About NCG Squad Builder

What Samsta AI^™ does, where its signal comes from, and how it decides who plays with who.

◆ 01

The Pitch

A private, signal-driven team builder for the NCG friend group of 15 League of Legends players, powered by Samsta AI^™. No captain's-pick drama, no ego drafting, no "well I feel like my team's better". Two clicks produce two teams the engine insists are as fair as possible.

Pick 10 of 15

Toggle the summoners in tonight's lobby. Everyone not playing sits out — no pressure, no last-minute shuffling.

Hit SUMMON THE RIFT

Samsta AI^™ evaluates every valid way to split the 10 into two role-complete teams and picks the fairest configuration. Deterministic, reproducible — same ten in, same two teams out.

Get a Deep Read

The Hextech Oracle, informed by Samsta AI^™'s analysis, writes up the matchup: who's paired with who and why, key head-to-heads, duo chemistry, x-factors, and a final verdict.

"This was built to solve the question of the ultimate 5 stack."

— Samsta, host

◆ 02

The Data

Every signal Samsta AI^™ uses comes from Riot's official match-v5 API — no scraping, no third parties. The cron pulls each player's latest matches, adds them to the archive, and the per-role profile updates itself.

Queues we count

Only serious Summoner's Rift modes. ARAM, URF, One-for-All and custom games are filtered out because they don't reflect real role skill.

Normal Draft Ranked Solo/Duo Ranked Flex Quickplay

What Samsta AI^™ looks at, per match, per role

Games played

Sample size per role — underpins how much confidence the engine gives any single signal.

Win rate

How often you win when you play this role. The cleanest outcome signal there is.

Combat contribution

KDA, kill participation, and damage output blended into a single measure of how much you affect fights.

CS per minute

Farm efficiency. Weighted more heavily in lane roles than in jungle or support.

Vision score

Ward placements + denials. The quiet stat that separates good supports from great ones.

Champion pool depth

How many distinct champs you've got meaningful games on in this role. Flexibility matters.

Rank signal

Current tier, division, and climb momentum across Flex and Solo/Duo.

Recent form

How your last run of games compares to your historical baseline. Catches hot streaks and slumps.

Duo chemistry

Your win rate as teammates with each other player in the roster — the pair signal the engine uses to prefer real-tested combos.

Head-to-head history

Your record against each other player when you've ended up on opposite teams. Flags matchups with a track record.

◆ 03

The Algorithm

Two layers: a Samsta AI^™ per-role skill model that rates every player in every role they credibly play, and a partition search that finds the fairest way to split 10 players into two teams of 5.

The per-role skill model

For each player in each role they've got credible reps in, Samsta AI^™ produces a single proprietary rating. The exact blend is the engine's secret sauce — what matters is what it considers:

Rank signal (Flex + Solo/Duo)

Role-specific win rate

Combat contribution

Champion pool depth

Recent form

Duo chemistry

Head-to-head history

Every signal is blended Samsta AI^™-side using a proprietary model that's been tuned against the group's actual match history. The output is a single rating per player per role that the partition search can compare apples-to-apples.

The partition search

With 10 players and 5 role slots per team, the number of valid role-locked partitions is manageable. Samsta AI^™ evaluates every one of them against a multi-factor fairness objective — team totals as close as possible, internal spread on each side kept tight, minimal forcing of anyone off their main role, and pair chemistry between teammates factored in. The best split wins.

Why this matters

It's balanced two ways at once

Team totals are close and each team's internal spread is tight. No team gets four studs and a throwaway — that's a stacked team, not a balanced one.

It respects who plays what

The engine discourages putting a support main in top lane. We could force it, but your friend will rage-quit.

It knows your duos

Players with a real history of winning together carry a chemistry bonus — Samsta AI^™ would rather put a proven duo on the same side than break them up for cosmetic balance.

It's exhaustive, not heuristic

The search actually looks at every valid partition — no "good enough" approximations. If a better split exists, Samsta AI^™ finds it.

It's deterministic

Same 10 players in, same teams out. Every time. No random seeds, no "re-roll for better dice." If you don't like the split, the answer is to swap a player.