Discrete Probability Distributions

Random Variables and Probability Distributions

  • Random Variable (XX): A function assigning numerical values to outcomes of an experiment to capture uncertainty.

  • Discrete Random Variable: Assumes a countable number of distinct values (e.g., number of employees).

  • Continuous Random Variable: Takes uncountable values across an interval (e.g., return on a mutual fund).

  • Discrete Distribution Requirements:

    • 0≤P(X=x)≤10 \le P(X = x) \le 1

    • ∑P(X=x)=1\sum P(X = x) = 1

  • Cumulative Distribution Function (CDF): Measures the probability that the random variable is less than or equal to a given value: P(X≤x)P(X \le x).

Summary Measures for Discrete Random Variables

  • Expected Value (Mean, μ\mu): Weighted average measuring central location:   E(X)=μ=∑xiP(X=xi)E(X) = \mu = \sum x_i P(X = x_i)

  • Variance (σ2\sigma^2): Weighted measure of dispersion about the mean:   Var(X)=σ2=∑(xi−μ)2P(X=xi)\text{Var}(X) = \sigma^2 = \sum (x_i - \mu)^2 P(X = x_i)

  • Standard Deviation (σ\sigma): Measure of risk or spread:   SD(X)=σ=Var(X)\text{SD}(X) = \sigma = \sqrt{\text{Var}(X)}

  • Consumer Risk Attitudes:

    • Risk-Averse: Demands positive expected gain as compensation for risk; may decline prospects with positive expected gain.

    • Risk-Neutral: Ignores risk completely; evaluates opportunities strictly on positive expected gain.

    • Risk-Loving: Accepts risky prospects even when expected gain is negative.

Portfolio Returns and Linear Combinations

  • Properties of Linear Combinations:   E(aX+bY)=aE(X)+bE(Y)E(aX + bY) = aE(X) + bE(Y)   Var(aX+bY)=a2Var(X)+b2Var(Y)+2abCov(X,Y)\text{Var}(aX + bY) = a^2\text{Var}(X) + b^2\text{Var}(Y) + 2ab\text{Cov}(X, Y)

  • Portfolio Expected Return:   E(Rp)=wAE(RA)+wBE(RB)E(R_p) = w_A E(R_A) + w_B E(R_B)   where wAw_A and wBw_B represent the proportion of wealth in assets AA and BB.

  • Portfolio Variance and Correlation:   Var(Rp)=wA2σA2+wB2σB2+2wAwBCov(RA,RB)\text{Var}(R_p) = w_A^2 \sigma_A^2 + w_B^2 \sigma_B^2 + 2w_A w_B \text{Cov}(R_A, R_B)   ρAB=σABσAσB\rho_{AB} = \frac{\sigma_{AB}}{\sigma_A \sigma_B}

  • Diversification: Weak or low correlation between asset returns reduces overall portfolio standard deviation (risk) below the individual asset risk levels.

The Binomial Distribution

  • Bernoulli Process Conditions:

    • Fixed number of trials (nn).

    • Identical and independent trials.

    • Exactly two outcomes per trial: success (pp) or failure (1−p1 - p).

    • Probability of success (pp) remains constant across trials.

Probability tree diagram illustrating success and failure sequences across three trials
  • Probability Mass Function (PMF):   P(X=x)=(nx)px(1−p)n−x=n!x!(n−x)!px(1−p)n−xP(X = x) = \binom{n}{x} p^x (1 - p)^{n - x} = \frac{n!}{x!(n - x)!} p^x (1 - p)^{n - x}   for x=0,1,2,…,nx = 0, 1, 2, \dots, n.

  • Summary Measures:   E(X)=μ=npE(X) = \mu = np   Var(X)=σ2=np(1−p)\text{Var}(X) = \sigma^2 = np(1 - p)   SD(X)=σ=np(1−p)\text{SD}(X) = \sigma = \sqrt{np(1 - p)}

The Poisson Distribution

  • Poisson Process Conditions:

    • Successes are counted over a specified interval of time or space (x=0,1,2,…x = 0, 1, 2, \dots).

    • Counts in non-overlapping intervals are mutually independent.

    • Probability of success is equal for intervals of equal size and proportional to interval length.

  • Probability Mass Function (PMF):   P(X=x)=e−μμxx!P(X = x) = \frac{e^{-\mu} \mu^x}{x!}   where μ\mu is the mean number of occurrences in the interval and e≈2.71828e \approx 2.71828.

  • Summary Measures:   E(X)=μE(X) = \mu   Var(X)=σ2=μ\text{Var}(X) = \sigma^2 = \mu   SD(X)=σ=μ\text{SD}(X) = \sigma = \sqrt{\mu}

The Hypergeometric Distribution

  • Conditions:

    • Sampling without replacement from a finite population of size NN.

    • Population contains SS successes and N−SN - S failures.

    • Trials are dependent; probability of success changes on each consecutive draw.

  • Probability Mass Function (PMF):   P(X=x)=(Sx)(N−Sn−x)(Nn)P(X = x) = \frac{\binom{S}{x} \binom{N - S}{n - x}}{\binom{N}{n}}   for sample size nn and integer successes xx.

  • Summary Measures:   E(X)=μ=n(SN)E(X) = \mu = n \left(\frac{S}{N}\right)   Var(X)=σ2=n(SN)(1−SN)(N−nN−1)\text{Var}(X) = \sigma^2 = n \left(\frac{S}{N}\right) \left(1 - \frac{S}{N}\right) \left(\frac{N - n}{N - 1}\right)   SD(X)=σ=Var(X)\text{SD}(X) = \sigma = \sqrt{\text{Var}(X)}

Statistical Software Functions

  • Binomial Distribution:

    • P(X=x)P(X = x): Excel =BINOM.DIST(x, n, p, 0); R dbinom(x, n, p)

    • P(X≤x)P(X \le x): Excel =BINOM.DIST(x, n, p, 1); R pbinom(x, n, p)

  • Poisson Distribution:

    • P(X=x)P(X = x): Excel =POISSON.DIST(x, μ, 0); R dpois(x, μ)

    • P(X≤x)P(X \le x): Excel =POISSON.DIST(x, μ, 1); R ppois(x, μ)

  • Hypergeometric Distribution:

    • P(X=x)P(X = x): Excel =HYPGEOM.DIST(x, n, S, N, 0); R dhyper(x, S, N - S, n)

    • P(X≤x)P(X \le x): Excel =HYPGEOM.DIST(x, n, S, N, 1); R phyper(x, S, N - S, n)