http://s32.postimg.org/gt4f54c01/Kill_Counter.png This gives a sense of where the player's performance fits in the set of all players who did that champion/lane combo.
More technically, its a display of the standard deviations from average, portioned into a five-bucket set and rounded to the nearest whole number for easy comprehension.
Mysticman892 votesGame length would be a confounding factor. Also "performance" covers a much much broader set of criteria than just kills, as in some cases (e.g. supports/splitpushers) players will be playing their role extremely well and end up with very few kills.
Incidentally, the already existing mastery grade already does account for both of those things, and compares you to all other performances in that champ/role combo. Getting an S+ grade for example indicates your performance was in the top 2% of all players in that role/champ combo, so the existing grade system already essentially works as you'd like.
It is not entirely clear to me that performances are necessarily normally distributed anyway. There could well be a bi-modal distribution for example on some champions where theres one peak around players who haven't necessarily mastered a particular combo, and another one further along for players who have figured out how to reliably do a champs combo.
BeeCuz1 votesThank you for contributing.
The purpose of this suggestion is to improve the player grasp and understanding of the Kill/Death/Assist scores and how their performance on those metrics compares to the community.
BeeCuz1 votesThe letter-grade system is a valuable feedback for the player -- I enjoy receiving such judgement on my performance -- but it is also opaque in terms of what the player did to achieve that score.
One of the consequences of opacity is a tendency to influence the score based on social politeness; not that it's a factor in this game, but it's so prevalent that lots of people take generic assessments with a grain of salt.
BeeCuz1 votesImplementation of Standard Deviation 1). The center green Normal kills bar would cover about 2/3 of the players performing that champion/lane, +/- 1 Sigma from Average
2). The upper and lower kill bars (blue and yellow) would each cover about 1/7 of players (13.6%) out to +/- 2 Sigma from average, accounting for about 95% of all players
3). The outer kill bars (purple and red) represent the performance of 1/40 of players each (2.5%) of +/- 3 Sigma from average, the final 5% of the player base
BeeCuz1 voteshttp://s32.postimg.org/u235jc65x/formula.png
The Sigma Algorithm 1). Start with a set of kills, for example {NA, Soraka, Support} 2). Calculate the total kills, totalKills = sum of all kills in the set 3). Calculate the average kills per game from the set of games played, averageKillsPerGame = totalKills / numberOfGamesInSet
From here we want some sense of how each game was different from that theoretical average, a deviation from average.
4). For each game in the set, calculate the distance between that game's kills and the set average gameDeviation = thisGameKills - averageKillsPerGame 5). Produce only positive distance numbers by squaring the result gameDeviationSquared = gameDeviation * gameDeviation 6). Store those gameDeviationSquared values and calculate the average of that new set of deviationsSquared averageDeviationSquared = totalDeviationsSquared / numberOfDeviationsSquared 7). Calculate Sigma, the standard deviation for the set: standardDeviation = squareRoot( averageDeviationSquared )
standardDeviation is a kind of average deviation away from the set average.
Maximum Morde1 votesWhat if my contribution was actually motivational speaking in team chat
