Overwatch 2 Dual-System ranking
OVERWATCH 2, SEASON 3
The competitive ranking system relies upon large datasets to attempt to fairly place players. It is a team centric game and as such, each randomly assembled team is it's own unique composition. Like a symphony played out in a single battle, the contribution of a single instrument is nearly impossible to separate from the whole effect of the piece of music.
Therefore when the difference between technical ability, game familiarity, strategic concepts and other factors vary wildly between players, the job becomes literally impossible without extremely large numbers of games played per player.
This problem of extreme variation does not confuse the ranking system in place at higher ranks of the game. Unfortunately, however, new player influx combined with "smurfing" can lead to constant unavoidable chaos at lower levels before the system has collected enough data. Finally, the behavior of "casual" players interacts undesirably with ranking methodology through low amounts of "per season" data upon which the system eventually must rely heavily to grant overall ranking stability.
The result is chaotic lower ranks with both low player ranking mobility and extremely chaotic "stomp-counterstomp" experience for those at the lower end which can reduce the fun of the overall experience. This identifies a need for a different approach to ranking methodology for lower rank players which is addressed below.
GENERAL UPDATES
Unlike higher ranks where the understanding of team interaction, timing, and coordination is known as the greatest factor, at lower levels of play team coordination cannot even be expected to occur with the slightest consistency. Therefore we recognize that the greatest contributing factor at low rank versus high rank flips in priority from individual skill to team coordination and tactics and can address the problem with a dual system.
At one end, individual performance statistics alone will be used for the lowest ranking divisions, including disregard of team win/loss, with Silver and Gold ranking system relying ever more heavily on team play and win/loss rates as the system returns completely to the normal method at Platinum and above.
At low rank, teamplay focused behavior is often "going along with" the wrong thing but the unfortunate "solo-play" style can completely destroy team coordination found at high levels and is a negative behavior that eliminates the net gains of self-sacrificing or supportive behavior good teams excel at. It must be eliminated to achieve the synergy of teamplay expected of high level competition which can only be captured via wins and losses, thus high performance ranking will remain unchanged.
- Individual statistics such as kill/death ratio, damage per second, healing per second, objective time per (specific) map, teammate saves, assists, etc are already tracked and adjusted by role. (though only slightly used) These stats will be compared to character averages at higher ranks for individual evaluation. There is already a "plays" evaluation system in place for the "play of the game" which will also be used for individual evaluation as well. Even teammate disconnect can still result in a positive adjustment to individual evaluation.
- For example: At Bronze rank and below, a single instance of "leavers, feeders, and griefers" which are far more common at this rank, can individually destroy a game for four other players. (especially with the introduction of single-tank gameplay) However, if a given player continues to do their job well, even a lost game could result in a higher ranking evaluation by the system.
- These changes are necessary, specifically, for data analysis techniques to allow greater mobility of ranking and consistency of matchmaking, via a smaller set of data from the more chaotic and inconsistent set of players.
BRONZE SPECIFIC UPDATES
Completely new players, veteran bronze players, "smurf accounts" or "rank boosting," and casual-but-skilled, represent a wild difference of skill not accountable in a single ranking. Discovering the differences via game performance data alone requires stronger focus on individual statistics over win/loss rates.
Therefore, the elemental ranks of Wood, Earth, and Water will be added to represent a greater spread of game familiarity. (Please don't call Water "potato" simply because it's below Earth)
Individual statistics can accidentally reward poor team play and has been an undesirable strategy for our prior single-system rank evaluation. Thus a double system that starts with individual performance statistics will be used solely for the three additional divisions but almost entirely for regular Bronze as well. However, team focussed play will still remain the priority evaluation metric of high level performance, thus each rank advancement will up to Platinum de-prioritize individual stats.
- Players previously evaluated as "Bronze" will first be placed in elemental ranks as a help system which will, at the end of every game, give a running list of per-game elemental rank as a game-by-game personal evaluation with constructive feedback on individual stats and tips such as practicing aim, survival, or other game basics so that players know what steps to take on their path to improve.
- Elemental players will be given positive reinforcement on individual skill improvement game-by-game as well as noting performances that indicate high levels of play in specific tasks like healing alone.
- Bronze and Silver ranks will be far more volatile in the speed of rank change to give rapid feedback as well as the feeling of mobility for long-term Bronze skill level players.
- Players of moderate or high individual skill and game familiarity will be far more rapidly sorted higher or lower to minimize game disruption from large technical and game-knowledge skill gaps that currently exist in various Silver and Bronze players.