BUSINESS STATISTICS • PROBABILITY MODELS

Probability & Expected Value — Probability Rules and Expected Value in Decision-Making

Quantifying uncertainty and weighing outcomes to make rational business decisions under risk.

Historical Context & Motivation

The formal study of probability emerged not from ivory-tower mathematics but from intensely practical questions about gambling, commerce, and insurance. Merchants in Renaissance Italy needed to price maritime insurance contracts—essentially placing bets on whether a ship would survive its voyage—and royal courts across Europe sought rational methods to divide stakes in interrupted games of chance. These early efforts to quantify uncertainty laid the groundwork for modern decision science, risk management, and the expected-value calculations that permeate every MBA curriculum today. Understanding this history clarifies why probability is not merely an abstract branch of mathematics: it is, at its core, a framework for making better decisions when outcomes are uncertain.

1654
The Pascal–Fermat Correspondence
Blaise Pascal and Pierre de Fermat exchanged letters on the problem of points—how to divide the stakes of an interrupted game fairly—establishing the combinatorial foundations of probability theory.
1713
Bernoulli's Ars Conjectandi
Jacob Bernoulli published Ars Conjectandi, which formalized the law of large numbers and extended probability beyond games of chance into commerce, law, and demography.
1738
Expected Utility & the St. Petersburg Paradox
Daniel Bernoulli proposed that decision-makers evaluate outcomes by expected utility rather than raw expected monetary value, resolving the famous St. Petersburg Paradox and anticipating modern behavioral economics.
1933
Kolmogorov's Axioms
Andrey Kolmogorov published Foundations of the Theory of Probability, providing the rigorous axiomatic framework that unified all probability rules into a coherent mathematical system still used today.
1944
Game Theory & Decision Science
John von Neumann and Oskar Morgenstern published Theory of Games and Economic Behavior, embedding expected value into strategic decision-making under uncertainty—a cornerstone of modern business strategy and finance.

The central question these thinkers addressed remains the same question every business leader faces: When outcomes are uncertain, how should we weigh possible gains against possible losses to choose the best course of action? This lesson equips you with the probability rules that structure such analysis and the expected-value machinery that converts those rules into actionable decision criteria.

Core Principles & Definitions

Before diving into formulas, it is essential to establish the vocabulary and foundational rules that govern all probability calculations. Every probability model rests on Kolmogorov's three axioms: probabilities are non-negative, the probability of the entire sample space equals one, and the probability of the union of mutually exclusive events equals the sum of their individual probabilities. From these axioms flow the addition rule, the multiplication rule, the complement rule, and the concept of conditional probability—tools that, when combined with expected-value analysis, provide a complete decision-making toolkit.

1

Sample Space & Events

The sample space (S) is the set of all possible outcomes. An event is any subset of S. For example, when launching a product, S might be {success, moderate, failure} and the event 'profitable' might be {success, moderate}.
2

Addition Rule (OR)

P(A ∪ B) = P(A) + P(B) − P(A ∩ B). When events are mutually exclusive, P(A ∩ B) = 0, so the formula simplifies to P(A) + P(B). This rule prevents double-counting overlapping outcomes.
3

Multiplication Rule (AND)

P(A ∩ B) = P(A) × P(B | A). When A and B are independent, P(B | A) = P(B), so P(A ∩ B) = P(A) × P(B). Independence means one event's occurrence does not affect the other's probability.
4

Conditional Probability

P(A | B) = P(A ∩ B) / P(B). Conditional probability updates your belief about event A given that event B has occurred—critical when new market data arrives and you must revise forecasts.
5

Expected Value

The expected value E(X) = Σ xᵢ × P(xᵢ) is the probability-weighted average of all possible payoffs. It represents the long-run average outcome if a decision were repeated many times under identical conditions.
KEY TAKEAWAY
Think of probability rules as the grammar of uncertainty. Just as grammatical rules let you construct any sentence from a finite set of parts of speech, the addition rule, multiplication rule, and complement rule let you compute the probability of any complex business event from simpler building blocks. Expected value then acts as the translator, converting those probabilities into a single dollar figure that makes alternatives directly comparable.

Visual Explanation — Probability Rules in Action

The Venn diagram shows two overlapping events A and B within a sample space S. The violet circle represents Event A and the cyan circle represents Event B. The overlap (A ∩ B) must be subtracted once when computing P(A ∪ B) to avoid double-counting customers who buy both products.

The diagram above captures the most common source of error in applied probability: double-counting the intersection. Suppose a retail analyst reports that 40% of customers purchased Product X (Event A), 30% purchased Product Y (Event B), and 10% purchased both. A naïve sum of 40% + 30% = 70% overstates the proportion who purchased at least one product because the 10% overlap is counted twice. The correct calculation is P(A ∪ B) = 0.40 + 0.30 − 0.10 = 0.60, or 60%. This principle scales directly: when evaluating the probability that at least one risk event occurs in a portfolio or supply chain, the addition rule ensures accurate aggregation.

Mathematical Framework

This section formalizes the rules introduced in Section 2 and derives the expected value formula that serves as the decision criterion in quantitative business analysis. Mastering these equations is essential because every capital budgeting model, risk assessment, and pricing strategy ultimately relies on probability-weighted payoffs.

COMPLEMENT RULE
P(A') = 1 − P(A)
P(A') is the probability that event A does not occur. This is often the fastest path to a solution: if it is hard to compute P(at least one), compute P(none) and subtract from 1.
GENERAL ADDITION RULE
P(A ∪ B) = P(A) + P(B) − P(A ∩ B)
For mutually exclusive events (P(A ∩ B) = 0): P(A ∪ B) = P(A) + P(B). Use the general form whenever events can co-occur—such as a customer defaulting on multiple loan products simultaneously.
GENERAL MULTIPLICATION RULE
P(A ∩ B) = P(A) × P(B | A)
For independent events: P(A ∩ B) = P(A) × P(B). Independence means knowing A occurred provides no information about B—e.g., two factories on separate power grids experiencing outages.
EXPECTED VALUE
E(X) = Σ xᵢ × P(xᵢ) = x₁P(x₁) + x₂P(x₂) + … + xₙP(xₙ)
Where xᵢ represents the payoff (or cost) of outcome i and P(xᵢ) is its probability. The sum runs over all n possible outcomes. E(X) is the long-run average payoff and serves as the rational choice criterion: among mutually exclusive alternatives, choose the one with the highest E(X), assuming risk-neutrality.
When Expected Value Is Not Enough
Expected value assumes risk-neutrality—that a decision-maker cares only about long-run averages. In practice, managers are often risk-averse: they would not bet the company on a coin flip even if the expected payoff is positive. That is why real-world analysis supplements E(X) with measures of dispersion (variance, standard deviation) and downside-risk metrics. Nonetheless, expected value remains the foundational benchmark from which all such refinements depart.

Detailed Breakdown — Decision Trees & Expected Payoffs

A decision tree is the primary visual tool for applying expected value in multi-stage business decisions. It maps out decision nodes (squares), chance nodes (circles), and terminal payoffs at the end of each branch. By working backward from the terminal nodes—a technique called folding back the tree—a manager computes the expected value at each chance node and selects the branch with the highest expected payoff at each decision node. The diagram below illustrates a firm choosing between launching a new product or investing the capital in a risk-free bond.

The decision tree shows a firm choosing between launching a product (with three uncertain demand scenarios) and a guaranteed bond return. By computing the expected value at the chance node ($160,000) and comparing it to the bond payoff ($60,000), the rational decision under risk-neutrality is to launch.

The decision tree above encodes every element of the expected-value framework. The square node on the left represents a decision point where the manager controls which branch to follow. The circle represents a chance event governed by probabilities that the manager does not control. Terminal payoffs are listed at the rightmost endpoints. The fold-back calculation multiplies each payoff by its probability, sums the products, and compares the resulting expected value ($160,000) to the certain payoff of the bond alternative ($60,000). Because $160,000 exceeds $60,000, a risk-neutral manager would launch the product.

Worked Example — Marketing Campaign ROI

A consumer-goods company is evaluating a $200,000 digital marketing campaign. Market research suggests three possible outcomes: a strong response (probability 0.25) generating $600,000 in incremental revenue, a moderate response (probability 0.50) generating $300,000, or a weak response (probability 0.25) generating only $100,000. The marketing VP wants to know whether the campaign's expected net payoff justifies the investment.

Expected Net Payoff of a Marketing Campaign
1
Step 1 — Define the Random VariableLet X represent the net payoff (incremental revenue minus the $200,000 campaign cost). The three outcomes are: strong response yields $600,000 − $200,000 = $400,000; moderate yields $300,000 − $200,000 = $100,000; weak yields $100,000 − $200,000 = −$100,000.
X ∈ {$400,000, $100,000, −$100,000}
2
Step 2 — Assign ProbabilitiesFrom market research: P(strong) = 0.25, P(moderate) = 0.50, P(weak) = 0.25. Verify the probabilities sum to 1: 0.25 + 0.50 + 0.25 = 1.00. ✓
ΣP(xᵢ) = 1.00
3
Step 3 — Compute Each Weighted PayoffMultiply each net payoff by its probability: 0.25 × $400,000 = $100,000; 0.50 × $100,000 = $50,000; 0.25 × (−$100,000) = −$25,000.
Weighted payoffs: $100,000; $50,000; −$25,000
4
Step 4 — Sum to Find E(X)E(X) = $100,000 + $50,000 + (−$25,000) = $125,000. The campaign's expected net payoff is positive.
E(X) = $125,000
5
Step 5 — Interpret the ResultAn expected net payoff of $125,000 means that if the company faced this exact decision repeatedly, the average net return per campaign would be $125,000. Since E(X) > 0, the campaign is worth pursuing under risk-neutral assumptions. The VP should additionally consider the 25% chance of a $100,000 loss when assessing risk tolerance.
Decision: Proceed with the campaign (risk-neutral criterion).

Strengths, Limitations & Comparisons

Expected value is arguably the most widely used decision criterion in business, but like any model it rests on assumptions that can break down in specific contexts. The table below contrasts the strengths and limitations of the expected-value framework, helping you recognize when it provides reliable guidance and when supplementary analysis is warranted.

Expected Value — Strengths vs. Limitations
DimensionStrengthLimitation
ComparabilityReduces complex, multi-outcome decisions to a single dollar figure, enabling direct comparison of alternatives.Collapses all distributional information into one number, ignoring variance and skewness of payoffs.
RepeatabilityOptimal for recurring decisions (pricing, inventory) where the law of large numbers ensures convergence to E(X).May mislead for unique, non-repeatable decisions (e.g., a one-time acquisition) where no long-run average applies.
Risk attitudeProvides a clear risk-neutral benchmark that anchors negotiation and budgeting.Assumes risk-neutrality; does not capture risk-averse or risk-seeking preferences without extension to expected utility.
Data requirementsRequires only outcome values and probabilities—inputs often available from historical data or expert judgment.Garbage in, garbage out: if probability estimates are biased (overconfidence, anchoring), E(X) will be systematically wrong.
ScalabilityExtends naturally to multi-stage decisions via decision trees and to continuous distributions via integration.Decision trees can become unwieldy with many stages or correlated uncertainties; simulation may be preferred.
KEY TAKEAWAY
Expected value is like a compass heading: it tells you which direction is north on average, but it says nothing about the terrain you will cross. A project with E(X) = $1 million but a 40% chance of bankruptcy is very different from a project with the same E(X) but guaranteed positive returns. In practice, pair expected value with variance analysis or scenario planning to get a fuller picture of risk.

Connection to Advanced Theory

The probability rules and expected-value framework presented in this lesson form the launching pad for several advanced topics in finance, operations, and strategy. Understanding how these concepts evolve prepares you for upper-division coursework and professional certifications like the CFA or FRM.

From Foundations to Advanced Applications
This Lesson's ConceptAdvanced ExtensionBusiness Application
Expected Value E(X)Expected Utility Theory — replace dollar payoffs with utility to model risk aversionInsurance pricing, portfolio allocation, executive compensation design
Conditional Probability P(A|B)Bayes' Theorem — update probabilities as new data arrivesSpam filters, medical diagnostics, A/B testing interpretation
Decision TreesReal Options Analysis — incorporate flexibility (delay, expand, abandon) into project valuationR&D investment staging, oil exploration, startup funding rounds
Addition & Multiplication RulesMonte Carlo Simulation — use probability rules inside computer-generated random samplesSupply-chain risk modeling, financial derivatives pricing, demand forecasting
Variance of a Discrete RVModern Portfolio Theory (Markowitz) — optimize the trade-off between E(X) and variance across assetsConstructing efficient investment portfolios, pension fund management

The key insight is that every advanced technique listed above builds on—rather than replaces—the foundational probability rules and expected-value logic you have learned here. Bayes' Theorem, for example, is simply the multiplication rule applied twice and then normalized. Monte Carlo simulation samples thousands of scenarios using the same addition and multiplication rules, then averages the results—which is nothing more than a brute-force estimate of E(X). Mastering these foundations now ensures that the advanced material will feel like a natural extension rather than an entirely new subject.

Practice Problems

PROBLEM 1CONCEPTUAL
A colleague claims that because the probability of a data breach at Server A is 0.10 and at Server B is 0.15, the probability of a breach at either server is 0.25. Under what condition is this calculation correct, and when would it overstate the true probability? Explain using the addition rule.
PROBLEM 2BASIC CALCULATION
A venture fund estimates three outcomes for a startup investment of $500,000: exit at $3,000,000 (probability 0.15), exit at $800,000 (probability 0.35), or total loss (probability 0.50). Compute the expected net payoff E(X) of this investment.
PROBLEM 3INTERMEDIATE
A supply chain has two independent links: a supplier with a 0.95 probability of on-time delivery and a logistics partner with a 0.90 probability of on-time delivery. (a) What is the probability that both deliver on time? (b) What is the probability that at least one fails to deliver on time? (c) If a late delivery from either link costs $50,000 in penalties and expediting fees while on-time delivery from both yields $0 in extra cost, what is the expected cost?
PROBLEM 4APPLIED
A pharmaceutical company must choose between two drug-development projects. Project Alpha costs $10 million upfront and has a 0.30 probability of FDA approval, which would generate $50 million in revenue, and a 0.70 probability of failure with zero revenue. Project Beta costs $4 million upfront with a 0.60 probability of approval generating $15 million in revenue and a 0.40 probability of failure. Using expected net payoff, which project should the company pursue? Discuss any additional factors a risk-averse CFO might consider.
PROBLEM 5CRITICAL THINKING
A firm is offered an exclusive licensing deal. Industry data shows that 20% of similar deals are highly profitable, 50% are moderately profitable, and 30% result in losses. However, a due-diligence report reveals that the licensor has a strong patent portfolio, and historically, deals with strong IP have a conditional probability of 0.40 of being highly profitable (versus the baseline 0.20). If P(strong IP) = 0.35 and P(strong IP | highly profitable) can be derived from the given data, use Bayes' theorem to find the updated probability of high profitability given strong IP. Explain how this updated probability would change the expected-value calculation for the deal.

Lesson Summary

This lesson established the essential toolkit for reasoning under uncertainty in a business context. We began with the historical foundations of probability—from Pascal and Fermat's correspondence through Kolmogorov's axioms—and then formalized the core probability rules: the complement rule P(A') = 1 − P(A), the addition rule P(A ∪ B) = P(A) + P(B) − P(A ∩ B), the multiplication rule P(A ∩ B) = P(A) × P(B | A), and conditional probability P(A | B) = P(A ∩ B) / P(B). These rules serve as the building blocks for computing the probability of any complex business event.

We then introduced expected value E(X) = Σ xᵢ × P(xᵢ) as the decision criterion that converts probability distributions into a single, comparable dollar figure. Using decision trees, we demonstrated how to structure multi-outcome decisions, fold back expected values, and select the optimal alternative. The lesson also addressed the limitations of expected value—particularly its assumption of risk-neutrality and its insensitivity to variance—and previewed advanced extensions including Bayes' theorem, expected utility theory, and Monte Carlo simulation. Armed with these tools, you can now systematically evaluate uncertain business scenarios and make decisions grounded in quantitative reasoning rather than intuition alone.

Varsity Tutors • Business Statistics • Probability & Expected Value — Probability Rules and Expected Value in Decision-Making