  1. #1
    Basic Member Bleako's Avatar
    Join Date
    Jan 2012
    Location
    Legendary skill bracket
    Posts
    2,280

    [Suggestion] A Performance Rating system that rates players for each skill category

    I know a lot of players have been asking Valve to display their MMR ranking in the past. There are multiple valid reasons in favor of and against displaying this statistic, but one thing's for sure: players interested in competitive games need to have some sort of stats system to not only give them a sense of progression but also to let them know where they stand within the community and what skill they should try to improve. This is why Dotabuff.com's short-lived rating system was praised by a portion of the community but ultimately would've proven toxic if it had stayed.

    Basically, we need stats but we also need to reduce the amount of abuse related to sharing that information with others (and perhaps even prevent it entirely).

    My suggestion: a Performance Ratings system with various categories of "skills", each rated with simple non-numerical one-word ranks.

    Because DotA 2 is a complex game with many different category of skills and many different roles, it's very hard to properly rank players with only one single numerical rating. Two players with the exact same Elo rating could have widely different areas of expertise - one player can be a phenomenal support player while the other player can be an all-around, jack-of-all-trades kind of guy. This is why we need multiple rating categories that encompass the wide variety of skills. But we also have to account for all the different roles in the game (which each require a different set of skills) meaning that we need certain roles to have more or less impact on each rating (for instance, games played as a Support should not affect your Last Hitting rating as much as if you were to play a Carry)

    Without further ado, here is my suggested ratings system. There are 11 different rating types, 3 levels of impact for 4 hero types and there are 6 expertise ranks (Average > Good > Great > Excellent > Expert > Master). Each level of rank would simply be determined by a hidden numbered rating (similar to the Elo rating, only specific to each category). These stats could also be displayed with a small graph next to them showing the player's curve of improvement over time. Additionally, small rewards could be granted when reaching one of the 3-4 highest ranks.

    Performance Rating
    Determined by averaging all other ratings

    K/D/R Rating
    Determined by observing:
    • Number of kills+assists compared to number of deaths

    Impact on rating per role type:
    • Carry: high
    • Roamer: high
    • Pusher: medium
    • Support: low

    Last Hitting Rating
    Determined by observing:
    • Number of last hits compared to total number of enemy creeps within range of the player (in the first 10 minutes of a match)
    • Total last hits per match

    Impact on rating per role type:
    • Carry: high
    • Roamer: medium
    • Pusher: high
    • Support: low

    Time Efficiency Rating
    Determined by observing:
    • Time spent farming (gold, xp), attacking towers, being in a fight, warding, doing rune control compared to total game time

    Impact on rating per role type:
    • Carry: high
    • Roamer: medium
    • Pusher: high
    • Support: low

    Map Control Rating
    Determined by observing:
    • Frequency of ward purchase and placement
    • Number of times player or ward granted vision of a rune spot (regardless of the rune's position)
    • Willingness to purchase invisibility detection for wards
    • Willingness to purchase invisibility detection for invisible heroes/Shadow Amulet/Shadow Blade holders (if applicable)

    Impact on rating per role type:
    • Carry: low
    • Roamer: medium
    • Pusher: medium
    • Support: high

    Team Fight Rating
    Determined by observing:
    • Number of times involved in a team fight compared to total number of 2-to-5 player fights
    • Average outcome of fights player was within a certain range of

    Impact on rating per role type:
    • Carry: medium
    • Roamer: high
    • Pusher: low
    • Support: high

    Team Support Rating
    Determined by observing:
    • Willingness to purchase wards
    • Frequency of TP reaction to enemy ganks in other lanes
    • Number of last hits compared to number of attempted last hits by your lane's carry (in the first 10 minutes of a match)
    • Willingness to stack jungle creeps or pull creeps to the lane (in the first 10 minutes of a match)
    • Willingness to share consumables with other players
    • Willingness to purchase or upgrade couriers

    Impact on rating per role type:
    • Carry: low
    • Roamer: medium
    • Pusher: low
    • Support: high

    Tower Pushing Rating
    Determined by observing:
    • Number of times spent attacking towers
    • Number of times wiping a wave of enemy lane creeps close to an enemy tower
    • Number of buildings destroyed
    • Frequency of split-pushing (number of time spent pushing/attacking an enemy tower while the rest of the team is attacking/defending a different tower)

    Impact on rating per role type:
    • Carry: medium
    • Roamer: medium
    • Pusher: high
    • Support: low

    APM Rating
    Determined by observing:
    • Number of commands given per minute (not calculated when not within a certain range of an enemy unit)
    • Control efficiency of secondary units (if applicable)

    Reflex Rating
    Determined by observing:
    • Number of enemy skill shots dodged compared to enemy skill shots taken.
    • Speed at which abilities are cast in succession in team fights
    • Frequency of animation canceling
    • Speed at which certain defensive spells are cast before damage is taken (Disruption, Shallow Grave, etc)
    • Speed at which the player reacts to an enemy popping out of fog of war/invisibility near them.

    Community Fame Rating
    Determined by observing:
    • Frequency of manual mute/player report
    • Number of commends given by other players
    • Number of commends given to others players
    • Average number of matches spectated by more than 60 people
    • Number of self-made cosmetics available in the store
    • Total number of matches played.
    • Whether or not the player has participated in a professional tournament

    Additional details:
    - "Willingness to" refers to the player doing a specific action at least once in a match
    - Heroes with multiple well-defined roles (for instance, Nature's Prophet and Abaddon) have an global impact rating of "medium".
    - For the sake of this rating system, "carry"-type heroes are any heroes who primarily deal autoattack damage, "roamer"-type heroes include junglers and gankers, "pusher"-type heroes refer to a specific subset of heroes who focus on destroying towers and wiping creep waves, "support"-type heroes is self-explanatory.
    - None of the above ratings should decay over time.

    Please feel free to suggest additional conditions for each rating category that I might have missed!
    Last edited by Bleako; 09-24-2013 at 11:10 AM.

  2. #2
    Basic Member DiGodKolya's Avatar
    Join Date
    Oct 2011
    Posts
    133
    I wouldn't feel satisfied knowing that alone.
    However, these stats should all be visible to you anyway, just make them private.
    I don't understand what's the huge deal with people not wanting to see whether or not they improved when it's PRIVATE.

    Have bracket systems added to the game already.

    more ontopic:
    Your stats seem good, but i dont want them to be the only stats i know. I want brackets and thus this idea isn't good enough for me.
    supporting it as add-on stats, but not as a stand alone stastics.

  3. #3
    Basic Member
    Join Date
    Jul 2012
    Posts
    37
    I could argue alot about the different catagories and how it should be done, but basicly I agree that we really need some sort of statistic and preformance rating.

    I would still like to add that assist on kills should be high for support (and thereby seperated by kills) and ofc denys on creep and towers should be there also (important for all tbh) also harass..how much dmg done to enemy players doing laning.

  4. #4
    Basic Member soprof's Avatar
    Join Date
    Dec 2011
    Location
    SF Bay Area
    Posts
    1,231
    Many of these categories are actually the "playfields", or PvP planes inside the game.

    You've have a very bad K/D/R versus v. good players in certain conditions; (1.)
    And you'll get no lasthits as well, (2.)
    Thus you'll be efficient in the category (3.)

    Other categories can be easily boosted:
    Mass obs+sentry in the end of the game (4.)
    Apm (8)
    Reflex (9)
    I am former gm sc2 player, and I know how people love to boost stats like this if someone gives any value to them =D

    And some of them will actually provoke players to play "the standart way"(like 5, 6 and 7), which means games will be played in a certain template.

    I will not comment your last category, maybe it is a good idea, but it is not related to the game itself.

    Sorry, but I generally strongly disagree with this being implemented. This clearly looks like HoN and will create looks of Statsmen.
    You're only as strong as your weakest link.
    -Thomas Reid


  5. #5
    Basic Member
    Join Date
    Jul 2012
    Posts
    37
    well mr soprof..what we miss is could also be an official ladder...so people can see their skill. Your are a former gm sc2 player...if Blizz dident have their ranking system..you would just be a former good sc2 player. Knowing that you were actually one of the best on the world (which gm rank is) means alot. And all those bronze league players they still keep playing sc2 even when they know they are the worst of the worst.

  6. #6
    Basic Member bob's Avatar
    Join Date
    Sep 2011
    Posts
    172
    bump

  7. #7
    Basic Member Bleako's Avatar
    Join Date
    Jan 2012
    Location
    Legendary skill bracket
    Posts
    2,280
    Bumpin' ain't workin'

  8. #8
    Basic Member vladn's Avatar
    Join Date
    Jun 2013
    Posts
    2,838
    Very nice... but VERY hard to implement.
    This is how Dota2 games should look like! Game 1_Game 2_Game 3_Game 4_Game 5_Game 6

  9. #9
    Basic Member
    Join Date
    Aug 2013
    Posts
    6
    Quote Originally Posted by Bleako View Post
    Team Support Rating
    Determined by observing:
    • Willingness to purchase wards
    • Frequency of TP reaction to enemy ganks in other lanes
    • Number of last hits compared to number of attempted last hits by your lane's carry (in the first 10 minutes of a match)
    • Willingness to stack jungle creeps or pull creeps to the lane (in the first 10 minutes of a match)
    • Willingness to share consumables with other players
    • Willingness to purchase or upgrade couriers

    I don' think it should be restricted to the first 10 minutes of the match.


    As far as giving adequate rating rewards based on what role you play, I think that should be defined by the player by where he goes the first 10 mins, what he uses money on, how many last hits are attempted etc. The reason being is that a hero can be played in many different ways. One example would be that of naga.

    The real conondrum is how these roles would be detected by the rating system/bots/whatever. The so-called 1-through-5 positions (which actually also are subject to change, such as warlock mid that gets aga refresher, but then proceeds to let someone else on the team that scales better, farm).

    If this system, or one that is similar should get implemented, I believe it should be done so only in the top-of-the-top brackets, aka high DBR, aka game 1 page 1.

    Possible scenario:

    Game starts, creepwave can't spawn before people have decided lanes and heroes, and chosen what position the will play, prompted with 1/2/3/4/5. If nobody objects to one players choice within 15 seconds, the choice is "locked". Players should and would be repeatedly informed how the rating rewards are calculated.

    As far as rating supports goes, what has to be taken into account when the end result is revealed, is that of the differential in experience points between the rest of the heroes on the map (opposing support, carry, etc), the efficiency in stacking neutrals (missed or stacked correctly), damage done to towers and amount of zoning during timing pushes (3:30, 7:00, aka catapult pushes). Bonus points when TPing to countergank a global ganker such as wisp or furion.

Posting Permissions

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