To answer this question, let's use League Points as a measure of Victory/Defeat, with negative LP being a loss and positive LP being a win. We can then plot the Kill, Deaths and Assists against the points awarded.
The seven games I've tracked so far http://s33.postimg.org/qjuzk8czz/KDAvs_Leage_Points.png
Mysticman891 votesYour question seems ill formed, did you leave out a word perhaps?
Or were you attempting to see if your kda influenced the amount of LP you get? (If not, why use LP versus a binary win/loss?)
If your goal is to find out if LP is influenced by KDA, then the answer is no, it's not. Your LP gains/losses depend only on your mmr relative to the division you're in (and of course if you win/lose). If loss forgiven is on, then losses are reduced to 0, and gains are reduced by half, but those are the only influences, KDA and similar performance metrics don't matter.
BeeCuz1 votes{quoted} If your goal is to find out if LP is influenced by KDA, then the answer is no, it's not.
That is not the question I asked and addressed.
To restate: Does the count of your Kills, Deaths, and Assists within the game have a measurable relationship to game outcome?
Mysticman891 votesMy point was why not just use the game outcome directly which is binary, rather than use a potentially misleading indicator?
People might look at your graph and think there's some significance to the points location on the x axis currently, which for your question there isn't any beyond if it's >0 (i.e. a win) or <=0 (i.e. a loss).
BeeCuz1 votesMy point was why not just use the game outcome directly which is binary, rather than use a potentially misleading indicator? People might look at your graph and think there's some significance to the points location on the x axis currently, which for your question there isn't any beyond if it's >0 (i.e. a win) or <=0 (i.e. a loss).
1). You are welcome to generate such a binary graph/table if it seems it would convey information to the audience
2). The question of significance that comes from examining the LP awarded vs the KDA would be another study. Perhaps you'd like to demonstrate how to approach such a hypothesis?
Mysticman891 votesI'm perfectly fine with intuiting that better kda ratios will be correlated with win/loss, and I expect most would be satisfied by that as well. Anyone who plays the game knows that a good kda doesn't assure victory (and indeed theres plenty of cases where the entire team can do poorly, but a 'lucky' ace can be enough to let a team kill the nexus from behind), so it's known it's merely a correlation and not a causal relationship. As such, I don't see what there might be to gain from exploring how strongly correlated kda ratios are to game outcome, but you are of course free to approach it if it interests you.
That said, I would propose that measurements of Kills/deaths/assists individually as you're doing would be a better indicator of game length than game outcome.
We already known that LP awarded versus KDA is unrelated, since Riot already told us what it does depend on (i.e. the things I mentioned in my first post). If you or anyone found that there was a strong correlation, then that would just indicate that there was a confounding factor in the data (or simply there wasn't enough data).
Science/research/etc is cool and all, but you have to be careful if you want it to be meaningful. Confounding variables are in abundance, particularly with such a complex game as league, so you really need to be careful to control for things. On the plus side, there's lots of things we know about league (since riot told us), so that can make things a little easier than the messy complex world that is real life, but it's still somewhat more complicated than slapping some variables together on a graph and looking for lines.
BeeCuz1 votesI'm perfectly fine with intuiting that better kda ratios will be correlated with win/loss, and I expect most would be satisfied by that as well.
This really gets to the heart of the matter and why I'm tracking this data, because I suspect that intuition to be a false positive.
