Oracle’s Elixir LoL Statistics v Analytics | Tim Sevenhuysen Mag1c

Tim Sevenhuysen “Mag1c” of talks with Mark Register the difference between Statistics, Analytics, and what he does here

Mark Register – Can you explain the difference between statistics and analytics using the work that you do as examples?

Tim Sevenhuysen – There a probably from a purely linguistic point of view there are people who would define each word in different ways.

I could throw up a Wikipedia definition possibly but in my head, which is hopefully more interesting…to me when I think of statistics I think of just here are some numbers, we’ve recorded these numbers, here’s what they are.

Analytics is taking those numbers and using it to learn something.

So we’re not just learning the most basic statistics, this game won ten games and lost eight games.

That’s a statistic, you have a count of wins, you have a count of losses.

The analytics are to me when I use that word…we’re using that to come up with some kind of conclusion or some kind of observation that says this team is mediocre or above average but not by much.

You could look at jungle control or something like this and you could say the statistic is they take 53% of all the jungle farm in the game on average.

But the analytics might go beyond that and start looking at ok, well which parts of the jungle farm do they take at what times, and you’re starting to tell a story about what this means.

Is it good is it bad?

What does that tell you about the team and how they actually play?

That’s just in my head how I kind of use the two terms.

I think in a lot of different disciplines especially more established statistical disciplines in business, medicine, or a lot of these places they probably have very different pictures of what those words actually mean.

Since League of Legends stats are still a relatively young discipline I don’t know necessarily that it will ever evolve past being a relatively simple discipline.

I tend to use analytics when I’m thinking of interpreting or when I’m thinking of actually trying to maybe test a theory or a hypothesis and come to something that is more than just counting and observing.

Mark Register – I mean it sounds like you’re having to…as you said, you’re creating that foundation so you’re getting, you’re building the pool of information in which you can then draw from and then make more predictions just like your 15 minute early game indicator which 60% of the time it works every time.

Tim Sevenhuysen – Slightly better than that actually 🙂

Mark Register – Do you think that is that the future of where you want to go which is that once you kind of get that foundation and large enough pool of…steady stream of information, do you think analytics and more prediction is something you want to get into?

Tim Sevenhuysen – Possibly.

I think it depends on the value that different aspects of the work adds to different people.

I’ve got a lot of different communities that use my work in different ways.

There’s the fan community that just wants to see the numbers and see oh here’s my favorite team, my favorite players, how well are they playing?

How do they stack up?

At the most basic level and I really hope nobody comes to my site just for this but how good is their KDA ratio compared to other players and if they’ve got the best KDA then they’re the best player YEAH!

But well, there’s a reason I’ve got 20 more stats lined up behind it.

So that’s the fan community.

Then there’s the writer or the content creator community they come to my site because they want to see…usually they have a deeper understanding of the game they want to see I’m writing an article about this team, player, or this group of players, how do they stack up against each other, where are they strong, where are they weak?

They take the numbers and they use them in their articles or videos.

90% of the content that you see out there that cites statistics whether it references my site or not…well maybe not 90% I’m being a little generous…people also use EsportsPedia a lot…but a lot of those numbers come from my site because it’s not available a lot of other places.

There’s also another site games of legends dot com that is quite good, has a lot of numbers, just a little differently set up than mine.

There’s that community and they’ll come to these sites and they tend to be a little bit more aware of the site and use the site for their work.

And then there’s the actual analyst community and by actual, by what I mean is the coaching community.

Team coaches, professional analysts and they come to my site because generally they look for these numbers more than anyone else.

Otherwise they’d be gathering these numbers by hand themselves potentially or maybe there’s some numbers here that they haven’t thought of and they want to understand when they’re scouting their opponents or when they’re scouting their own team to understand how are we playing well, how are we playing poorly, maybe they already have a feeling but they just want to see it in numbers so they can get a second opinion.

They come to my site for the really in depth numbers and that’s what I’m thinking about in terms of innovation project and maybe some predictive work or things like that.

Usually it’s more oriented towards the analyst community but the kind of journalist community, the content creator community is more kind of the medium complexity and the relatively basic, the ones that people know the acronyms for already, the ones that are more straightforward for evaluating a player or team and how “good they are”…those are more for the fan community.

When I’m thinking about the different kinds of projects I can come up with, when I have time beyond writing articles, gathering stats, and publishing on my site I am thinking about which community I’m trying to serve at any given time.

When I came up with my win prediction model which I used to generate my early game rating and mid late rating numbers on my site, that’s probably the most complicated work I’ve done, the most complicated model I built but they way I presented the early game rating is actually a fan orientated number, mostly, and also somewhat to the medium complexity.

It’s a number that is built to be higher is better, lower is worse, it’s that simple.

It’s meant to be you can see it in a list and you know if it’s good or bad, it’s not all relative.

It is supposed to be one that is simple to interpret.

It’s also one that people aren’t…it takes a lot of time to adopt which involves a few people championing it out there, and I have seen it get a little more traction lately but it’s going to take time.

It just depends what community I’m trying to serve at any given time.

I do try and serve the analyst community with some of the work that I do and usually if I add a new stat to my tables it’s usually I’m adding more complicated ones rather than simpler ones to aim towards that community.

I don’t have anything really complicated in terms of exploration or new projects on the go right now just because I don’t really have time.

But it is something that I would love to do more of assuming I do get time to really get exploratory with it, it just depends on the time and the opportunity.

Leave a Reply

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