So the dataset is pretty weak at the moment. I am currently working on the data model for synergies. It's very complicated to do in a way that is somewhat future proof; what I want is for a synergy to be of any two of something. Those somethings could be:MrMystery314 wrote: ↑4 years agoWhat I'm surprised by is that the main game plan being Worldgorger combo isn't something relevant by the ranking system (Grusilda is partially the commander because it wins off a full graveyard and infinite mana from the command zone without needing anything extra in hand), or Kiki-Jiki being counted as a value piece and not because it's part of the combo win from a single Buried Alive. Recognizing some of those combos (Kiki-Jiki without a way to combo is far weaker) could result in a more effective ranking system. Personally, I'd rank the deck as a strong 8 or a weak 9, effectively being cEDH if not solely optimized for being a combo deck.
Any ranking system is going to excel at spotting generally good cards, but without lots of effort won't recognize synergies and that sort of thing, which players looking at a deck may use to rank it either higher or lower than any automated system.
1) Card types and subtypes (ex. an instance of Enchantment may modify Verduran Enchantress of Tuvasa as a commander)
2) Card tags (ex. ramp cards or tutors)
3) Card as a commander (ex. thrasios)
4) List of Card(s) in deck (ex. kiki-jiki) (
This is probably a weekend project to finish up, but my basic thinking is basically a 3 column row of:
criteria 1, criteria 2, points
With criteria being prefixed by something to identify the type of criteria, which would allow me to add new criteria in the future with additional prefixes. e.g.
cards:["Kiki-Jiki, Mirror Breaker"]
cards:["Dramatic reversal", "Isochron Scepter"]
commander:"Thrasios, Triton Hero"
tag:"Ramp"
type:"enchantment"
subtype:"elf"
Your point value is then per instance; for example:
tag:"Ramp" + tag:"Bomb" = 1 point (this could be say 100 points pretty easily if you have 10 ramp spells and 10 bombs)
card:"Serra's Sanctum" + "type:"enchantment"" = 2 points (so 2 points per enchantment in your deck if you have serra's sanctum)
There are a lot of gaps in this approach. Mainly that it doesn't allow you to do gradients or whatever you want to call it, where the more ramp you add the more your bombs count -- so 10 pieces of ramp makes bombs worth 5, and 20 pieces of ramp makes bombs worth 10, or whatever.
But so far it feels like the best combination of simple to maintain and effective.
One thing I am seeing is the need for a form front end for all this nonsense too, since Excel/Sheets is going to get messy with people typing; really need a front end with some intelligence and a card selector at minimum for mass consumption. But I don't want to do that