“This sort of innovation occurs when you set out to improve an electric tea kettle and misplace a decimal.”
- Cosmotronic Wave
Ponder uses a Linear Programming techniques to take a large set of cards (more than the deck size), a bunch of constraints, and find the optimal* combination of cards.
Optimal deck not guaranteed.*
The cards are input as a CSV, with names in the first column, categories for the cards as the header, and for each combination a score of some sort.
| Card Names | Card Draw | Ramp | Damage |
|---|---|---|---|
| Ponder | 1 | 0 | 0 |
| Cultivate | 0 | 3 | 0 |
| Lightning Bolt | 0 | 0 | 3 |
Here, our constraints are the 'Card Draw', 'Ramp' and 'Damage'.
After 'Solving' the constrains and cards, the program outputs three different pieces of information.
Vertices are the solutions to individual constraints, which can act as proposed decks.
Any cards with a score of 0 are not included in the proposed decks.
The midpoint is the arithmetic mean of all decks.
A summary of the mins and maxes of all decks.
Tuning the decks can be done inside index.js. Add or remove constraints to the tacticsConstraints entries, this can be informed from the ranges object, and the midpoint. For instance, often a certain amount of Ramp is recommended for all decks, which could be enforced with [['Ramp', { min: 10 }]].
The scores in the sample_data csv can be adjusted. Currently they are scored 0-3, but if you only included good cards, ranking them 0 - 1 could allow you to enforce 10 ramp cards regardless of quality.
Deck size, card limit and land count can be adjusted as well inside index.js
To run the program, first check it out from github, and install all dependencies:
npm install
Then call the main program with
npm run opt
This programe does not (yet) do the following:
- Account for mana costs
- Prevent you from spending all night searching Scryfall for more cards
- Account for pet cards
- Lockdown due to pandemics
- Having to read papers on Multi-objective Knapsack problems
- Account for converted mana cost
- Unfair moderation on Reddit
- Terrible choice of cards in the example data
- Account for a good Mana Curve
- Convince you to run more lands in EDH
- Stop WotC from concealing top 8 lists
- Prevent being posted too often on Reddit
- Account for cards already in your collection