Markov and Mr.

Monopoly Make Millions

Sp.268 Spring 2010

Probability

Probability is key to many fields, a mere few of which are econometrics,

quantum mechanics, signal processing, statistical physics, and theoretical
computer science. This section will be a gental introduction to probability,
more than necessary for how it is applied to Monopoly. We will also look
at how probability can be used to analyze strategies of and find solutions
to some fun games and puzzles. Problems involving probability are either
continuous or discrete, and the two are usually very similar besides some
minor adjustments in notation. The differences will be mentioned when
applicable.

Sets

A set is a collection of items. An example of a set can be all the Course XIV
classes offered at MIT: {14.01, 14.02, 14.04, 14.05, 14.32, 14.33, 14.36 . . . }. For
the following definitions and examples, let A and S be arbitrary sets.

An element of a set is something belonging to that set. We write a ∈ A if

a is a member of the set A, and a ∈
/ A if a is not a member of the set A.

A subset is a set contained within another set, in other words, if all members
of a set belongs to another set. A is a subset of S if all members of A belong
to S, and we write A ⊆ S. Note that:

1
SP.268 Monopoly

• A = S if and only if A ⊆ S and S ⊆ A.

• The empty set ∅, or a set with no elements, is a proper subset of every

set.

• A proper subset is a set that is strictly contained in another set.

That is, A is a proper subset of S if and only if there is at least one
element contained in S that is not contained in A, and all elements of
A are contained in S.

The cardinality of a set, denoted |A| here, is the number of elements in that
set. If A ⊆ S, then |A| ≤ |S|. If A ⊂ S, then |A| < |S|.

Probability and Sets

Now that we have defined sets generally, let’s look at how sets are used when
applied to probability. The ‘things’ or ‘items’ that we’re concerned with are
outcomes–outcomes from flipping coins, dealing hands or cards, etc. The
sample space is the set of all possible outcomes, denoted Ω. A subset of a
sample space consists of the outcomes that we’re interested in, called events.

Suppose we have the events A, B, and C. The interesction of two events

is the event that they both occur. C = A ∩ B if the event C represents both
A and B occuring. If A ∩ S = φ, then the two sets are called disjoint or
mutually exclusive. The union of two events is the event that either one
or the other occurs, denoted C = A ∪ B.

Laws of set operations:

• Commutative:

A∪B =B∪A
A∩B =B∩A

2
SP.268 Monopoly

• Associative:

(A ∪ B) ∪ C = A ∪ (B ∪ C)
(A ∩ B) ∩ C = A ∩ (B ∩ C)

• Distributive:

(A ∪ B) ∩ C = (A ∩ C) ∪ (B ∩ C)
(A ∩ B) ∪ C = (A ∪ C) ∩ (B ∪ C)

The probability of an event corresponds to the map from the set of events to
the interval [0,1]. When we talk about the probability of some event A in Ω,
it will always conform to the following axioms:

1. The probability of the sample space, Ω, is P (Ω) = 1.

2. P (A) ≥ 0.

3. If A1 and A2 are disjoint, then

P (A1 ∪ A2 ) = P (A1 ) + P (A2 ).

More generally, if Ai for i = 1, 2, 3, . . . are disjoint, then

[∞  X ∞
P Ai = P (Ai ).
i=1 i=1

The inclusion-exclusion principle is very useful in calculating probabilities.

It states that for two events, A1 and A2 , not necessarily disjoint as in the
third axiom above,

|A1 ∪ A2 | = |A1 | + |A2 | − |A1 ∩ A2 |.

The third term in in the equation above subtracts the overlap in A1 and A3
that gets counted twice. The probability version of the inclusion-exclusion
principle is
P (A1 ∪ A2 ) = P (A1 ) + P (A2 ) − P (A1 ∩ A2 ).

3
SP.268 Monopoly

The complement of a set A, commonly denoted A0 , Ac , or A, is all elements

in the sample space that don’t belong to that set.
Ac = Ω − A,
|Ac | = |Ω| − |A|,
P (Ac ) = P (Ω) − P (A)
= 1 − P (A).

For most problems, the goal will be to find the likelihood that an event E
happens, or P (E), out of the set of possible outcomes S. When all the
outcomes are equally likely,
|E|
P (E) = .
|S|
We’re adding up all the elements in E and all the elements in S, then dividing
them. This leads us to the topic of counting, which is used when dealing with
discrite, finite sample spaces.

Counting

We make the assumption that all the outcomes are equally likely, also known
as the assumption of uniform probability. All that needs to be done then is
add up the number of outcomes that we care about and divide that by the
number of all possible outcomes. The trickiest part is defining the event and
sample space and making sure that we count everything the right number of
times.

Counting Rules

We’ve seen the Sum Rule already, just not labeled with the name. If
A1 , A2 , . . . , An are disjoint sets, then
|A1 ∪ A2 ∪ . . . ∪ An | = |A1 | + |A2 | + . . . + |An |.
What’s the probability version of the Sum Rule?

4
SP.268 Monopoly

Example 1

Excuse this somewhat lame example, but its purpose is to show the sum rule
at work. In a group of 150 students, 15 use Internet Explorer as their web
browser of choice, 80 use Firefox, 15 use Safari, and 40 use Chrome. If being
“cool” means you use Firefox or Chrome as your main web browser, what is
the probability that we pick one student who is “cool?”

Recall the definition of set intersections, when one event OR the other hap-
pens. Let the set C be the set of “cool” students; there are 80 + 40 students
in C, by the sum rule. Let S be the set of all students; there are 150 students
in total, as stated in the problem. Therefore, the probability of picking a cool
student is:
|C| 120 4
P (picking a cool student) = = = .
|S| 150 5

The multiplication rule states that for a length-k sequence, where the first
term is chosen out of set S1 , the 2nd term is chosen out of set S2 . . . the last
term is chosen out of Sk , then

|Total # of sequences| = |S1 × S2 × · · · × Sk |

= |S1 | · |S2 | · · · |Sk |.

Example 2

The athena combination lock just got changed again. You’re far from any
Quickstation and there’s no one else nearby. Suppose you wanted to try
your luck at guessing the combo (and SIPB’s hint board is not there). As
you try the different combos that pop up in your head, you wonder, how
many possible combination of buttons are there?

The athena door locks have 6 buttons, 3 on the top row and 3 on the bottom
row. The athena passcode is 5 digits. Let Di , for i = 1, 2, 345, represent
the set of buttons possible for each digit; each digit can be any of the first 5

5
SP.268 Monopoly

buttons.

|Total # of combinations| = |D1 × D2 × D3 × D4 × D5 |

= |D1 | · |D2 | · |D3 | · |D4 | · |D5 |
=5·5·5·5·5
= 55

Permutations

The set of permutations on a collection of objects is an example of the

algebraic structure called a ‘group,’ which was covered in the Rubik’s Cube
lecture. Here we’ll use ‘a set of ordered objects’ as a working definition of
permutations. For a collection of n objects, there are n(n−1)(n−2) · · · (1) =
n! different orderings of the objects.

Example 3: The Birthday Problem

You’re in a room with a bunch of people, say n ≤ 365 people.

a) What is the probability that two people in the room have the same
birthday? Ignore complications with leap years and assume there are
365 days in a year. We also assume that birthdays are random (not
exactly true).
This problem is best approached the other way around, because the
probability that no two people have the same birthday would be easier
to find.
Let A be the event that two people have the same birthday. Then Ac
is the event that no two people have the same birthday. Note that
P (A) = 1 − P (Ac ). We start with person 1; this person can have any 1
of 365 days out of the year. A second person can only have a birthday
on the 364 days out of the year that hasn’t been ‘taken.’ By assumption
of random birthdays, and of uniform probability, the chance that this

6
SP.268 Monopoly

person has any of the 364 birthdays is 364365

. A third person can only
have a birthday out of the 353 days not ‘taken,’ and the corresponding
probability of such an event is 363
365
. This continues until we’ve covered
all n people.
365 · 364 · 363 · · · (365 − n + 1)
P (Ac ) =
365n
c
P (A) = 1 − P (A )
365!
=1−
365n · n!
b) What is the probability that someone shares your birthday?
1
Each person can have your birthday with probability 365 . There are
n−1 people other than yourself, so the probability that someone shares
your birthday is n−1
365
.

The answers to part a) and part b) are quite different, but the way the ques-
tions were phrased were only slightly different. Half the work in probability
questions is usually figuring out what the question wants from you...

What happens if n > 365? You can answer part a) without doing any math,
by the Pigeonhole Principle. The Pigeonhole Principle states that in a
mapping from set X to set Y , if |X| > |Y |, then more than one element of
X map to soem element in Y .

Combinations

We will also want to deal with collections that are unordered. How many
ways are there to take r objects out of a set of n objects?

For the first object, we have n to choose from. For the 2nd object, we have
n − 1 to choose from. For the rth object, we have n − r + 1 to choose from.
But note that once we’ve selected r objects this way, they are in some kind
n!
of order, and the answer n(n − 1)(n − 2) · · · (n − r + 1) = (n−r)! is not correct.
We must divide by r!, which is the number of ways you can order (permute)
r objects.

7
SP.268 Monopoly

The number of ways that we can take r objects out of a set of n objects is
therefore  
n! n
= .
r!(n − r)! r

Example 4: Hands of cards

In this example we’re using a standard 52-card deck.

a) How many ways are there to deal a 5-card hand?

 
52 52!
= = 2598960
5 5!(47)!

b) How many ways are there to deal a flush, a 5-card hand with all cards
the same suit?
There are 41 ways to choose the suit, and 13



5
ways to choose the 5
cards out of that suit.
 
13
# of ways to deal a flush = 4 = 5148
5

c) How many ways are there to deal a 5-card hand with 1 pair?
There are 13 4



1
ways to choose the card value of the pair, and 2
ways
50


to choose the suits of the pair; then there are 3 ways to choose the
remaining 3 cards of the 5-card hand.
  
4 50
# of ways to deal a hand with 1 pair = 13 = 1528800
2 3

d) How many ways are there to deal a 5-card hand with only 1 pair?
As
before, there are 13


1
ways to choose the card value of the pair, and
4
2
ways to choose the suits of the pair. But the problem specifies only
1 pair. The remaining 3 cards in the hand cannot contain a pair. So

8
SP.268 Monopoly

there are are 12

3
ways to choose 3 different values besides the value
that’s already a pair, and they can be from any suit.

  
4 12 3
# of ways to deal a hand with only 1 pair = 13 4 = 1098240
2 3

e) How many ways are there to deal a 3-of-a-kind?

There are 13 4



1
ways to choose the card value of the 3-of-a-kind, and 3
ways to choose the suits of the pair; then there are 48


2
ways to choose
the remaining 2 cards of the 5-card hand, making sure that the value of
the 3-of-a-kind doesn’t get chosen (otherwise we’d get a 4-of-a-kind).
  
4 48
# of ways to deal a 3-of-a-kind = 13 = 58656
3 2

f) How many ways are there to deal a full house, a 5-card hand with 3 of
one kind and 2 of another?
13


As before, there are 1
ways to choose the card value of the 3-of-a-

kind, and 3 ways to choose the suits of the pair.
Then there are 12
4


1
ways to choose the value of the 2-of-a-kind and 42 ways to choose the
suit.
   
4 12 4
# of ways to deal a full house = 13 = 3744
3 1 2

Conditional Probability

If we’re interested in the probability that some event A occurs given that
some event B has already occurred, the sample space becomes B. The prob-
ability of A conditioned on B becomes a probability on the space B.

The Multiplication Law states that P (A ∩ B) = P (A|B)P (B).

9
SP.268 Monopoly

So we get that
P (A ∩ B)
P (A|B) = , for P (B) 6= 0.
P (B)

With some rearranging, we get Baye’s Rule, which is commonly seen in

many different forms:

P (B|A)P (A) = P (A|B)P (B).

The Law of Total Probability gives us the ability to isolate the probability
of one event on a partitioned probability space. Given a space Ω that is
partitioned by Bn : n = 1, 2, . . . , and an event A,

n
X
P (A) = P (A ∩ Bi )
i=1

Example 5

Melissa and I are going to assign your P/F grades for this seminar by picking
them out of a hat. We take 100 slips of paper and mark P on half of them,
F on the other half. Then we put the slips of paper in two hats, and pick
a slip of paper from one of the hats. Whatever we pick will be your grade.
But, being as merciful and fair as we so obviously are, and curious how much
you got out of this class, we’ll leave it up to you to place the slips of paper
into the two hats any way you want. How will you do it?

Monopoly

The game of MONOPOLY R came about during the Great Depression, orig-
inating from Charles Darrow of Germantown, Pennsylvania. It started out
as handmade sets sold in a shop in Philadelphia, and as people grew to love

10
SP.268 Monopoly

the game, Darrow approached Parker Brothers to enlarge the production

scale (he’d actually been rebuffed by the Parker Brothers the first time in
1933 due to 52 ‘fundamental playing flaws’). Today, MONOPOLY R is the
best-selling board game in the world, distributed in 103 countries and 37
languages. Some fun facts from the MONOPOLY R website:

• The longest MONOPOLY

R game ever played was 1, 680 hours long.

• The MONOPOLY
R man isn’t a Parker Brother. His name is Mr.
Monopoly.

• Parker Brothers once sent an armored car with a million MONOPOLY R dollars
to Pittsburgh because a marathon game there had run out of money.

• MONOPOLY
R comes in a Braille version.

• The four most-landed-on squares are Jail, Illinois Avenue, “Go”, and
the B&O Railroad.

The last in the list of fun facts above is more than meets the eye. What
makes certain game squares more likely to be landed-on than others? Illinois
Avenue doesn’t seem to be special compared to other properties. . . It turns
out that we can model the MONOPOLY R game board to calculate the exact
probabilities of landing on a certain square.

Rules

The objective of the game is to bankrupt all opponents, though most games
played with family and friends end when it is apparent that someone will
win. A typical game of MONOPOLY R uses the following items:

• 1 game board

• 2 dice

• token for players (11 official MONOPOLY

R ones)

11
SP.268 Monopoly

• 32 houses

• 12 hotels

• 16 Chance cards

• 16 Community Chest cards

• property deeds for each of the 22 MONOPOLY

R properties

• $15140 in MONOPOLY
R money

The Chance and Community Chest cards are placed face down on the game
board, and a player must pick one of the cards when he lands on the Chance
or the Community Chest game squares. Each player is given $1500 to be-
gin the game. All remaining money, game piece, houses, hotels, and deeds
of unsold property go to the Bank. The Bank collects all taxes, fines,
loans, and interest. The Bank never goes ‘broke.’ If the Bank runs out
of MONOPOLY R money, then more can be issued (see fun fact above).

Players begin on the Go square, roll two dice, and advance as many steps as
dots displayed on the the two dice. A player can buy any property, utility,
or railroad that isn’t already owned by another player, or must to draw
Chance/Community Chest cards, pay rent, fines, or go to Jail as dictated
by the square he lands on. If a player throws a double, then he moves his
token the number of steps, is subject to whatever privileges or penalties of
the square he lands on, and then tosses the dice again. If a player tosses
three doubles in a single turn, he must go to Jail.

Landing on the Jail square is just ‘visiting Jail’, while landing on the ‘Go to
Jail’ square, drawing a ‘Go to Jail’ card, and tossing doubles 3 times during
a turn are actual Jail sentences. Any Jail term lasts 3 turns. A player tosses
dice at each turn, and if he tosses a double, then he is free to get out of jail
and advances the number of steps as his double shows. That player does not
take another turn. A player gets out of Jail if he has a ‘Get out of Jail Free’
card, or if another player is willing to sell him a ‘Get out of Jail Free’ card
at a negotiated price, or if the player pays a $50 fine.

12
SP.268 Monopoly

If a player lands on a property owned by another player, then the owner

collects rent based on the information on the property deed. Rents are much
higher for properties with houses or hotels. When a player owns all the
MONOPOLY R properties in a color group, then he has the option to build
houses on those properties. If he buys one house, he can put them on any
one of those properties. The next house he buys must be erected on one of
the unimproved properties of that or any other complete color group, and so
on. Thus, players must build evenly across all his properties in a color group.

More details of the rules of the game will unfurl as we analyze the game.
First, a bit of linear algebra.

Matrices, Eigenvalues, and Eigenvectors

The linear equation Ax = b is at the heart of most introductory linear

algebra courses. A is a matrix, and x and b are vectors; the matrix A
‘operates’ on x to give b; x and b lie on the same vector space but are in
different directions unless A is the identity matrix I.

Eigenvectors are special vectors associated with every operating matrix.

These vectors don’t change directions when multiplied by the matrix, and
we get the equation Ax = λx. Each eigenvector has its own eigenvalue
λ. Most 2 × 2 matrices have two eigenvectors and their two corresponding
eigenvalues.

What happens when A operates on x more than once? As in, what’s A2 x?

A3 x? A100 x?

The number λ is an eigenvalue of A if and only if A−λI (which is a matrix) is

singular, or det (A − λI) = 0. A singular matrix is a square matrix that has
no inverse, or det A = 0. Then, for each eigenvalue, we solve (A − λI)x = 0,
or Ax = λx to find the eigenvector x.

13
SP.268 Monopoly

Example 6

1 2
Find the eigenvalues and eigenvectors of A = 2 4 .

 1−λ 2

A − λI = 2 4−λ

Take the determinant of this matrix.

2
det 1−λ 2
 
2 4−λ = (1 − λ)(4 − λ) − (2)(2) = λ − 5λ.

Set the determinant to 0, and solve for λ.

λ2 − 5λ = 0 gives λ1 = 0 and λ2 = 5.

Solve (A − λI)x = 0 separately for λ1 = 0 and λ2 = 5.

 1 2  x1  0  2

(A − 0I) = x2 = gives eigenvector −1
 12 24  x1   00  1
(A − 5I) = 2 4 x2 = 0 gives eigenvector 2

As a side note, because the vectors that make up A are constant multiples of
each other, we know that A itself is a singular matrix. The determinant of
a matrix can be found by taking the product of all its eigenvalues, so if the
determinant is zero, then we know one of the eigenvalues must be zero.

Markov Chains

Markov chains are the probabilistic versions of deterministic finite automata.

For our analysis of MONOPOLY R , we’ll consider each of the 40 game
squares to be a state H. At each time step n, the probabilistic state dis-
tribution Xn will be a 40 × 1 vector, with each element representing the

14
SP.268 Monopoly

probability that a player ends his turn in that state. Our state, H, belongs
to the set S of the state space of size 40. The Markov chain is described
in terms of its transition probabilities pij , which is the probability that
we’ll go from state i to state j at a time step. The transition probabilities
sum to 1.
pij = P (Hn+1 = j|Hn = i), i, j ∈ S
X
pij = 1
j

The probability that we’re in a certain state at time step n depends only
on our state at time step n − 1, and is independent of all states besides the
previous state:

P (Hn+1 = j|Hn = i, Hn−1 = i − 1, . . . , Ho = io ) = P (Hn+1 = j|Hn = i) = pij

The transition matrix captures all the transition probabilities and operates
on our state distribution vector. Such a matrix is called a Markov matrix,
and it is also a square matrix.
 
p00 p01 . . . p0m
 p10 p11 . . . p1m 
 
 .. .. . . .. 
 . . . . 
pm0 pm1 . . . pmm

Special matrices will have special eigenvalues and eigenvectors, and for Markov
matrices, all entries are positive and every row adds to 1. Can you see why
they must add to 1? The largest eigenvalue is 1, and the corresponding
eigenvector is the state that comes out at the end. The eigenvectors of other
eigenvalues fall to 0 over time.

Example 7

The ESG elevator has two states: FAIL and MEGAFAIL (we pretend that
those dark days of a completely broken elevator never happened). If it is
FAIL today, then the probability that it becomes MEGAFAIL tomorrow

15
SP.268 Monopoly

is 0.6, and the probability that it stays FAIL is 0.4. If it is MEGAFAIL,

then the probability that MIT Facilities comes to try to fix it (and thus
make it just FAIL) is 0.2, but most likely, with probability 0.8, it’ll just stay
MEGAFAIL. If we start the school year in the fall with a FAIL elevator,
what kind of elevator will we have at the end of the school year?
 x1 
Our initial state distribution is x2 .
 .6 
Our transition matrix is A = .4 .2 .8 . It has eigenvalues 1 and .2, but over
time, the eigenvector associated with .2 will go to 0. We consider the eigen-
vector with eigenvalue 1.

 x1 
(A − I)x = 0 gives us eigenvector x2 wherex1 = x2

So, surprisingly, although it seems more likely that we would end up with a
MEGAFAIL elevator, the odds are actually even between FAIL and MEGAFAIL.

MONOPOLY
R

Ian Stewart, a math professor at the University of Warwick, wrote a column

in the April, 1996 issue of Scientific American seeking to answer the question:
‘Is Monopoly fair?’ In other words, is every MONOPOLY R square equally
likely to be occupied? His initial analysis was only a mathematical exercise,
and his model abstracted many of the realistic playing rules.

Initial Analysis

We abstract away the rules about rolling doubles, Chance/Community Chest

squares, and the complications involving going to Jail. Then each of our 40
states are equally likely, and on each roll of our dice the number of steps we
could possibly take (sum of rolling two dice) is distributed as follows:

16
SP.268 Monopoly

Number on the two dice Probability

6
7 36
5
6, 8 36
4
5, 9 36
3
4, 10 36
21
3, 11 36
1
2, 12 36

The initial state probability vector is a 40-dimensional-vector with 1 as the

0th element and 0 everywhere else. If we index the board from 0 through
39, beginning at ‘Go’, then on the first toss of the dice, our state probability
vector H1 would be:
1 2 3 4 5 6 5 4 3 2 1
[0, 0, , , , , , , , , , , , 0, 0, 0, . . . , 0]0 .
36 36 36 36 36 36 36 36 36 36 36 | {z }
27 zeros

The 1st row of our 40 × 40 Markov transition matrix M would be H10 . The
second row of the matrix would be this vector with 3 zeros before the begin-
ning of the fractions and 27 zeros after, in essence, the same vector ‘shifted’
over by 1. The third row would be shifted over again, so on until we’ve
completed all 40 rows of the matrix. We won’t enumerate all 1600 entries.

To get our state probability vector at time step n, we would just calculate
An Hn . For sufficiently large n, our state probability vector would be the
eigenvector of A that has eigenvalue 1. Our distribution slowly approaches
1
one with all entries equal to 40

The Real Game

We made a few simplifying assumptions in the last section: tossing doubles,

Jail, and Chance/Community Chest cards.

To deal with the rule about tossing doubles, we can modify X1 to have 35
nonzero elements. The maximum number of spaces a player can move is
35, (if a player rolls a {(6,6),(6,6),(6,5)}). Note that the ‘Go to Jail’ square

17
SP.268 Monopoly

has probability 0 because our state distribution is “the probability that the
player ends a turn on that square”. If a player lands on ‘Go to Jail’, then he
would go immediately to Jail and end his turn there.

Jail is a little more complicated. The player has several options when he
goes to Jail: (1) stay there for three turns, (2) roll doubles and get out of
Jail early, (3) pay $50 fine and get out of Jail early, and (4) use a ‘Get out
Jail Free’ card. For a single transition matrix, we can only take one of the
cases. Then we can switch transition matrices depending on what part of
the game we’re in. Early on in the game, it is beneficial to move around as
much as possible to buy up property; later in the game, ‘Jail’ is like ‘Free
Parking’–you don’t have to pay sky-high rents for a bunch of turns while
other players duke it out. If we’re dealing with the case where the player
chooses to stay in Jail, then we’ll have to break up the ‘Jail’ state into three
separate states: ‘Entered Jail on Last Turn’, ‘Entered Jail Two Turns Ago’,
‘Will Leave Jail on Next Turn’. On any of those states, it is possible to roll a
double and transition to another game square. The transition matrix should
reflect this.

The Chance/Community Chest cards are actually not as complicated as

they seem. There are 16 Chance cards, 10 of which tell the player to
move to another square. The probability of staying in Chance is thus 81
the probability it had before, and each of the 10 destinations is increased
1
by 10 × P (probability of landing on Chance). The same goes for Commu-
nity Chest, which only has 2 cards that send players to other squares. The
probabilities would be adjusted accordingly.

As a consequence of Jail, tossing doubles, and Chance/Community Chest

cards sending players to different squares, the probability distribution is no
longer uniformly distributed over all 40 squares. Instead, it is skewed toward
certain squares. Players are almost twice as likely to be in Jail than in any
other square; the next-most-frequented square is Illinois Avenue, and GO
is the third most likely square. B&O Railroad is the most-often occupied
railroad.

18
SP.268 Monopoly

Where and When to Build?

Rent-collecting is when things actually start to get interesting. After all, the
whole point of the game is to bankrupt the other players. What strategy
should we take in building houses and hotels? What can we use from our
probabilistic analysis? If we take the actual decimal values of the probabil-
ities and analyze the time of the break-event point (total cost of buildings
divided by expected earnings from property per turn; how would you calcu-
late the expected earnings?) which is when rents collected becomes greater
than the cost of building the houses and hotels, we find that with 2 houses
or fewer, it typically takes 20 moves or more to break even. With 3 houses,
the chances are significantly better. It is even better than building 4 houses
or a hotel. This is preferable strategy because one of the principle strategies
of MONOPOLY R is to deplete accounts of other players fast while accumu-
lating fast yourself (so that you can purchase more property and build more
buildings). If the break-even point takes too long, then we are wasting valu-
able resources that could have been allocated to buildings on other properties
and raising the rents of those properties.

Remarks

Other similar board games can be modeled in the same way. The premesis of
Markov chains is that the next state is independent of all previous states–it
ony depends on the current state.

19