Elemental Skirmish Queue Bot

A competitive queue + matchmaking bot used inside the Elemental Esports Discord to run fair, fast in-house Valorant games with account linking, party queues, base + seasonal MMR, and seasonal leaderboards.

Note: Valorant data is currently fetched via HenrikDev (unofficial) while the official Riot API application is under review.

What it does

Queues: join solo or as a party; region/server-aware; configurable queue types
Matchmaking: targets equal total MMR; caps team gap (≤ 350)
MMR model: Base MMR (from current/peak Valorant rank) + Seasonal MMR (earned)
Seasons & ranks: tiered seasonal ranks, reset controls, history, leaderboards
Admin ops: start/cancel matches, set results, migrate season, audit rating changes

Why it matters

Fair games: underdog/favorite balancing, upset bonuses, anti-farm dampening
Less admin overhead: automated queues, results, deltas, and persistence
Visible progress: seasonal ranks with meaningful climbs; history you can point to

Flow (end to end)

Link account → user verifies their Valorant profile (Discord ↔ Valorant)
Base MMR → conservative seed from current/peak rank (+ tiny RR)
Join queue → solo or party; region-compatibility & MMR window enforced
Matchmaking → two teams formed with similar (base + seasonal) totals
Play & report → admins set score / winner; draws & cancels supported
Apply deltas → adaptive K-factor updates seasonal MMR; history recorded
Leaderboards → live seasonal standings + per-queue ladders

Matchmaking & MMR (short version)

Total MMR = Base (1000–3000 cap at link) + Seasonal (unbounded)
Gap scaling: at 350 MMR diff → underdog +25%, favorite −35% gain/penalty
Upset bonus: underdog win → +15%
Premades: full 3-stack −5% dampen (coordination edge)
Spread dampening: big internal spread reduces win gains (up to −40%)
K-factor: newcomers 50; experienced 16–64 (adaptive by form/consistency)

Core snippet

function expected(rA: number, rB: number) {
  return 1 / (1 + Math.pow(10, -(rA - rB) / 400));
}

const MAX_DIFF = 350;

function gapScale(teamAvg: number, oppAvg: number) {
  const diff = Math.min(Math.abs(teamAvg - oppAvg), MAX_DIFF);
  const ratio = diff / MAX_DIFF;
  const underdog = teamAvg < oppAvg;
  const maxUnderdog = 0.25;
  const maxFavorite = 0.35;
  return underdog ? 1 + maxUnderdog * ratio : 1 - maxFavorite * ratio;
}

Efe