[Official] Dimensional Shield | Phase Shield

[OldTrees]

Why is dusk 'better', because you dont need to carry as many

This is wrong. You have to play test before you can make claims like this. There may be some decks where 1 dusk shield is better than 1 dim shield, but in general, this is not true. Based on simulations, I would estimate that 1 dusk shield improves your chance to win by 17% against straight rush, where as 1 dim shield improves your chance to win against straight rush by 26%. A single dim shield unanswered will win the game for you. A single dusk shield unanswered might give you an edge.

Add CC to the simulation. I suspect that 1 Dusk shield would be better than 1 Dim shield if the offense is restrained.

[majofa]

1 turn of Dimensional Shield before it gets destroyed will ALWAYS be better than 1 turn of Dusk Mantle before it gets destroyed.
Same with 2 and 3 turns. It's not until 4 turns that Dusk Mantle is better.

[jawdirk]

Add CC to the simulation. I suspect that 1 Dusk shield would be better than 1 Dim shield if the offense is restrained.

I will try this when I get the chance. I suspect that it depends on the damage ramp over the CC. If you are killing the creatures always on their 1st turn out, then you are probably right, because it is uneconomical to maintain a shield chain when there are no creatures. However, if your CC is insufficient, and your opponent is still ramping damage (at a reduced rate) then I think dim shield could still be better. It may depend on adjusting the AI to start playing dim shield at 30 HP instead of 50 though.

It would be interesting if we discovered that dusk shield is only better when your deck contains a significant amount of CC.

[storyteller]

I dont think you have 'found' anything/. your code is flawed by oversimplification.

[jawdirk]

I dont think you have 'found' anything/. your code is flawed by oversimplification.

Yah science sucks am I rite?
'Merica
\S

[OldTrees]

I dont think you have 'found' anything/. your code is flawed by oversimplification.

Yah science sucks am I rite?
'Merica
\S

It is inappropriate of a scientist to throw Ad Hominem attacks when people remind them that all models have flaws as a direct result of their simplification of reality.

[Quantumage]

Honestly, it's clear that from the simulation that dusk shield is a little UP. Dusk shield costs 6 for 50%, while fog shield costs 2 for 40%. I suspect that one could determine phase shield's true cost if they modified the simulation comparing it to fog shield instead.

[jawdirk]

It is inappropriate of a scientist to throw Ad Hominem attacks when people remind them that all models have flaws as a direct result of their simplification of reality.

Completely inappropriate, scientist or otherwise, I would say. 

Any way, here is some more of that science stuff:

The code:

Code: [Select]


def log *strings
  # uncomment to output blow-by-blow
  # puts *strings
end

class Game
  attr_accessor :player1, :player2, :turn_count
 
  def initialize(player1, player2)
    @turn_count = 0
    @player1, @player2 = player1, player2
  @player1.game = self
    @player2.game = self
  end
 
  def turn
    @turn_count += 1
  log "Turn #{@turn_count}:::"
    @player1.turn(@player2)
  @player2.turn(@player1) unless over?
  end
 
  def over?
    @player1.hp < 1 || @player2.hp < 1 || @player1.cards.size == 0 || @player2.cards.size == 0
  end
 
  def deck_out?
    @player1.cards.size == 0
  end
 
  def to_s
    "#{@player1}\nvs.\n#{@player2}"
  end
end

class Player
  attr_accessor :hp, :quanta, :quanta_per_turn, :creatures_in_play, :card_hash,
    :cards, :hand, :name, :shield, :destroy_shield_on_turn, :game
 
  def initialize(name, hp, card_hash)
    @card_hash = card_hash
    @name = name
    @hp = hp
  @shield = 1
  @quanta = @quanta_per_turn = 0
    @creatures_in_play = []
    @cards = @card_hash.keys.map{|ca| ca * @card_hash[ca]}.flatten
    @cards = @cards.shuffle
  @cards = @cards.shuffle if cards[0..6].all?{|c| !c.is_a?(Pillar)}
  @cards.push(Dead.new)
  @hand = []
  7.times { draw }
  end
 
  def shield?
    shield < 1
  end
 
  def destroy_shield!
    self.shield = 1
  end
 
  def damage_per_turn
    creatures_in_play.inject(0){|acc,c| acc + c.damage }
  end
 
  def turn_count
    game.turn_count
  end
 
  def draw
    hand << cards.shift
  end
 
  def turn(enemy)
    destroy_shield! if destroy_shield_on_turn == turn_count
    draw
  log inspect
  @hand.sort_by{|c| c.cast_order}.each{|c| c.try_cast(self, enemy) }
  log "Boom!: #{enemy.hp} -= #{damage_per_turn * enemy.shield} (#{enemy.shield})"
  enemy.hp -= damage_per_turn * enemy.shield
  self.quanta += quanta_per_turn
    log inspect, "\n"
  end
 
  def deck_size
    cards.size - 1
  end
 
  def inspect
    #"~#{name} hp: #{hp}" q: #{quanta} cs: #{deck_size} c_in: #{creatures_in_play.inspect}~"
    "~#{name} hp: #{hp}  qpt: #{quanta_per_turn} q: #{quanta} cs: #{deck_size}
dpt: #{damage_per_turn} in play: #{creatures_in_play.inspect}
in hand: #{hand.inspect}~"
  end
 
  def to_s
    "#{name}(#{card_hash.inspect})"
  end
end

class Card
  def try_cast(me, enemy)
    cast!(me, enemy) if cast?(me, enemy)
  end
 
  def cast!(me, enemy)
    card_cast(me, enemy)
  me.hand.slice!(me.hand.index(self))
  end
 
  def to_s
    self.class.to_s
  end
end

class Dead < Card
  def cast?(me, enemy);  true ; end
  def cast!(me, enemy); end;
  def cast_order; 0; end
end

class Pillar < Card
  def cast?(me, enemy); true ; end
  def card_cast(me, enemy)
    me.quanta_per_turn += 1
  end
 
  def cast_order; 0; end
end

class Creature < Card
  attr_accessor :damage, :health
  def initialize(cost, damage, health)
    @cost, @damage, @health = cost, damage, health
  end
  def cast?(me, enemy); me.quanta >= @cost; end
  def card_cast(me, enemy)
    me.quanta -= @cost
    me.creatures_in_play << self
  end
  def cast_order; 100 - damage; end
  def to_s
    "Creature(#{damage}|#{health})"
  end
end

class Bolt < Card
  attr_accessor :cost, :damage
  def initialize(cost, damage)
    @cost, @damage = cost, damage
  end
  def cast?(me, enemy); me.quanta >= cost && enemy.creatures_in_play.detect{|cc| cc.health <= damage }; end
  def card_cast(me, enemy)
    me.quanta -= cost
    target = enemy.creatures_in_play.sort_by{|cc| (cc.health > damage ? 1000 : 0) - cc.damage }.first
    log "Bolt!!!!! #{target}"   
    enemy.creatures_in_play.slice!(enemy.creatures_in_play.index(target))
  end
 
  def cast_order; 2; end
end

class Explosion < Card
  attr_accessor :cost
 
  def initialize(cost)
    @cost = cost
  end
 
  def cast?(me, enemy); me.quanta >= cost && enemy.shield != 1; end
 
  def card_cast(me, enemy)
    me.quanta -= cost
    log "explosion!!!!!"
    enemy.destroy_shield!
  end
 
  def cast_order; 1; end
  def to_s; "Explosion(#{cost})"; end
end
 
class DimShield < Card
  def initialize(lasting, cost)
    @cost = cost
    @lasting = lasting
  end

  def cast?(me, enemy); !me.shield? && me.quanta >= @cost && me.hp < 50; end
  def card_cast(me, enemy)
    me.quanta -= @cost
    log "dim shield!!!!!!!!!!!!! #{me.turn_count}"
    me.destroy_shield_on_turn = me.turn_count+@lasting
    me.shield = 0
  end

  def cast_order; 1; end

  def to_s
    "DimShield(#{@lasting},#{@cost})"
  end
end

class DuskMantel < Card
  def cast?(me, enemy); !me.shield? && me.quanta >= 6; end
  def card_cast(me, enemy)
    me.quanta -= 6
    log "dusk mantel!!!!!!!! ";
    me.shield = 0.5
  end
  def cast_order; 1; end
end

test_decks = [

 ->() {Player.new('control-4bolt', 100.0, [Pillar.new] => 15, [Creature.new(5,5,5)] => 11, [Bolt.new(2,5)]  =>  4)},
 ->() {Player.new('1dusk-4bolt', 100.0, [Pillar.new] => 15, [Creature.new(5,5,5)] => 10, [DuskMantel.new]  =>  1, [Bolt.new(2,5)]  =>  4)},
 ->() {Player.new('3dusk-4bolt', 100.0, [Pillar.new] => 15, [Creature.new(5,5,5)] => 8, [DuskMantel.new]  =>  3, [Bolt.new(2,5)]  =>  4)},
 ->() {Player.new('1dim-4bolt', 100.0, [Pillar.new] => 15, [Creature.new(5,5,5)] => 10, [DimShield.new(3,6)]  =>  1, [Bolt.new(2,5)]  =>  4)},
 ->() {Player.new('3dim-4bolt', 100.0, [Pillar.new] => 15, [Creature.new(5,5,5)] => 8, [DimShield.new(3,6)]  =>  3, [Bolt.new(2,5)]  =>  4)},
 ->() {Player.new('6dim-4bolt', 100.0, [Pillar.new] => 15, [Creature.new(5,5,5)] => 5, [DimShield.new(3,6)]  =>  6, [Bolt.new(2,5)]  =>  4)},
 ->() {Player.new('1dim10-4bolt', 100.0, [Pillar.new] => 15, [Creature.new(5,5,5)] => 10, [DimShield.new(3,10)]  =>  1, [Bolt.new(2,5)]  =>  4)},
 ->() {Player.new('3dim10-4bolt', 100.0, [Pillar.new] => 16, [Creature.new(5,5,5)] => 7, [DimShield.new(3,10)]  =>  3, [Bolt.new(2,5)]  =>  4)},
 ->() {Player.new('6dim10-4bolt', 100.0, [Pillar.new] => 16, [Creature.new(5,5,5)] => 4, [DimShield.new(3,10)]  =>  6, [Bolt.new(2,5)]  =>  4)}, 
 ->() {Player.new('1dim20-4bolt', 100.0, [Pillar.new] => 15, [Creature.new(5,5,5)] => 10, [DimShield.new(3,20)]  =>  1, [Bolt.new(2,5)]  =>  4)},
 ->() {Player.new('3dim20-4bolt', 100.0, [Pillar.new] => 17, [Creature.new(5,5,5)] => 6, [DimShield.new(3,20)]  =>  3, [Bolt.new(2,5)]  =>  4)},
 ->() {Player.new('6dim20-4bolt', 100.0, [Pillar.new] => 17, [Creature.new(5,5,5)] => 3, [DimShield.new(3,20)]  =>  6, [Bolt.new(2,5)]  =>  4)},
 # ->() {Player.new('control', 100.0, [Pillar.new] => 16, [Creature.new(5,5,5)] => 14)},
 # ->() {Player.new('1dusk', 100.0, [Pillar.new] => 16, [Creature.new(5,5,5)] => 13, [DuskMantel.new]  =>  1)},
 # ->() {Player.new('3dusk', 100.0, [Pillar.new] => 16, [Creature.new(5,5,5)] => 11, [DuskMantel.new]  =>  3)},
 # ->() {Player.new('1dim', 100.0, [Pillar.new] => 16, [Creature.new(5,5,5)] => 13, [DimShield.new(3,6)]  =>  1)}, 
 # ->() {Player.new('3dim', 100.0, [Pillar.new] => 16, [Creature.new(5,5,5)] => 11, [DimShield.new(3,6)]  =>  3)},
 # ->() {Player.new('6dim', 100.0, [Pillar.new] => 16, [Creature.new(5,5,5)] => 8, [DimShield.new(3,6)]  =>  6)}, 
]
test_decks.each do |test_deck|
  damage = 0
  wins = 0
  losses = 0
  deck_outs = 0
  reps = 10000

  start = ->() {
    Game.new(
      Player.new('rush', 10000, [Pillar.new] => 16, [Creature.new(5,5,5)] => 14),
      #Player.new('blast', 10000, [Pillar.new] => 15, [Creature.new(5,5,5)] => 13, [Explosion.new(3)] => 2),
      test_deck.call
    )
  }

  reps.times do
    game = start.call
    game.turn until game.over?
    damage += 10000 - game.player1.hp
    wins += game.player1.hp <= 9900 ? 1 : 0
    losses += game.player1.hp > 9900 && game.player2.hp <= 0 ? 1 : 0
    deck_outs += game.player1.hp > 9900 && game.player2.hp > 0 ? 1 : 0
  end
 
  overkill = damage*1.0/reps
  win_p = wins*100.0/reps
  loss_p = losses*100.0/reps
  deck_out_p = deck_outs*100.0/reps

  puts "#{start.call.inspect} (#{reps} games)",
    "overkill: #{overkill}",
    "win%: #{win_p}",
    "loss%:#{loss_p}",
    "deckout%: #{deck_out_p}\n\n"
end


Notes:
Bolts are never used against the opponent, only against creatures.

About the results:
overkill is the average damage ignoring wins but not losses until all of player 1's cards are exhausted.
Wins, losses, and deck outs are recorded separately.

Results for Rush deck vs. shields + 4 bolts:

Spoiler for Hidden:

rush({[Pillar]=>16, [Creature(5|5)]=>14})
vs.
control-4bolt({[Pillar]=>15, [Creature(5|5)]=>11, [Bolt]=>4}) (10000 games)
overkill: 99.9075
win%: 49.37
loss%:50.63
deckout%: 0.0

rush({[Pillar]=>16, [Creature(5|5)]=>14})
vs.
1dusk-4bolt({[Pillar]=>15, [Creature(5|5)]=>10, [DuskMantel]=>1, [Bolt]=>4}) (10000 games)
overkill: 120.3415
win%: 66.66
loss%:33.34
deckout%: 0.0

rush({[Pillar]=>16, [Creature(5|5)]=>14})
vs.
3dusk-4bolt({[Pillar]=>15, [Creature(5|5)]=>8, [DuskMantel]=>3, [Bolt]=>4}) (10000 games)
overkill: 124.0275
win%: 74.02
loss%:25.98
deckout%: 0.0

rush({[Pillar]=>16, [Creature(5|5)]=>14})
vs.
1dim-4bolt({[Pillar]=>15, [Creature(5|5)]=>10, [DimShield(3,6)]=>1, [Bolt]=>4}) (10000 games)
overkill: 127.5435
win%: 70.21
loss%:29.79
deckout%: 0.0

rush({[Pillar]=>16, [Creature(5|5)]=>14})
vs.
3dim-4bolt({[Pillar]=>15, [Creature(5|5)]=>8, [DimShield(3,6)]=>3, [Bolt]=>4}) (10000 games)
overkill: 205.901
win%: 90.65
loss%:9.35
deckout%: 0.0

rush({[Pillar]=>16, [Creature(5|5)]=>14})
vs.
6dim-4bolt({[Pillar]=>15, [Creature(5|5)]=>5, [DimShield(3,6)]=>6, [Bolt]=>4}) (10000 games)
overkill: 303.576
win%: 97.27
loss%:2.72
deckout%: 0.01

rush({[Pillar]=>16, [Creature(5|5)]=>14})
vs.
1dim10-4bolt({[Pillar]=>15, [Creature(5|5)]=>10, [DimShield(3,10)]=>1, [Bolt]=>4}) (10000 games)
overkill: 117.927
win%: 61.0
loss%:39.0
deckout%: 0.0

rush({[Pillar]=>16, [Creature(5|5)]=>14})
vs.
3dim10-4bolt({[Pillar]=>16, [Creature(5|5)]=>7, [DimShield(3,10)]=>3, [Bolt]=>4}) (10000 games)
overkill: 170.6665
win%: 78.56
loss%:21.44
deckout%: 0.0

rush({[Pillar]=>16, [Creature(5|5)]=>14})
vs.
6dim10-4bolt({[Pillar]=>16, [Creature(5|5)]=>4, [DimShield(3,10)]=>6, [Bolt]=>4}) (10000 games)
overkill: 240.534
win%: 92.87
loss%:6.87
deckout%: 0.26

rush({[Pillar]=>16, [Creature(5|5)]=>14})
vs.
1dim20-4bolt({[Pillar]=>15, [Creature(5|5)]=>10, [DimShield(3,20)]=>1, [Bolt]=>4}) (10000 games)
overkill: 105.462
win%: 50.28
loss%:49.72
deckout%: 0.0

rush({[Pillar]=>16, [Creature(5|5)]=>14})
vs.
3dim20-4bolt({[Pillar]=>17, [Creature(5|5)]=>6, [DimShield(3,20)]=>3, [Bolt]=>4}) (10000 games)
overkill: 135.6045
win%: 62.25
loss%:37.75
deckout%: 0.0

rush({[Pillar]=>16, [Creature(5|5)]=>14})
vs.
6dim20-4bolt({[Pillar]=>17, [Creature(5|5)]=>3, [DimShield(3,20)]=>6, [Bolt]=>4}) (10000 games)
overkill: 168.567
win%: 80.49
loss%:16.85
deckout%: 2.66

TL;DR Bolts help out a lot, but dim shields are still a lot better than dusk shields across the board with or without bolts, in small numbers, or in big numbers.
With bolts, 6 Dim shields that cost 20 are better than 3 dusk shields. The results DO suggest that expensive dim shields are less "versatile" - 1 and 3 expensive
dim shields didn't fair so well. I'm not sure if that is a true signal or just a problem with the quanta balance or AI. More investigation is warranted.

I also simulated shield decks vs. a deck with 2 explosions. I noticed that tweaking the AI to start playing dim shields at 75 hp instead of 50 was highly beneficial to the win percentage against explosions, not surprisingly. The results below are with that tweak:
Results for Rush deck + 2 explosions vs. shields (no bolts this time):

Spoiler for Hidden:

blast({[Pillar]=>15, [Creature(5|5)]=>13, [Explosion(3)]=>2})
vs.
control({[Pillar]=>16, [Creature(5|5)]=>14}) (10000 games)
overkill: 92.1715
win%: 40.5
loss%:59.5
deckout%: 0.0

blast({[Pillar]=>15, [Creature(5|5)]=>13, [Explosion(3)]=>2})
vs.
1dusk({[Pillar]=>16, [Creature(5|5)]=>13, [DuskMantel]=>1}) (10000 games)
overkill: 90.43
win%: 37.59
loss%:62.41
deckout%: 0.0

blast({[Pillar]=>15, [Creature(5|5)]=>13, [Explosion(3)]=>2})
vs.
3dusk({[Pillar]=>16, [Creature(5|5)]=>11, [DuskMantel]=>3}) (10000 games)
overkill: 92.3605
win%: 42.28
loss%:57.72
deckout%: 0.0

blast({[Pillar]=>15, [Creature(5|5)]=>13, [Explosion(3)]=>2})
vs.
1dim({[Pillar]=>16, [Creature(5|5)]=>13, [DimShield(3,6)]=>1}) (10000 games)
overkill: 94.0565
win%: 41.12
loss%:58.88
deckout%: 0.0

blast({[Pillar]=>15, [Creature(5|5)]=>13, [Explosion(3)]=>2})
vs.
3dim({[Pillar]=>16, [Creature(5|5)]=>11, [DimShield(3,6)]=>3}) (10000 games)
overkill: 119.437
win%: 55.83
loss%:44.17
deckout%: 0.0

blast({[Pillar]=>15, [Creature(5|5)]=>13, [Explosion(3)]=>2})
vs.
6dim({[Pillar]=>16, [Creature(5|5)]=>8, [DimShield(3,6)]=>6}) (10000 games)
overkill: 207.0755
win%: 78.46
loss%:21.54
deckout%: 0.0

TL;DR Dim shields are still way better than dusk shields vs. explosions. I haven't run tests with expensive dim shields yet, but my intuition is that they will be worse than in other scenarios, since a 3 for 20 quanta trade is bad, mkay.

I highly encourage you to download ruby (if you haven't already) and play around with the code yourself. I would consider the results above somewhat preliminary because I didn't have time to fully optimize the decks. This is significantly more difficult with bolts and explosions in the mix, and it is harder to verify that the decisions made by the AI are acceptable. Future work, for you or me.

[Marsu]

Dim Shield is ridicoulusly strong. The main arguement against a nerf is that it "can be countered".

However, something can be counterable and still OP.

An example:

If SoSac was to suddenly give 20 turns of the "sacrifice" Status instead of 2, with the old HP & Quantum cost, players would complain that it's op. With the same logic that is used to defend Dim Shields, some people would then say: "It's not op, it can be countered. Splash some Water Pendulums and 6 Purifies in your deck and anyone who uses SoSac is a free win." While the second half of that arguement is true, this change of SoSac would influence the meta so strongly that it basically were to become rock-paper-scissors. (SoSac > No Purify > Purify > SoSac > [...])

Thus, the influence on the meta is a criterion to say wether something is OP and needs to be nerfed or not. Dim Shield isn't a card that's very often seen in the highest levels of unupped, unlimited meta. But the only reason not every second deck uses Dim Shield is BECAUSE literally everyone thinks - and has to think - of this card and how to counter it when building a deck. It's way too much rock-paper-scissors as it is at the moment, and it doesn't do the game any good.
Dim Shield needs a nerf.

[Cheesy111]

Spoiler for Hidden:

Dim Shield is ridicoulusly strong. The main arguement against a nerf is that it "can be countered".
However, something can be counterable and still OP.
An example:
If SoSac was to suddenly give 20 turns of the "sacrifice" Status instead of 2, with the old HP & Quantum cost, players would complain that it's op. With the same logic that is used to defend Dim Shields, some people would then say: "It's not op, it can be countered. Splash some Water Pendulums and 6 Purifies in your deck and anyone who uses SoSac is a free win." While the second half of that arguement is true, this change of SoSac would influence the meta so strongly that it basically were to become rock-paper-scissors. (SoSac > No Purify > Purify > SoSac > [...])

That it can be countered may be the most common argument, but it is certainly not the most valid.  Many others have already debunked it - I think it would be more constructive to focus on other lines of argument.

Spoiler for Hidden:

Thus, the influence on the meta is a criterion to say wether something is OP and needs to be nerfed or not.

This is interesting, however you need more substance to back it up than simply saying "since it is not this, it is this".

[Marsu]

That it can be countered may be the most common argument, but it is certainly not the most valid.  Many others have already debunked it - I think it would be more constructive to focus on other lines of argument.

It is an argument I still hear, and not (only) from new players. I don't feel that this topic is "closed" already.

Spoiler for Hidden:

Thus, the influence on the meta is a criterion to say wether something is OP and needs to be nerfed or not.

This is interesting, however you need more substance to back it up than simply saying "since it is not this, it is this".

Actually, saying "since it is not this, it is this" is perfectly fine in this case, as there are only two options.

The necessary criterion is the absolute strength of a card. A 5/5 for x  is stronger than a 4/4 for x  is stronger than a 3/3 for x  .
A 1/1 for 10  can't be OP if there is a 10/10 for 10  .

The sufficient criterion is the relative strength of a card. Remove Deflagration from this game and Dim Shield becomes even more op. Add PC to Elements that don't have any PC yet and Dim Shield becomes weaker.
This interaction/countering of cards, thus the relative strength, (Deflag>Dim) is exactly "influence on the meta".

So, aside from absolute & relative strength, what else could be a criterion?
Nothing, in my opinion.

(The only thing left are the different fields of application of a card, but to my understanding, everything should be balanced for PvP. Additional rules can be implemented for PvE, see SoSac)

[OldTrees]

That it can be countered may be the most common argument, but it is certainly not the most valid.  Many others have already debunked it - I think it would be more constructive to focus on other lines of argument.

It is an argument I still hear, and not (only) from new players. I don't feel that this topic is "closed" already.

Most of the time that argument is actually: "There exist counters so the assertion that 'It is OP because it lacks counters.' is false."