An Introduction to the Commander Map

The Commander Map is a visualization of the wide world of Magic’s Commander format! Each point on the map represents an individual Commander deck collated by EDHREC. The decks are algorithmically organized based on the cards they contain: closer points share more cards. From this starting point, the algorithm uncovers trends including colors, commanders, deck strategies, budgets, and other forces that influence deck design.

Use the map to explore the full range of the Commander format, or focus in on single themes or commanders to discover hidden diversity.

The map is generated algorithmically starting with the decklists of every deck on EDHREC. While the mathematical details are complex, the principle is straightforward.

We can think of the presence or absence of each card, in a mathematical sense, as a dimension and plot each deck in a many-dimensional space. To start, we could plot deck on a two-dimensional scatter plot based on the presence or absence of two cards, for example, making one axis Counterspell and the other Rampant Growth. Every decklist can be plotted as a point on this graph in one of the four corners based on whether they play one, both, or neither cards. We can add a third card to the graph, expanding it into three dimensions. If we do this for every Magic card, we can plot every Commander deck in a 20,000+ dimensional space!

While a many-dimensional graph is mathematically rigorous, it’s not very useful in itself to humans who can’t visually perceive more than three dimensions. To generate an readable map, we have to flatten the data into a two-dimensional form while maintaining as much of the patterns that define the Commander format as possible. Flattening dimensions is something we experience daily in 2 dimensional shadows of 3d objects. By rotating an object, we can find a shadow that preserves it’s most important details. Creating a flat map from the globe is another way we project data into a readable shape. A map won’t perfectly replicate exact distances, sizes, or relative positions of features on a globe, but we can see all those features in an understandable way and preserve their individual shapes.

The algorithm we’re using, called UMAP, is used in a wide range of scientific fields to better understand the shape of complex data. With the shift in our perspective to think of cards as dimensions, we can use decklists as the algorithm’s input to generate the map.

Taken as a whole, the map visualizes the breadth and depth of the world of Commander, showing the vibrant creativity of deckbuilders. Zooming in, the map reveals details and trends that aren’t visible when we visualize just the most impactful features of the format.

As you explore the map, highlighting decks will reveal what different regions are all about. The commanders and colors will give you a quick impression of the region, and you can inspect everything from themes and defining cards all the way down to individual decklists. Working on the opposite direction, highlight aspects of decks like colors, commanders and themes to see how they’re distributed across the map and how they form trends and clusters across larger geographic structures.

To get even more detail, you can focus on any one particular color, commander, tribe, or theme, and see a map of just those decks. Each of these maps uses the algorithm on a subset of decks, often revealing more nuanced patterns where the most obvious, distinguishing characteristics aren’t in play. For example, in the map for a particular commander, decks may separate by budget, theme, or entirely different restrictions. If you look hard enough, you might even be able to find the Chair Typal cluster on the Oloro, Ageless Ascetic map!

Want to know more about the math behind generating the map? Read our more in-depth article about making a map from the 20k+ dimensional card space.

We’ve also applied the same algorithm to every Cube on Cube Cobra to make the Cube Map. Since Cubes tend to include all colors and a multiplicity of themes, even more complex and nuanced patterns emerge in this map.