Compare

What this surface is

Compare puts two deck states side-by-side and explains what differs strategically — not just what cards changed. Inputs are two analyzed decks: two versions of the same deck, your deck vs a benchmark, two builds of the same archetype, or a budget vs full-power version.

Where to find it in the app

Navigate to /compare from the main nav, or enter from a deck’s version timeline (Deck versions). URL pattern after pairing: /compare/<a-id>/<b-id>.

Screenshot pending /img/docs/reference/compare-overview.png

The compare view — read top-down. Takeaways are the conclusion, diff is the evidence.

Layout

Region	Purpose
Top picker	Left + right deck slots, swap, change picker
Takeaways panel	Strategic-language summary of the delta
Coverage delta	Per-dimension before/after with deltas highlighted
Simulation delta	Observed-rate changes
Findings delta	Pegasus-detected pattern changes
Suggestions panel	Diagnosis-paired swap proposals
Card diff	Raw additions / removals / quantity changes at the bottom

Pair picker

Field	Type	Notes
Left deck	dropdown	Pick a deck, then a version
Left version	dropdown	Defaults to current
Right deck	dropdown	Same
Right version	dropdown	Same
Benchmark mode	toggle	Replaces right side with curated benchmark lists
Compare button	action	Opens the compare view

Useful pairings:

• Your v1 vs your v2 — testing a specific change
• Your list vs a benchmark meta list — measuring distance from a known-good build
• Two builds of the same archetype — comparing engine variants
• Budget version vs full-power — quantifying what budget costs you

Takeaways panel

Each takeaway is one strategic-language sentence. Common shapes:

Takeaway pattern	Example
”X density up / down"	"starter density increased"
"X risk up / down"	"brick risk decreased"
"X matchup improved / worsened"	"matchup vs combo improved (interaction density up)"
"Package change"	"swapped Adventurer package for Sky Striker engine"
"Style shift"	"build shifted from combo toward stun”

Takeaways are the conclusion of the comparison. Read them before the raw card diff — the diff tells you WHAT, takeaways tell you WHY it matters.

Coverage delta panel

Visualization	What it shows
Side-by-side radars	Both decks’ radar overlaid
Delta heatmap	Per-dimension +/- in cell color
Per-axis deltas	Numerical change with significance markers

A green-only delta is suspicious — usually means the model didn’t see the cost of the change. Look for what got worse.

Simulation delta

Metric	Format	Notes
Starter access change	+X%	Sub-percent deltas inside noise
Extender access change	+X%	Same
Interaction access change	+X%	Same
Dead hand change	+X%	Sub-percent deltas inside noise
Engine online change	+X%	Sub-percent deltas inside noise

Sample-size context: at 10,000 hands, ≤2% deltas are noise. Pegasus flags significant deltas with a confidence chip.

Suggestions panel

Each suggestion has a strict three-part shape:

Part	Required content
Problem	The specific weakness diagnosed (with the metric backing it)
Suggestion	The candidate swap
Tradeoff	What you give up to take it

A suggestion without a stated problem is just an opinion. Pegasus should never show one — if you see a bare “add card X” suggestion, the problem statement is hiding somewhere; expand the row.

Card diff

Column	What it shows
Card	Card name + set code
Left qty	Copies in the left deck
Right qty	Copies in the right deck
Delta	Right minus left (positive = added, negative = removed)
Role tag	Pegasus’s inferred role (starter / extender / payoff / interaction / etc.)

States and badges

Badge	Meaning
”Same archetype”	The two decks share the engine identification
”Different formats”	Comparison still works but interpretation needs care
”Stale analysis”	One side’s analysis predates the deck list
Confidence chip on a delta	High / medium / low — based on sample size and effect size

Interactions

• Swap left/right — toolbar button reverses the comparison
• Save the right side as a new version — toolbar action if right is a draft
• Apply a suggestion — opens a new draft version with the swap pre-applied
• Click a takeaway — drills into the evidence (metric + cards)
• Export comparison as PDF — toolbar action (sprint-future)

Edge cases

Achtung

• Comparing decks with no analysis runs. The comparison degrades to just card diff — you lose 80% of the value. Run Quick Analysis on both sides first.
• Comparing unrelated archetypes. Coverage deltas will read extreme on every axis; takeaways become “everything changed”. Pair against something at the same maturity level / archetype.
• Reading card diff before takeaways. Biases interpretation toward “cards changed” thinking instead of “structure changed”.
• Accepting suggestions without the problem statement. Don’t. If you can’t see the diagnosis, expand the row.

Related workflows

Compare and improve decks End-to-end workflow for pair → delta → swap → save.

Manage deck versions The versioning pattern that supplies pair inputs.

Related references

• Quick Analysis — required input on both sides
• Deck versions — the versioning surface that pairs into compare
• Scoring model — coverage and simulation methodology