K/D/A indexes and KDA ratio

When talking about how good or bad we did play in a certain game, we often refer to K/D/A indexes and KDA ratio as one way of judgement. Why are these indexes and what do they actually mean?

1. K/D/A indexes v.s. KDA ratio

First off, for clarity, we have four different indexes of discussion: K, D, A, and KDA. The KDA index or KDA ratio is calculated by:

KDA = (K + A)/D

These indexes convey different meanings which are not simply "the numbers of kills, deaths, assists, and the ratio". Especially when being combined into the index of KDA ratio, they really have insightful meanings, which are very important for us, since they may serve as one of the ways to judge how bad or good we are on playing League from different perspectives.

The first thing we can easily notice is that, if we make no death (mathematically, D = 0) this index of KDA ratio will be infinite. Which meanings does this case convey? Since we cannot eliminate the case, would the KDA ratio and all the indexes convey completely wrong senses?

2. The meanings: single points on a process

Basically for every team game, every goal or achievement must be the work of the whole team, not just of a certain single player. We have ten players for each game, five players each team. If one of our team mates gets a kill, he cannot get the kill alone: if he is at top, for example, he still needs other team mates to do their work at bot and mid and jungle, so that the opponents will be kept there and not go to top. A team cannot miss any player, and so any single kill taken must be considered to be the work of the whole team of five players for most exactly.

So, for the most exactly, one kill must be:

+) counted one kill for the player who did the killing blow

+) counted four assists for the four other team mates

This point, a problem appears: who is the player who did the killing blow? The player who made the last-hit, the player who dealt the most damage on the slain opponent, or even the player who dealt the most damage on all the opponents in total, etc? The other problem may be: why would it count an assist for those who had died before the kill and dealt no damage to the slain opponent?

One other way, not that much exactly in nature but more reasonably, is to consider those counts for just the players who did participate in the fights. In the meaning that, the players must deal damage within the fights, or at least help other players deal damage or protect them within the fights.

Obviously, each way of choosing the indexes has its own reasons and paradoxes. I can safely say that no way is perfect. But we have to choose one way nonetheless. For some reasons, Riot Games chose the second way, which in general is[1]:

The K index (kill) is counted for the only player who did the killing blow of last-hitting.

The A index (assist) is counted for the player(s) who did like any amount of damage on the slain opponent within a pretty short duration right before the skill.

The D index (death) is counted after a player gets slain, of course.

Note that, this choice of counting is of Riot Games' decision, it still has paradoxes[2] like I have just said. For example, the players who dealt the largest amount of damage on all the opponents may not be those who get the most kills, while the players who dealt the least damage on all the opponents may have highest KDA ratio like supporters for example.

Relatively, we will be considered to be a better player with more kills and assists and less deaths (greater K and A indexes with a smaller D index).

+) For just a single game, these three indexes of K and D and A may be enough to judge how good or bad a player did on the particular game. In this case of a single game, we may not need any additional index for more complicated.

+) The complexity is when we consider how good or bad we are on a set of a very large number of games, like hundreds or thousands of games with the same scales for the indexes of K and D and A. How can we say how good or bad a player is from like hundreds or thousands of games, with hundreds or thousands of K and D and A? One of the best ways, which Riot Games did choose, is to consider how effective the player do in one life; this is where the index of KDA ratio enters the League.

KDA ratio represents how effective a player plays one single life (in the meaning of before a single death). With one single life (or before each of his own death), how many times the player participates in taking down an opponent. On this perspective, the K and A indexes (kills and assists) should be treated as the same, since as said before: the work of taking down an opponent should be considered to be the achieve of all the players participating in the fight.

If your KDA ratio is 3:1, for example, that means for one single life (before a single death) you can make three times of participating in taking down an opponent.

Single points on a process: Basically, we can consider all the games we have played as a process of playing League, and each single game as a point on the process (which together consists of the process as a whole). When we say "how good or bad we are on playing League" (or some other games as well), we often make the reference to the process rather than a certain single match of the game.

The KDA ratio can be described as the characteristic of a process, while the K/D/A indexes should be described as the characteristics of a certain single match. These two kinds of indexes (KDA ratio and K/D/A indexes) are in fact very different in nature, due to the reason.

3. Some important conclusions

In case of a single game, the three indexes of K/D/A are enough to judge how good or bad we did play the game. To mention KDA ratio of a single game is not always a good idea, since this index may be very different for different games, and may get the infinite value in case a player makes no death (which is difficult and unnecessary to explain).

On the other hand, in case of a process (a set of hundreds or thousands of games), we must use KDA ratio to judge how good or bad we were in the process (at playing the set of games). To mention the separate indexes of K/D/A in this case is not a good idea, since it should makes nonsense to compare the sets of three numbers of hundreds or thousands of different players, as well as to talk about the meanings they convey in comparison.

Also for that reason, the number of deaths would never be zero for a set of hundreds or thousands of games. Or more strictly, we can say that the KDA ratio should only apply to the case of many games, many enough for the D index to get a non-zero value.

Different roles may have pretty different K and A indexes (the same for the D index). Support and assassin for example: support role would spend more time on warding and observing map, so may less participate in fighting relatively in comparison with AD assassin (whose main duty is to find a victim to take down).

Although players always favor few specific roles and always play very many games as their favorite roles, KDA ratio still serves as the relative indication of a player's skill level. This is especially true when it is used to compare those players favoring the same roles, and play the same numbers of games on their favorite roles.

Because the difficulty increases for higher Divisions and Tiers, KDA ratios of a player would be different and decrease for higher Divisions and Tiers. It would be hard to say which value of KDA ratio would be considered to be excellent, so the scale below is just for reference:

KDA = 3:1 is fair

KDA = 4:1 is good

KDA = 5:1 is excellent

Finally, due to the differences between the champions, the KDA ratios of a player would be different on different champions as well. A player may be excellent on playing Lux, meanwhile may be very bad on playing Jax, for example.

This post was completed; there would be just minor adjustments if any on this post

Last update: March 10, 2016

The Silencekeeper

References

[1] Note that, this discussion may have a gap in comparison with what Riot Games really do and did choose to do. Like they may choose whether or not an assist is counted for those players who didn't deal any damage on the slain opponent, but did help protect the one who made the killing blow. The nature of the issues doesn't change, nonetheless.
[2] Also note that, paradoxes are not always bad, and not always wrong