PvP Matchmaking & ELO System

## ROLE
You are a competitive game systems designer who has built matchmaking systems for ranked PvP games with millions of concurrent players. You understand Elo, Glicko-2, TrueSkill, and custom rating systems.

## OBJECTIVE
Design a matchmaking and ranking system for [GAME TITLE]'s [GAME MODE: 1v1/Team/Battle Royale] competitive mode.

## TASK

### Rating System Selection
- Elo system: simple, transparent, best for 1v1
- Glicko-2: adds rating deviation and volatility for more accurate matching
- TrueSkill: Microsoft's Bayesian system, excellent for team games
- Custom hybrid: combine approaches based on game-specific needs
- Placement matches: initial calibration period (8-10 matches)
- Rating decay: inactive players gradually lose rating certainty

### Matchmaking Algorithm
- Search radius: start narrow (±50 MMR), expand over time
- Maximum wait time: balance match quality vs queue time
- Party size handling: premade groups vs solo players
- Role queue: if applicable, match by role preferences
- Ping-based filtering: regional server selection for fair latency
- Backfill logic: handling disconnects and leavers mid-match

### Rank Tiers & Progression
- Tier structure: Bronze through Grandmaster (6-8 tiers with subdivisions)
- Promotion/demotion: match series at tier boundaries
- Loss protection: safety net at new rank for reduced frustration
- Visible rank vs hidden MMR: when and how they diverge
- Season length: optimal reset cadence (2-3 months)
- Season rewards: exclusive cosmetics and titles per rank

### Anti-Abuse Systems
- Smurf detection: rapid win streaks trigger accelerated MMR gains
- Win trading detection: suspicious patterns between accounts
- Boosting prevention: party MMR restrictions and performance analysis
- Dodge penalties: escalating punishments for queue dodging
- AFK/griefing detection: behavioral scoring affects matchmaking
- Hardware/IP bans: for repeat ban evaders

### Player Experience
- Match quality score: post-game assessment of balance quality
- Comeback mechanics: close matches are more engaging than stomps
- Streak bonuses/protection: reward hot streaks, cushion cold streaks
- Performance-based adjustments: individual performance modifies MMR gains
- Transparency: show players why they gained/lost specific rating amounts
- End-of-season analytics: personal performance reports

## OUTPUT FORMAT
Complete matchmaking design document with algorithm pseudocode, tier structure, anti-abuse systems, and player-facing UX specifications.

## CONSTRAINTS
- System must handle 100K+ concurrent players in queue
- Average queue time target: under 3 minutes for most ranks
- Match quality target: less than 5% of matches should be severe mismatches
- Support cross-platform play with input-based filtering
- Plan for low-population edge cases (very high/low ranks, off-peak hours)