Marvel Champions Codex
About this site
The Codex is a reference and research tool for Marvel Champions: The Card Game. It looks at thousands of decks that real players have shared, finds the patterns in them, and hands those patterns back to you so you can look something up, dig into how a hero gets built, or get unstuck on your next deck. This page explains how all of that works, in plain language. None of it is magic, and none of it is the final word.
Where the numbers come from
Everything here is built from public decklists on MarvelCDB. Once a night, a build job fetches the latest decks, cleans them up, and works out all the patterns ahead of time. The results get saved as plain data files that your browser downloads when you open a page.
That has a few nice consequences. There is no account and no login. Nothing you type is sent anywhere; the searching and matching all happen on your own device. And because the heavy lifting is done overnight, the pages stay fast. The trade-off is that the data is as fresh as last night's build, not this second.
We also fold reprints together, so a card that has appeared in several products counts as one card, and we set aside decks that look incomplete or broken so they don't skew the analysis.
How to read the numbers
One thing worth saying up front: this is a description of what people play, not a ranking of what is good. A card being in 90% of decks means it is popular, which usually means it is reliable, but popularity and power are not the same thing, and a fresh idea that nobody has tried yet will look unpopular precisely because it is new.
So treat the numbers as a well-informed starting point. They are great for "what do people usually do here?" and "what am I missing?" They are not a substitute for your own read of a card or a meta. When a number is built on only a handful of decks, we try to say so, because a pattern from five decks is a hint, not a fact.
Search
Why: the most common question is simply "show me decks like the one I have in mind." Search answers it.
You describe a deck the same way you would build one: pick a hero, pick an aspect, list a few cards it should run, and list any it should not. Search then sweeps the whole corpus, keeps the decks that fit, and ranks them. Alongside the matching decks you also get the clusters, packages, and individual cards that fit your description, so one query doubles as a research tool.
The strictness control decides how much of your card list a deck has to share to count. At full strictness a deck must run everything you listed; loosen it and decks that match most of your list come through too.
Good for: finding real decks to copy or learn from, and seeing what a hero-and-aspect pairing tends to look like.
Worth knowing: if you ask for a very specific combination, you may only get a few decks back. That is real information, not a bug. A small number means few people have built it that exact way.
Card suggestions
Why: when you are building, the useful question is "what else goes in here?" We answer it three different ways, because there are three different things you might mean.
Popularity
The share of decks in your context that run a card. If you have picked Spider-Man and Justice, a card's popularity is how often Spider-Man Justice decks include it. This surfaces the staples: the cards almost everyone runs, the safe includes.
Synergy
How much more a card shows up in your context than in decks generally. A card that is everywhere will look ordinary here even if it is popular, while a card that is special to your hero or aspect rises to the top. This is the "what makes this build distinctive?" view.
Lift
Similar idea to synergy, but as a multiplier rather than a difference: a lift of 5 means a card appears five times more often in your context than at large. Lift is good at spotlighting tight pairings, but it gets jumpy when a card is rare, so we only trust it once there are enough decks behind it.
The popularity basis
Newer cards have a built-in disadvantage: a card released last month cannot appear in a deck from last year, so a plain all-time percentage makes everything new look unpopular. To handle that, you can choose the basis the percentages are measured against:
- Smoothed (the default) leans toward fairness for new cards while staying steady on small samples. A sensible middle ground.
- Co-availability measures a card only against decks built after it existed. Honest, but it can read very high on a brand-new card that only a few decks have had the chance to try.
- All-time is the raw share across every deck. Simple, but it understates anything recent.
Good for: Popularity answers "what should I almost certainly include," synergy answers "what makes this deck its own thing," and they are meant to be read together.
Worth knowing: these describe habits, not laws. A staple being popular does not mean it is right for your plan, and a card with no synergy score yet may just be too new or too rare to have a pattern.
Packages
Why: cards are rarely added one at a time. They come in little bundles that work together, like the X-Men cards or the Ready for a Fight cards. Packages are us finding those bundles automatically.
We look at which cards keep showing up together across thousands of decks and group the ones that travel as a unit. Unlike clusters, packages overlap on purpose: a single deck is usually a mix of two or three packages, the way a real deck is a base plus a couple of themes.
We build packages three ways, and you can switch between them:
- By aspect (the default) finds bundles inside each aspect on its own, so everything in a package is legal to play together. This is the best answer to "what pairs with this in my aspect?"
- Whole-pool looks at the entire card pool at once and sorts every card into exactly one package. A cleaner big-picture map, but it can lump a lot together.
- Overlapping lets a card belong to several packages at once, which is good for the glue cards that bridge a few different ideas.
Good for: "I have this one card, what comes with it?" and for understanding a deck as a few building blocks rather than 40 separate choices.
Worth knowing: a package is a tendency, not a rulebook. You do not have to run all of one, and the line between two related packages is a judgment call the math makes, so it will not always match how you would split them.
Clusters
Why: people talk about "archetypes," but real decks do not fall into tidy boxes. Clusters are our attempt to find the recurring ways of building from the decks themselves, instead of deciding the categories in advance.
We compare decks by the cards they share and group the ones that look alike. Each group is a cluster: a way of playing that the community keeps returning to. We prefer the word "cluster" over "archetype" on purpose, because it is honest about what it is. It is a cloud of similar decks with a fuzzy edge, not a fixed label, and a single deck can sit near the border of two of them.
There is more than one way to slice the corpus, so clusters come in a few flavors:
- The default groups decks at a comfortable middle resolution: broad, recognizable styles.
- Fine-grained turns the dial up, so smaller and more specific styles form their own clusters instead of being folded into a bigger neighbor. Good for spotting newer or niche ideas.
- Mixture-based groups each deck by the package it leans on most, which tends to produce tighter, more focused clusters.
The Frontier
Some decks do not fit any cluster well. The Frontier collects those: the decks our grouping explains least, ranked by how unusual they are. It is where to look for genuinely off-beat ideas, with the caveat that "unusual" sometimes just means "unfinished or quirky."
Good for: getting the lay of the land for a hero, and finding a established style to build toward.
Worth knowing: the boundaries and the names are generated, so they can shift as the corpus grows, and a cluster's name is just its most defining card, not an official title.
Similar decks & recommendations
Why: once you are looking at one deck, two follow-up questions come up naturally: "what else is like this?" and "what would people add?"
Similar decks finds the decks whose card lists overlap most with the one you are viewing. Decks like this also run looks at those neighbors and surfaces cards they share that yours is missing, along with how strong and how common that pattern is.
Good for: branching out from a deck you like, and spotting an obvious include you forgot.
Worth knowing: it is evidence, not advice. It tells you what tends to go together, and the call on whether it fits your plan is still yours.
The card map
Why: sometimes you want to wander rather than search. The card map is a picture of how cards relate.
Cards that often appear together sit near each other, so the map forms little neighborhoods that line up with the packages. It is a way to explore by eye: start at a card you know and see what it keeps company with.
Good for: open-ended browsing and getting a feel for how a corner of the card pool hangs together.
Worth knowing: flattening a complicated web of relationships onto a flat picture always loses some detail, so read it as a sketch, not a blueprint.
Rules & errata
Why: mid-game you just want the answer, fast.
The rules reference is the official Rules Reference Guide, broken into one short article per topic and cross-linked, with search that matches as you type. The errata section collects official rulings and community clarifications so you can check whether a card works the way you remember.
Good for: settling a question at the table without leaving the page.
Worth knowing: always defer to the latest official documents; we link out to the sources so you can confirm.