Portfolio Expected Value and Risk

Expected Returns

  • The discussion revolves around calculating expected returns and risks associated with investments in stocks X and Y, both individually and as a portfolio.

  • Initial data includes potential losses and gains for stock X (−25,000-25,000, 50,00050,000, 100,000100,000) and stock Y (−200,000-200,000, 60,00060,000, 350,000350,000).

  • Probabilities are assigned to each of these return values, forming a joint probability distribution.

Portfolio Definition

  • A portfolio (P) is defined as a linear combination of stocks X and Y,:
    P=wX+(1−w)YP = wX + (1-w)Y
    where ww represents the weight or proportion invested in stock X, and (1−w)(1-w) is the proportion invested in stock Y.

  • This is analogous to constants A and B from a prior lecture.

Risk Assessment

  • Risk is quantified using standard deviation.
  • The goal is to calculate standard deviation for X, Y, and the portfolio.

Formulas

  • The presenter refers to an uploaded example with formulas for expected value, variance, covariance, and correlation.

  • The expected value of X is calculated using the formula:
    E[X]=∑x<em>i∗P(x</em>i)E[X] = \sum x<em>i * P(x</em>i), where P(x<em>i)P(x<em>i) is the marginal probability of x</em>ix</em>i.

  • The variance of X is calculated using: Var(X)=E[X2]−E[X]2Var(X) = E[X^2] - E[X]^2

Simplified Table

  • The table is simplified with marginal probabilities for X and Y.
  • Joint probabilities are considered, noting that some probabilities are zero, simplifying calculations.

Calculation Examples

  • Expected value of X is calculated as:
    E[X]=(−25,000∗0.2)+(50,000∗0.3)+(100,000∗0.5)=50,000E[X] = (-25,000 * 0.2) + (50,000 * 0.3) + (100,000 * 0.5) = 50,000

  • Expected value of Y is calculated similarly.

  • Variance is calculated using the formula, then standard deviation (risk) is derived.

Risk vs. Return

  • Stock Y has a higher expected return (95,00095,000) but also higher risk (148.32148.32).
  • Stock X has a lower return (50,00050,000) and lower risk (43,01243,012).
  • Investing in a portfolio might offer a balance between risk and return.

Portfolio Expected Value

  • The expected value of the portfolio is a weighted average of the expected returns of X and Y:
    E[P]=w∗E[X]+(1−w)∗E[Y]E[P] = w * E[X] + (1-w) * E[Y].

Portfolio Variance

  • The formula for the variance of the portfolio is:
    Var(P)=w2∗Var(X)+(1−w)2∗Var(Y)+2∗w∗(1−w)∗Cov(X,Y)Var(P) = w^2 * Var(X) + (1-w)^2 * Var(Y) + 2 * w * (1-w) * Cov(X,Y).

Covariance Calculation

  • Covariance between X and Y is calculated using the formula:
    Cov(X,Y)=∑∑(x<em>i−E[X])∗(y</em>i−E[Y])∗P(x<em>i,y</em>i)Cov(X,Y) = \sum \sum (x<em>i - E[X]) * (y</em>i - E[Y]) * P(x<em>i, y</em>i).

  • Due to zeros in the joint probability table, the calculation simplifies.

Portfolio Example

  • If 40% is invested in X and 60% in Y, the expected return of the portfolio is 77,00077,000.
  • The risk is also somewhere in the middle, potentially making the portfolio an attractive option.

General Concepts

  • Expected values and standard deviations are used to measure predictions and volatility.
  • Confidence intervals can be estimated using expected value and standard deviation, helping assess risk.

Visualization

  • Investment returns can be visualized over time, with a predicted value and confidence intervals representing risk.

Chapter 6 Introduction

  • Chapter 6 will cover continuous variables.
  • Key distributions: Normal, Exponential and Uniform.

Normal Distribution

  • The normal distribution (bell-shaped curve) is common in nature, e.g., weight distribution.
  • It is foundational for many statistical techniques, including Ordinary Least Squares (OLS) regression.

Other Distributions

  • Logistic Regression: Used for binary classification problems.
  • Poisson Regression: Used to model count data (e.g., number of phone calls).
  • Exponential Distribution: Used to model lifetimes or time duration (e.g., length of phone calls, battery life).
  • Uniform Distribution: Primarily used for simulation purposes, can be transformed to generate data for other distributions.

Simulation

  • Simulation involves generating data based on a theory to make predictions.
  • It can be used for weather forecasting, where different scenarios are simulated to estimate future conditions.