Scribd
Upload
12 views

Stats ch1

Uploaded by

Elias Macher
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
12 views

Stats ch1

Uploaded by

Elias Macher
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 22

Chapter 1: Probability Theory

1.1 Set Theory

Def (1.1.1): The set, S, of all possible outcomes of a particular experiment is called
the sample space for the experiment
Ex: tossing a coin, SAT results, reaction time

Countable if ∃ a 1-1 correspondence with a subset of integers


(finite sets are always countable)
Def (1.1.2): An event is any collection of possible outcomes of an experiment (i.e.
any subset of S ). An event A “occurs” if the outcome of the experiment is in the set.
Ex: prime numbers when rolling a dice
∈ means “belongs to”
Relations: Containment: 𝐴 ⊂ 𝐵 ⟺ 𝑥 ∈ 𝐴 ⟹ 𝑥 ∈ 𝐵 (“A is a subset of B”)
Equality: 𝐴 = 𝐵 ⟺ 𝐴 ⊂ 𝐵 & 𝐵 ⊂ 𝐴 (“A equals B”)

Operations:
Union: set of elements that belong to A or B or both;
𝐴 ∪ 𝐵 = {𝑥: 𝑥 ∈ 𝐴 𝑜𝑟 𝑥 ∈ 𝐵}
Intersection: set of elements that belong to A and B ;
𝐴 ∩ 𝐵 = {𝑥: 𝑥 ∈ 𝐴 & 𝑥 ∈ 𝐵}
Complement of A: set of all elements that are not in A; 𝐴∁ = {𝑥: 𝑥 ∉ 𝐴}
Ex: empty set ∅ is 𝑆 ∁
Thm (1.1.4): For any events 𝐴, 𝐵 and 𝐶 defined on sample space 𝑆,
a. Commutativity: 𝐴 ∪ 𝐵 = 𝐵 ∪ 𝐴 and 𝐴 ∩ 𝐵 = 𝐵 ∩ 𝐴
b. Associativity: 𝐴 ∪ (𝐵 ∪ 𝐶 ) = (𝐴 ∪ 𝐵 ) ∪ 𝐶
and 𝐴 ∩ (𝐵 ∩ 𝐶 ) = (𝐴 ∩ 𝐵) ∩ 𝐶
c. Distributive law: 𝐴 ∩ (𝐵 ∪ 𝐶 ) = (𝐴 ∩ 𝐵) ∪ (𝐴 ∩ 𝐶 )
𝐴 ∪ (𝐵 ∩ 𝐶 ) = (𝐴 ∪ 𝐵 ) ∩ (𝐴 ∪ 𝐶 )
d. DeMorgan’s laws: (𝐴 ∪ 𝐵)∁ = 𝐴∁ ∩ 𝐵∁ and (𝐴 ∩ 𝐵)∁ = 𝐴∁ ∪ 𝐵∁

Def (1.1.5): 𝐴 & 𝐵 are disjoint (or mutually exclusive) if 𝐴 ∩ 𝐵 = ∅


The events A1, A2, … are pairwise disjoint if 𝐴𝑖 ∩ 𝐵𝑗 = ∅ ∀𝑖 ≠ 𝑗

Def (1.1.6): If A1, A2, … are pairwise disjoint and ⋃∞


𝑖=1 𝐴𝑖 = 𝑆, then the collection A1,
A2, … forms a partition of S.
1.2: Probability Theory

Def (1.2.1): A collection of subsets of S is called a Borel field (or sigma algebra)
denoted by 𝓑, if it satisfies the following three properties:
a. ∅ ∈ ℬ (the empty set is an element of 𝓑)
b. If 𝐴 ∈ ℬ, then 𝐴∁ ∈ ℬ (or 𝓑 is closed under complementation)
c. If A1, A2, …∈ ℬ, then ⋃∞𝑖=1 𝐴𝑖 ∈ ℬ ( 𝓑 is closed under countable unions)
thus, 𝑆 ∈ ℬ and ⋂∞ 𝑖=1 𝐴𝑖 ∈ ℬ

Ex: trivial sigma algebra {∅, 𝑆}


Ex (1.2.2): if S countable, ℬ = {𝑎𝑙𝑙 𝑠𝑢𝑏𝑠𝑒𝑡𝑠 𝑜𝑓 𝑆, 𝑖𝑛𝑐𝑙𝑢𝑑𝑖𝑛𝑔 𝑆 𝑖𝑡𝑠𝑒𝑙𝑓 }
→in the case that S is finite, this is the power set
Ex (1.2.3): let 𝑆 = (−∞, ∞) = ℝ (thus S uncountable) ,
ℬ = 𝑠𝑒𝑡 𝑖𝑛𝑐𝑙𝑢𝑑𝑖𝑛𝑔 [𝑎, 𝑏], [𝑎, 𝑏), (𝑎, 𝑏] 𝑎𝑛𝑑 (𝑎, 𝑏) ∀ 𝑎, 𝑏 ∈ ℝ
𝑎𝑛𝑑 𝑎𝑙𝑙 (𝑝𝑜𝑠𝑠𝑖𝑏𝑙𝑦 𝑐𝑜𝑢𝑛𝑡𝑎𝑏𝑙𝑦 𝑖𝑛𝑓𝑖𝑛𝑖𝑡𝑒) 𝑢𝑛𝑖𝑜𝑛𝑠 𝑎𝑛𝑑 𝑖𝑛𝑡𝑒𝑟𝑠𝑒𝑐𝑡𝑖𝑜𝑛𝑠
Def (1.2.4): Given a sample space S and an associated Borel field ℬ, a probability
function is a function P with domain ℬ that satisfies:
a. 𝑃(𝐴) ≥ 0 for all 𝐴 ∈ ℬ
b. P(S)=1
c. If A1, A2,… ∈ ℬ are pairwise disjoint, then 𝑃(⋃∞ 𝐴
𝑖=1 𝑖 ) = ∑ ∞
𝑖=1 𝑃 (𝐴𝑖 )
called the Axiom of Countable additivity, however for us it is enough with
Axiom of Finite additivity: if 𝐴 ∈ ℬ and 𝐵 ∈ ℬ are disjoint, then
𝑃(𝐴 ∪ 𝐵)= 𝑃(𝐴)+𝑃 (𝐵)

Typically, we interpret 𝑃: ℬ → [0,1] as “frequency of occurrence”, but also


“beliefs”, others

Thm (1.2.6): Let 𝑆 = {𝑠1 , … , 𝑠𝑛 } be a finite set. Let ℬ be any sigma algebra of subsets
of 𝑆. Let 𝑝1 , … , 𝑝𝑛 be nonnegative numbers such that ∑𝑛𝑖=1 𝑝𝑖 =1
If for any 𝐴 ∈ ℬ, we define 𝑃(𝐴) = ∑{𝑖: 𝑠𝑖 ∈𝐴} 𝑝𝑖 , then 𝑃 is a probability function; even
if 𝑆 = {𝑠1 , 𝑠2 … } but 𝑆 is countable and ∑∞
𝑖=1 𝑝𝑖 =1.
Ex. risk neutral (RN) probabilities

Replicate with portfolio nS stocks & nB bonds


By no-arbitrage,

we can price any security paying CF in states u & d not knowing the actual
probabilities that the market assign to u & d. Instead, you can see in the numerator
0.6 & 0.4 that are RN probabilities, capturing the risk premium and allowing to
discount future payoffs by the risk free rate (i.e. as if agents were risk neutral)

For instance, if the market risk premium is 0.005, then the actual Pu is 0.65.
Thm (1.2.8): If P is a probability function and A is any set in ℬ, then:
a. 𝑃(∅) = 0
b. 𝑃(𝐴) ≤ 1
c. 𝑃(𝐴∁ ) = 1 − 𝑃(𝐴)

Thm (1.2.9): If P is a probability function and A & B are any sets in ℬ, then
b. 𝑃(𝐴 ∪ 𝐵) = 𝑃(𝐴) + 𝑃(𝐵) − 𝑃(𝐴 ∩ 𝐵)
a. 𝑃(𝐵 ∩ 𝐴∁ ) = 𝑃(𝐵) − 𝑃(𝐴 ∩ 𝐵)
c. If 𝐴 ⊂ 𝐵, then 𝑃(𝐴) ≤ 𝑃(𝐵)

Thm (1.2.10): Bonferroni’s inequality:


𝑃(𝐴 ∩ 𝐵) ≥ 𝑃(𝐴) + 𝑃(𝐵) − 1 =.95 + .95 −1 = .90

Thm (1.2.11): If P is a probability function, then:


a. 𝑃(𝐴) = ∑∞𝑖=1 𝑃 (𝐴 ∩ 𝐶𝑖 ) for any partition 𝐶1 , 𝐶2 , …
b. 𝑃(⋃∞ ∞
𝑖=1 𝐴𝑖 ) ≤ ∑𝑖=1 𝑃 (𝐴𝑖 ) for any sets 𝐴1 , 𝐴2 , … (Boole’s inequality)
When all outcomes in S are equally likely (classical probability), then we can
calculate the probabilities of events by simply counting:

Ex (1.3.1): Instead, two aces out of a hand of two poker cards


1.3 Conditional Probability and Independence

Ex (1.3.1) Revisit, two aces out of a hand of two poker cards

Def (1.3.2) If 𝐴 and 𝐵 are events in 𝑆 and 𝑃 (𝐵) > 0, then the conditional
probability of A given 𝐵 is:
𝑃(𝐴 ∩ 𝐵)
𝑃 (𝐴 | 𝐵 ) =
𝑃(𝐵)

Thm (1.3.5) Bayes’ Rule: Let 𝐴1 , 𝐴2 , … be a partition of the sample space, and let 𝐵
be any set. Then, for each 𝑖 = 1, 2, …
𝑃 (𝐵|𝐴𝑖 )𝑃(𝐴𝑖 )
𝑃 (𝐴 𝑖 | 𝐵 ) =
∑∞
𝑗=1 𝑃(𝐵|𝐴𝑗 )𝑃(𝐴𝑗 )
Def (1.3.7): Two events, 𝐴 & 𝐵 are statistically independent if
𝑃(𝐴 ∩ 𝐵) = 𝑃(𝐴)𝑃(𝐵)

Thm (1.3.9) If 𝐴 & 𝐵 are independent events, then the following pairs are also
independent:
a. 𝐴 and 𝐵∁
b. 𝐴∁ and 𝐵
c. 𝐴∁ and 𝐵∁
Ex (1.3.10)

Def (1.3.12): A collection of events 𝐴1 , , …, 𝐴𝑛 are mutually independent if for any


subcollection 𝐴𝑖1 , …, 𝐴𝑖 𝑘 , we have
𝑘 𝑘
𝑃 (⋂ 𝐴𝑖 𝑗 ) = ∏ 𝑃 (𝐴𝑖𝑗 )
𝑗=1 𝑗=1
1.4 Random variables

Def (1.4.1) A random variable is a function from a sample space 𝑆 into the real
numbers ℝ

Ex (1.4.3): Number of heads when tossing three coins


Ex (1.4.3): Number of heads when tossing three coins

and
1.5 Distribution Functions

Def (1.5.1): The cumulative distribution function (cdf) of a random variable 𝑋


denoted 𝐹𝑋 (𝑥 ) is defined by 𝐹𝑋 (𝑥 ) = 𝑃𝑋 (𝑋 ≤ 𝑥 ), for all 𝑥
Thm (1.5.3); The function 𝐹(𝑥) is a cdf if and only if
a. lim 𝐹(𝑥) = 0 and lim 𝐹(𝑥) = 1
𝑥→−∞ 𝑥→∞
b. 𝐹(𝑥) is a nondecreasing function of 𝑥
c. 𝐹(𝑥) is right-continuous; i.e. ∀ 𝑥0 lim 𝐹(𝑥) = 𝐹(𝑥0 )
𝑥→𝑥0
Def (1.5.7): A random variable 𝑋 is:
a. Discrete: if 𝐹𝑋 (𝑥 ) is a step function of 𝑥 (ex. geometric)
b. Continuous: if 𝐹𝑋 (𝑥 ) is a continuous function of 𝑥 (ex. normal)

Def (1.5.8 & 1.5.10): The random variables 𝑋 & 𝑌 are identically distributed if
𝐹𝑋 (𝑥 ) = 𝐹𝑌 (𝑥 ) ∀𝑥
1.6 Density and Mass functions

Def (1.6.1): The probability mass function (pmf) of a discrete random variable 𝑋
is given by 𝑓𝑋 (𝑥 ) = 𝑃(𝑋 = 𝑥 ) ∀𝑥

Def (1.6.3): The probability density function (pdf) of a continuous random


variable 𝑋, denoted 𝑓𝑋 (𝑥 ), is the function that satisfies:
𝑥
𝐹𝑋 (𝑥 ) = ∫ 𝑓𝑋 (𝑡) 𝑑𝑡 ∀𝑥
−∞

Thm (1.6.5): A function 𝑓𝑋 (𝑥 ) is a pdf (or pmf) of a random variable 𝑋 if and only
if:
a. 𝑓𝑋 (𝑥 ) ≥ 0, ∀𝑥
∞
b. ∑𝑥 𝑓𝑋 (𝑥 ) = 1 (pmf) or ∫−∞ 𝑓𝑋 (𝑥 )𝑑𝑥 = 1 (pdf)
Exponential distribution 𝑋~exp(𝜆) continuous analogue to geometric
𝑋 is the time it takes until an event happens (or time between events)
The event follows a Poisson process THUS they occur:
independently from each other
at a constant rate ("𝜆" occurrences per unit of time)

Given the cdf: 𝐹𝑋 (𝑥 ) = 1 − 𝑒 −𝜆𝑥 ∀ 𝑥 ∈ [0, ∞)


the pdf is 𝑓𝑋 (𝑥 ) = 𝜆𝑒 −𝜆𝑥 ∀ 𝑥 ∈ [0, ∞) although in book
https://en.wikipedia.org/wiki/Exponential_distribution & https://homepage.divms.uiowa.edu/~mbognar/applets/exp-like.html

You might also like