Methodology

All tournament data is scraped from two publicly available sources:

• MTGGoldfish — tournament search results and individual event pages for Modern and Standard.
• MTGTop8 — tournament listings with a 2-month rolling window, including player counts and deck placements.

The scraper runs automatically once daily to pick up newly published events. Tournaments with at least 8 players are included. Events with fewer than 8 players are excluded as they provide insufficient data.

Meta percentages are not simple deck counts. We use a weighted system to give more influence to larger, more competitive events:

This means a deck placing in a 200-player Regional Championship carries twice the weight of a Challenge 32, and four times the weight of a small 8-player local event. Including smaller events provides more recent decklist data while keeping large competitive events dominant in the meta calculation. The default view uses 14 days of data, but you can adjust this from 7 to 60 days on the homepage.

Each archetype page displays a trend chart showing that deck’s meta share week-by-week over the past 8 weeks. The same weighted calculation applies — each week’s percentage is the archetype’s weighted deck count divided by the total weighted deck count for all archetypes in that week.

The chart gives a quick visual of whether a deck is rising, falling, or holding steady in the meta, which meta percentages alone don’t convey.

Win rates are sourced from mtgdecks.net, which tracks head-to-head match results from competitive tournaments. This data supplements our meta percentages by showing not just what’s popular, but what’s actually winning.

For each archetype, we display:

• Overall win rate — the archetype’s aggregate win percentage across all recorded matches.
• Head-to-head matchups — win rates against specific opposing archetypes, with confidence intervals indicating data reliability.
• Match counts — the sample size behind each win rate. Higher match counts mean more statistically meaningful results.

Win rates reflect tournament match results only — not ladder, league, or casual play data. Confidence intervals (shown as a range like 46%–53%) indicate statistical uncertainty; wider ranges mean fewer matches and less certainty.

Color coding: green (≥52%) indicates a favorable matchup, red (≤48%) indicates unfavorable, and neutral colors indicate an even matchup.

Archetype names come directly from how the source sites categorize each deck. Because different sites sometimes use different names for the same deck (e.g., “Boros Energy” vs “Mardu Energy”), we run an automated deduplication process:

1. Each archetype’s representative decklist is compared against every other archetype’s list.
2. Land cards are excluded from comparison so the focus is on the functional core of the deck.
3. If two archetypes share 90%+ of their non-land cards, the less popular one is merged into the more popular one.
4. The merged name is kept as an alias so existing links and searches still work.

Every decklist shown on an archetype page is a real deck played in a real tournament — never an average or synthetic composite. The specific list displayed is selected dynamically on each page load using a short-window clustering algorithm:

1. We collect all downloaded tournament decklists for that archetype from the past 7 days (expanding to 14 then 30 days if fewer than 3 lists are available).
2. Land cards are stripped from each list. Two lists that share the same 60-card non-land configuration but differ only in mana base are considered the same variant.
3. We count how many times each distinct non-land configuration appears. The most frequently occurring configuration is the dominant current build.
4. The best-placed real deck from that group is displayed, complete with its actual lands and sideboard.

When this live computation succeeds, a Live build badge appears in the page header. If insufficient recent data exists (e.g., a newly-tracked archetype or a format with few recent events), the page falls back to a stored representative updated by the daily scrape.

Card metadata (names, types, mana costs, images) is sourced from the Scryfall API.

Sideboard guides are AI-generated using your deck’s mainboard and sideboard as input, matched against each of the top meta archetypes. Each matchup incorporates the opponent’s typical decklist and available head-to-head win rate data to produce a concrete boarding plan: what to bring in, what to cut, and the strategic reasoning. Meta archetype guides are refreshed automatically each week, and guides for user-imported decks are generated in the background shortly after import.

Guides are ordered by live meta percentage so the most relevant matchups appear first. Each guide card shows how long ago it was last updated. Guides older than 21 days display an amber ⚠ warning to signal that the advice may not reflect recent meta shifts.

Registered users can submit suggestions for guide edits — proposing different cards in or out with optional reasoning. Other users can upvote or downvote suggestions. Approved suggestions are shown inline below the relevant guide. Community feedback with strong consensus will be incorporated into future guide regenerations.

Each archetype page also includes a What They Bring tab, which inverts the perspective: instead of showing how your deck boards against the field, it shows what each meta deck boards in and out when they face you — drawn from those archetypes’ own sideboard guides.

The Matchup Lookup tool lets you query a specific head-to-head directly: select the deck you’re playing and the deck you’re facing, and you get the sideboard plan and win rate for that exact matchup without navigating the full archetype page. A “Flip” link shows the reverse perspective.

Tournament data is refreshed once daily. Each scrape picks up any new tournaments published since the last run. Meta percentages are recalculated on every page load using the latest data, so the numbers you see are always current.

The representative decklist is also recomputed on every archetype page load — it always reflects the most common recent build at the time you visit, not a cached snapshot. Archetype deduplication runs with each scrape to catch newly-created duplicates.

Win rate data is refreshed each scrape cycle. Since mtgdecks.net aggregates match data over a longer period, win rates tend to be more stable than meta percentages.

• We do not create “average” or “optimal” decklists. Every list shown was actually played in a tournament.
• We do not include data from Magic Online leagues or Arena ladder — only tournament events with public results.
• Win rates are not adjusted for metagame context — a 55% win rate against a rarely-played deck means less than 55% against a top-5 archetype.