← Back to Home

About cEDH Analytics

Methodology, statistics, and technical details.

See data limitations for known caveats.

Elo details: cEDH Skill Rating methodology

Data model: Supabase schema + curated views

Data Inclusion

All findings are based on games played on the Topdeck platform.

  • Tournament and League data sourced from TopDeck.gg API
  • Only completed tournaments/leagues with published standings are included
  • Decklist data parsed and normalized for card frequency analysis
  • Partner commanders are tracked as a single combined commander identity
Primary Statistics

Win Rate

Win Rate = Wins / Total Games

The percentage of games won by a commander. In a 4-player pod, the expected (baseline) win rate is 25% - anything above this indicates above-average performance.

Example: A commander with 150 wins in 500 games has a 30% win rate (150/500 = 0.30), which is 5 percentage points above expected.

Conversion Rate (Top 16 / Top 10 / Top 4)

Conversion Rate = Top Bracket Finishes / Total Entries

The percentage of tournament entries that result in a top-bracket finish. Under 64 players, some events use a Top 10 cutoff, and for 34 players or fewer we only count Top 4 finishes.

Example: A commander with 25 top-bracket finishes from 100 entries has a 25% conversion rate.

Top Cut Conversion

Top Cut Conversion = Top Cut Finishes / Total Entries

The percentage of tournament entries that make the event's top cut bracket.

Example: A commander with 12 top cuts from 80 entries has a 15% top cut conversion rate.

Points per Game

Points per Game = (Wins * 5 + Draws) / Total Games

Weighted scoring that values wins at 5 points, draws at 1 point, and losses at 0 points.

Example: A commander with 10 wins, 5 draws, 25 losses across 40 games scores 1.375 points per game.

Resiliency

Resiliency = (Wins + Draws) / Total Games

The share of games that are not losses. Higher resiliency indicates a stronger ability to avoid losing.

Example: A commander with 20 wins and 10 draws across 50 games has 60% resiliency.

Inclusion Rate

Inclusion Rate = Decks with Card / Total Decks

For card analysis, this measures how often a card appears across all decklists for a given commander (or globally). Cards are tiered based on their inclusion rates.

Example: If Sol Ring appears in 95 of 100 decks, its inclusion rate is 95%.

Card Tiers

Core

80%+ inclusion

Essential

60-79%

Common

30-59%

Flex

10-29%

Spice

<10%

Statistical Significance

Not all observed differences are meaningful. Statistical significance helps us determine whether an observed effect (like a commander's win rate being above 25%) is likely real or just due to random chance.

Sample Size Requirements

We require minimum sample sizes before drawing conclusions. A commander with 5 entries and a 60% win rate is far less reliable than one with 500 entries and a 28% win rate.

Confidence Levels

★High100+ games - Strong statistical confidence
★Medium30-99 games - Moderate confidence, interpret with caution
★Low10-29 games - Low confidence, high variance expected

P-Value

p < 0.05 indicates statistical significance

The p-value represents the probability of observing results at least as extreme as the actual results, assuming the null hypothesis is true. In our context, if a commander's win rate appears higher than 25%, the p-value tells us how likely we'd see this by random chance.

Trap & Spice Analysis

Trap Score

Trap Score = Inclusion Rate × |Baseline WR - Card WR|

Identifies popular cards that underperform. Cards with high inclusion rates but below-baseline win rates are 'traps' - widely played despite hurting your chances. The trap score weights by inclusion rate so commonly-played underperformers rank higher.

Spice Identification

Spice = Low Inclusion Rate + High Win Rate Delta

Hidden gems are cards with <10% inclusion but significantly above-baseline win rates. These rarely-played cards may offer competitive advantages that the meta hasn't discovered yet.

Important Caveats

  • Correlation ≠ Causation:A card's correlation with win rate doesn't mean it causes wins
  • Confounding factors: Better players may play certain cards, skewing results
  • Meta context:A card's effectiveness depends on the current meta
  • Sample size: Low-inclusion cards have high variance in their statistics
Technology Stack

Frontend

  • Next.js 16 - React framework with App Router
  • TypeScript - Type-safe development
  • Tailwind CSS - Utility-first styling
  • Recharts - Data visualization
  • Radix UI - Accessible component primitives

Backend

  • Supabase - PostgreSQL database + API
  • Materialized Views - Pre-computed statistics
  • RPC Functions - Complex queries in PL/pgSQL
  • Edge Functions - Serverless compute

Data Pipeline

  • TopDeck.gg API - Tournament data source
  • Python ETL - Data extraction and loading
  • Scheduled Jobs - Regular data refreshes

Infrastructure

  • Vercel - Frontend hosting + CDN
  • Supabase Cloud - Managed PostgreSQL
  • GitHub Actions - CI/CD pipeline
Questions & Feedback

Have questions about the methodology or found an issue with the data? We welcome feedback and contributions.

View on GitHub