Scoring cards for Ashes: Rise of the Phoenixborn

Scoring cards for Ashes: Rise of the Phoenixborn

I used the results of the Summer Slam Jam 2 online Ashes tournament to experiment with a rough system of scoring the relative strength of cards. I tried two different approaches and hope to have some feedback from players who are more active than I am: Are the results interesting, and which of the two approaches seems more accurate?

Edit @ 2018-07-18: niu pointed out that this Summer Slam Jam 2 was in the Ashes 500 format. So, consider the scoring values here limited to that format.

The full results are browseable in this Google Sheet:

Ashes - Summer Slam Jam 2 Card Scoring

Screenshot of spreadsheet

How the scores are generated

I started by examining the decklists for (nearly) every deck in the Summer Slam Jam 2 tournament published on the deckbuilding site, Ashes.live. Using the tournament results, I built a scoring system in which every card in the winning decks "beats" every card in the losing decks.

Let's take the first matchup between Mbauers and jbensley as an example.

Screenshot of Round 1 Match 1 Challonge entry for Mbauers and jbensley

Given that Mbauers wins that round of play, the scoring system takes each card of Mbauers' deck and says that it "wins" against each card of jbensley's deck. For example, Mbauers' first copy of Chant of Revenge beats jbensley's Hypnotize, Magic Syphon, the first Summon Glider, the second Summon Glider, and so on:

Mbauers' winning deck card wins jbensley's losing deck card losses
Chant of Revenge #1 1 beats Hypnotize 1
Chant of Revenge #1 2 beats Magic Syphon 1
Chant of Revenge #1 3 beats Summon Glider #1 1
Chant of Revenge #1 4 beats Summon Glider #2 2
...
Chant of Revenge #1 30 beats Redirect #3 (final card in jbensley's deck) 1

The scoring system moves on to the next card: Mbauers' second copy of Chant of Revenge beats jbensley's Hypnotize, Magic Syphon, the first Summon Glider, the second Summon Glider, and so on.

Mbauers' winning deck card wins jbensley's losing deck card losses
Chant of Revenge #2 31 beats Hypnotize 2
Chant of Revenge #2 32 beats Magic Syphon 2
Chant of Revenge #2 33 beats Summon Glider #1 3
Chant of Revenge #2 34 beats Summon Glider #2 4
etc.
Chant of Revenge #2 60 beats Redirect #3 (final card in jbensley's deck) 2

We have full control of what it means for a card to win and lose in a matchup. Given the pattern above, one of the easiest things to tweak is the score value awarded for a win and deducted for a loss. There are two approaches profiled in the results, which appear to have some variation―a minor tweak in the scoring approach (wins worth 2 points vs. wins worth 1 point) produces these distinct Top 11 cards:

Approach 1
  1. Final Cry
  2. Summon Salamander Monk
  3. Hidden Power
  4. Call Upon The Realms
  5. Chant of Revenge* (missing from other approach's Top 11)
  6. Squire
  7. Join The Hunt
  8. Redirect*
  9. Fire Archer*
  10. Crescendo*
  11. Hunt Master*
Approach 2
  1. Hidden Power
  2. Call Upon The Realms
  3. Final Cry
  4. Spiked Armor* (missing from other approach's Top 11)
  5. Anchornaut*
  6. Frost Bite*
  7. Join The Hunt
  8. Squire
  9. Summon Salamander Monk
  10. Iron Worker*
  11. Summon Weeping Spirit*

Here's how the approaches work in greater detail.

Approach 1

In approach 1, a winning card's score goes up by two points for each card it beats. A losing card's score goes down by one for each card that defeats it. In this model, a card that wins as much as it loses can still score highly, especially when the card is played in a lot of matchups.

Statline for Redirect in Approach 1
rank card title picks score W L W-to-L
8 Redirect 1080 540 540 540 1.0

The "Picks" counter is incremented every time a card is involved in a matchup on the winning or losing side.

Approach 2

Alternatively, in approach 2, a winning card's score goes up by one point instead of two, and it decreases by one point when it loses. Since wins and losses exactly offset each other in this scheme, this scoring cares more about efficiency―it penalizes cards for losing and prevents them from climbing up the score chart.

Statline for Redirect in Approach 2
rank card title picks score W L W-to-L
41 Redirect 1080 0 540 540 1.0

Why try scoring cards?

As a Netrunner player, with occasional stints into Ashes, I really admire the Ashes 500 format, a community-generated set of deckbuilding rules for the game. The format involves assigning point values to every card, then setting a limit of 500 "points" for the decks used to play the game.

This homegrown approach enabled the community to keep the game interesting during an extended lull. By ratcheting up the point values of certain cards and making them more "expensive" to play with, the format's adminstrators could nudge the metagame around and provide players with new challenges. "We've all seen the Frostback Bear an awful lot, so if you put that card in your deck, you may not be able to use other tools."

With that in mind, I've always wondered if there would be a way to design a self-correcting scaffolding around such a point-based system for Ashes or any customizable game. Could deckbuilding function more like a working market, where "oversold" cards would become increasingly expensive with each inclusion and with each victory?

In such a setting, a tournament's winning cards would rise in cost proportionate to their value, which would constantly challenge players to come up with a less expensive way to win. It would be a way to balance deckbuilding on an ongoing basis using data that competitive card game players are already avidly consuming and producing―the results of major tournaments.

The Summer Slam Jam card scoring experiment above is almost certainly too naive to fit this imagined use case, even though the concept was the inspiration. I imagine that there's tons of room for improvement and experimentation. Maybe the model could be refined to incorporate more of the things that human decision-makers tend to consider when making metagame management decisions, like assessing the prevalence of broken combos (what are the winningest pairs, trios, any combinations of cards across the competitive landscape), imagining the impact of restricting or banning cards.

Questions for players

I'm having fun and would be keen to hear any feedback. Let me know what you think! In particular: