Esports Statistics Polarized by Context | Tim Sevenhuysen Mag1c

Tim Sevenhuysen “Mag1c” of talks with Mark Register about how context changes the interpretation of statistics

Mark Register – What measurements are straightforward indicators and which ones are polarized by context?

Tim Sevenhuysen – There’s more of the latter.

I kind of joke about KDA ratio sometimes and how people overuse it or misuse it or think it says more than it does.

It is relatively straightforward, the main thing you have to be careful about with that one is you’re not comparing different roles against one another.

A top laner’s KDA may tell a very different story than an AD carry’s KDA.

A certain number may be good for a top laner and bad for an AD carry.

That’s the kind of context you have there.

One of people’s favorite kind of groupings of numbers is damage output because it’s the most fun.

The most gratifying thing to do when your playing is to do lots of damage and carry the game.

It’s fun when you’re looking at your favorite players the more damage they do kind of better they are.

But there is a lot of context to those numbers that people aren’t necessarily aware of or that they don’t want to think about too much.

Things like game length, things like the champions that you’re playing, some champions just naturally have higher damage numbers than others because of their play patterns.

A champion that does a lot of long range poke damage that builds up over time but not every time they’re doing that damage, which they’re not threatening to kill someone it’s just they’re making them back off.

Those kinds of champions produce very high damage numbers but that doesn’t mean that they’re having more impact than a champion who does less frequent damage like an assassin but every time they do damage it’s very meaningful damage in terms of the outcome of the game.

So those kinds of interpretations, more people are picking up on them and becoming familiar with them and when you misuse damage stats and it comes out on Reddit or somewhere else, people are pretty quick to call you on it, which is nice to see as long as it’s done in a civil way.

That is a much more context heavy set of numbers than some people realize.

Gold is very similar, gold income levels, that one has a lot of complexity to it that people don’t necessarily realize.

One of the ones that comes up very frequently is the gold share statistic so how much of team’s overall gold you take in on average and it’s very common for people to say high gold share means that the team is intentionally funneling resources into that player.

Well you also get gold from kills so you may actually be getting a very small amount of the resources of the team relative to the average but if you’re still getting a lot of kills, then you’ll get a lot of gold, you’re going to get a high gold share but maybe you’re doing it with low resources which is actually even more impressive.

Or maybe you’re getting a ton of resources and you’re not getting kills so you’ve got a medium gold share when you really should have a much higher one if you were performing well.

So that’s actually not the best number to tell the story most people wanted to tell which is intentional resource distribution.

There are a lot of numbers I would say they all have a lot of context to them, that’s one of the big challenges with League of Legends stats.

I used to like warding numbers as a relatively straightforward one, they changed the whole warding system this year so there’s different types of wards and they last different lengths of time, those have more context to them then they used to which is unfortunate in some ways but the game takes precedence over the stats so I can’t complain too much about it.

In general the rule of thumb is any statistic that you’re going to use in League of Legends is going to have context to it and you should understand however much of the context you can before you start throwing the numbers around.

One thought on “Esports Statistics Polarized by Context | Tim Sevenhuysen Mag1c

  1. I totally agree with you. The context is often forgotten and it is hard to implement it in your analysis. How do you propose data-analysis can be improved? I think data-analysis can value from research conducted about the context in which stats are settled and the creation of different stats for different positions/ strategies…

Leave a Reply

Your email address will not be published. Required fields are marked *