Magic Math – A New Way to Determine an Aggregate Deck List: RG Dragons

Back in 2008, I made the Top 8 of the World Championships with an aggregate Faeries list. Coming into the event, I knew I wanted to play Faeries but didn’t have sufficient time to find the best build via playtesting. Instead, I used the wisdom of nearly a hundred Faeries players before me: Out of the lists available online, I took the statistical average and registered it for the event. The deck featured some seemingly awkward choices, such as 1 Ponder and 1 Broken Ambitions, but it still ran smoothly enough to carry me to the Top 8.

Nowadays, average lists of top-tier Standard archetypes are readily available. It’s a great way to communicate what a deck typically looks like and a quick source for a decent list, but the pure average has some issues. In this article, I will explain three of them, propose a new method that circumvents them, and apply it to RG Dragons in Standard.

Issue 1: The average list may contain inferior cards

Consider the following three Modern Burn deck lists (obviously simplified to illustrate my point):

The average burn list contains 3 Searing Blaze, 1 Searing Blood, and 56 Mountain, even though none of the decks chose the first Searing Blood over the fourth Searing Blaze. The three decks disagreed on how many of these effects to run, but all of them favored Searing Blaze over Searing Blood. Yet this preference is not captured in the average list.

Issue 2: The average list may lack synergy

Consider the following four Standard Abzan deck lists:

The average Abzan list contains 2 Sylvan Caryatid, 2 Elspeth, 2 Read the Bones, 1 Fleecemane Lion, 1 Rakshasa Deathdealer, 1 Dromoka’s Command, 1 Whip of Erebos, 1 Satyr Wayfinder, 1 Hornet Queen, 4 Siege Rhino, and 44 Sandsteppe Citadel. This is a awkward mess of a deck. It lost all the synergies of the individual lists because it combined cards that don’t yield a coherent game plan together.

The issue is particularly blatant here because I mingled distinct archetypes for illustration, but distinguishing between archetypes is tough in general, and there can even be enough variation within a single archetype to run into this type of problem.

Issue 3: The average list contains fractional amounts of cards

If the average list tells you to play 1.4 copies of one card and 0.6 copies of another, then what? I would happily construct a deck with a pair of scissors and a ruler if I could, but we have to adhere to natural numbers in Magic.

An alternative way to determine an aggregate deck list

To mitigate the issues described above, I’ll propose a different approach:

  • Step 1: In all of the deck lists you wish to take the aggregate from, split up multiple copies of any one card into separate entries. So, 3 Searing Blaze becomes 1 first Searing Blaze, 1 second Searing Blaze, and 1 third Searing Blaze. Eventually, any 60-card deck will be split up into 60 distinct “numbered” card names.
  • Step 2: Count how often every “numbered” card name occurs in all of the main decks, and then sort them in descending order of occurrence (with ties broken by alphabetical ordering of the card name). So, we’d get 3 first Searing Blaze, 2 second Searing Blaze, and so on.
  • Step 3: Take the first 60 “numbered” card names from the top of the sorted list—these form the aggregate main deck. Finally, repeat all steps for the sideboard.

With this method, the resulting deck lists for the examples above would be:

I believe these results make more sense than the pure average list. My method can be improved upon, in particular with a better way of breaking ties than alphabetical ordering, but the idea behind it seems solid.

As a final illustration, I’ll apply this method to all eight RG Dragons deck lists that finished in the Top 32 at last weekend’s SCG Open. It has quickly become one of the most popular Standard decks, and here is the aggregate list:

The 61st card in the main deck would be the third Surrak, and the 16th card in the sideboard would be the first Barrage of Boulders or the first Arc Lightning. I don’t know whether this aggregate list is better or worse than Chris VanMeter’s winning version with Courser of Kruphix and Sylvan Caryatid, but if you’re interested in RG Dragons and trust the wisdom of the crowd, then pick the above aggregate list over an awkward average with 1 Courser of Kruphix and 1 Sylvan Caryatid.


Scroll to Top