Player Scouting — Utilizing Analytics and the Adaptive Unconscious

Renato Perdigão (@shakarez) & (@ClutchGaming)

When scouting players, how does a team assess and determine their strengths and weaknesses? How much data is used and are there relevant stats that help with scouting?

“CG has already built out one of the largest scouting systems of any NA LCS team, and is using findings from Challenger and pro games around the world to develop an unprecedented competency in analytics, giving them an upper hand in building a competitive squad from Day 1.” — lolesports

While CG is delving into the use of analytics like its NBA counterpart, the team has brought on a staff of scouts and analysts to put in the work to develop a keen eye for spotting potential in professional League of Legends players.

I was brought on back in June to help grow the scouting process and prepare Clutch Gaming for free agency should the Rockets get into the NA LCS. We recognized immediately that the time window to go after key talent would be small and that we would want to get the legwork completed before the league even started. We also brought on Ryan Friedman (@TheAzotal) to help the scouting staff and its efforts in building the a successful new NALCS team.

First things first. The organization realized early on that coming in as a new team meant understanding that there’s a lot of groundwork to do to in order to catch up with established organizations (endemic teams). All endemic teams already had a roster, staff, and infrastructure. They had the optionality to solidify their team through minor roster and/or staff changes.

On the other hand, non-endemic organizations would have to start from scratch. This inherent disadvantage would manifest itself on many different fronts (from game knowledge to player expertise to even coaching/analyst staffing).

So when faced with an inherent disadvantage of being a non-endemic organization, uncertainty in whether or not we’d be a NA LCS partner, and only a few months until we had to move quickly, what should we do?

The solution was to take every action one step at a time and start scouting players.

One of the biggest questions people have asked is how did we scout players? It’s simple really. We watched a lot of vods. We talked at length to determine certain metrics that we defined to focus on for each role (our original version had every player in each role evaluated on the same metrics, but we quickly threw that iteration away. Different roles have different value functions and thus different metrics. How to define what metrics we evaluate our players on is a debate that has raged internally for a long time and one that’s still being had today) , made sure we stated our biases up front, and began the task of ranking players as accurately as we possibly could.

When looking at support players we examined how they would use lane priority for example

Is this a perfect solution? Absolutely not. The whole process of scouting a single player was extremely hard. There’s no way for us to access player comms or a way to only focus on a position when watching a competitive match (as there are no POV vods or replays). It didn’t matter whether it was the Challenger Series, the LCS or an International Tournament, we only had the content we had.

However, when faced with adversity and a steep hill to climb to reach the levels of the top teams in League of Legends, we had to use whatever we had and focus on that to make the best assessments we could.

Where do the Analytics come in?

Yuhan and our team have been working on a number of different tools and techniques to extract info from games to help better our understanding of League of Legends. Just like in our scouting, you have to start somewhere with whatever you have.

One of the biggest things that Clutch Gaming wants to fight against is this overconfidence in “known facts” without testing them. It’s rare that something is completely wrong all the time or completely right all the time.

We’re not that confident in anything other than this one truth: given enough time, energy, and resources, Clutch Gaming will figure something out. And with that, hopefully our resource allocation into analytics will net us a competitive edge in shaping our team as a force to be reckoned with.

We’re not there yet. But as we explore the space, we’ll share what we find with everyone (with the exception of the other 9 teams!)

Renato Perdigão (@shakarez) is a Scout for Clutch Gaming.