1 of 8

Amiibo Probability 掉寶保底

Drawn by Simple

2 of 8

3 of 8

Scan

20% GreatHit

80% BigHit

Scan

20% GreatHit

80% BigHit

Scan

20% GreatHit

80% BigHit

Scan

20% GreatHit

80% BigHit

Scan

20% GreatHit

80% BigHit

Scan

100% GreatHit

Reset counter

4 of 8

Scan

20%, reload if bad bow

80%, save if ore

Scan

20%, reload if bad bow

80%, save if ore

Scan

20%, reload if bad bow

80%, save if ore

Scan

20%, reload if bad bow

80%, save if ore

Scan

20%, reload if bad bow

80%, save if ore

Scan

reload if bad bow

Get Twilight Bow

Example: How to get

Twilight Bow 黃昏之光弓

5 of 8

Scan

20%, reload if sword/cloth

80%, save if arrow

Scan

20%, reload if sword/cloth

80%, save if arrow

Scan

20%, reload if sword/cloth

80%, save if arrow

Scan

20%, reload if sword/cloth

80%, save if arrow

Scan

20%, reload if sword/cloth

80%, save if arrow

Scan

reload if sword/cloth

Get Biggoron's Sword

Example: How to get

Biggoron's Sword 大鼓隆之劍

6 of 8

def MarkovScan( ProvableBigHit, DesiredTreasure ):

PBH = ProvableBigHit/100 * 4/5 # secure a big hit

DT = DesiredTreasure/100 * 1/5 # obtain the desired item

IC = 1 - PBH - DT # inconclusive

M = matrix(QQ, [ # transition matrix of the markov chain

[ IC,PBH, 0, 0, 0, 0, DT ], # initially no big hit

[ 0, IC,PBH, 0, 0, 0, DT ], # accumulate one big hit

[ 0, 0, IC,PBH, 0, 0, DT ], # accumulate two big hit

[ 0, 0, 0, IC,PBH, 0, DT ], # accumulate three big hit

[ 0, 0, 0, 0, IC, PBH, DT ], # accumulate four big hit

[ 0, 0, 0, 0, 0,1-DT*5,DT*5 ], # accumulate five big

[ 1, 0, 0, 0, 0, 0, 0 ] # dummy state for treasure

])

# Solve for the invarnat state

# = eigenvector of "lambda=1"

# = solve xM = 1x

# = solve x(M - I) = 0

M_I = (M - matrix.identity(7)).left_kernel()

InvState = M_I.basis()[0]

InvState /= sum(InvState)

InvState = vector(RR, InvState)

# print(InvState) # for curious cats

# Recurrent time = 1 / long-term probability

# -1 because we added a dummy state

ExpSteps = 1 / InvState[6] - 1

return ExpSteps

7 of 8

ProvableBigHit is the probability that, from the BigHit drop table,�we get an item that only appears in the this drop table.

DesiredTreasure is the probability that, from the GreatHit drop table,�we get an item that we want.

For example, assume that�the BigHit drop table is {itemA: a%, itemB: b%} (where a + b = 100),�the GreatHit drop table is {itemB: c%, itemD: d%}, (where c + d = 100),�and that itemD is what we want,�then ProvableBigHit = a and DesiredTreasure = d.

8 of 8

For Twilight Bow, the BigHit-exclusive items are Amber 5.00%, Opal 30.00%, Topaz 8.00%, Sapphire 8.00%, Ruby 8.00%, Diamond 2.00%, Star Fragment 2.00%, Luminous Stone 5.00%; sum = 68 = ProvableBigHit.

The Twilight Bow has chance 10% in the GreatHit drop table,�so 10 = DesiredTreasure.

MarkovScan(68, 10) gives 17.6,�meaning that every 17.6 scans gives you one Twilight Bow.