I Read a CBLOL Study on AI Roster Building — Here's Why Team Synergy Is Becoming Esports' Real Meta

#gaming #esports #ai #technology

Photo by Jade Chambers on Unsplash

There is a point where "sign the biggest name available" stops being strategy and starts being lazy management.

A recent CBLOL-focused study makes that point in a way esports teams can actually use. Instead of treating roster building like a star hunt, the researchers built a recommendation system around fit: role-specific performance, competitive network depth, and team synergy.

When they tested it against CBLOL 2024 Split 2, the predicted ranking reached a 0.89 Pearson correlation with the actual standings.

That is the kind of result that makes you look at roster construction very differently.

The old superteam idea is aging fast

Esports has spent years falling in love with the same fantasy:

Put enough elite players on one lineup and the wins will sort themselves out.

Sometimes that works for a while.
Most of the time, it ignores how games are actually won.

The article points to LOUD's run from mid-2022 to early 2024 as a reminder that cohesion can beat pure name value. A team that reads the map the same way, rotates with the same timing, and trusts the same fight windows is often stronger than a lineup built for announcement graphics.

What this model actually values

1. Role fit instead of generic stat worship

A support should not be judged like a mid laner.
A jungler should not be graded like an ADC.

The model gives different weights to different metrics depending on the role, which is exactly how real roster analysis should work.

2. Competitive network depth

The researchers used Leaguepedia data to map player history, match records, and competitive relationships.

That matters because performance is not only about isolated numbers. It is also about what kind of environments a player has already survived, adapted to, and learned from.

3. "Faker distance"

This is the part that makes the whole study memorable.

Borrowing from the logic of an Erdős Number, the model uses "Faker distance" to estimate how close a player is to the highest tiers of competitive League experience through the broader network.

It is a clever way to turn experience and pedigree into something more measurable than vibes.

Why this matters right now

Two outside signals make this feel bigger than one smart paper:

Signal	Why it matters
Grand View Research estimated the global esports market at $2.6B in 2025	Bad roster decisions now burn real money, not just fan goodwill.
Riot made Faker the first Hall of Legends inductee in 2024	The most iconic player in LoL esports is now also a useful anchor point for modeling experience.

There is a bigger lesson hiding in that table.

Esports management is maturing. Fast.

The orgs that keep optimizing only for highlight clips, name recognition, and short-term hype are going to lose ground to teams that treat roster building like systems design.