Why Riot Could Easily Create A "Troll" Detection System
The following is a list of things (albeit not comprehensive) you would need to consider in order to decide whether or not someone was intentionally disrupting or creating a negative experience for their teammates.
Primarily, there are eleven topics:
- Known player history with champion (Win rate, KDA, etc).
- Known player history of enemy players (Win rate, KDA, etc).
- MMR / Tier (How successful do most players do at this MMR w/ selected champion?).
- Selected Role (Is the champion commonly used in that role?).
- Items (What are the most common items for this role?).
- KDA (Does the player have an excessive amount of deaths or low Kill Participation?).
- Source of death (Is the player dying because of a fed opponent?).
- Location of deaths (Is the player dying in unusual locations?).
- Weighted win rate of champion vs each enemy champion (Does this champion tend to struggle against a certain enemy or team composition?)
- Target and location of abilities (Is some champion using an ability in an abnormal spot? Such as: Anivia walling off the fountain).
- Number of casts of abilities (Is the player not casting their spells during team fights or at all?)
Riot already has all of this information. Literally all they would need to do is score these against the "average" of the entire player base and anytime there were multiple abnormal scores it would be clear that the player was either not trying or griefing in some manner. I know everyone here talks about how "difficult" it is, but it's really not as difficult as you believe. The very information Riot uses to balance their champions can easily be used in other applications, such as this above.
