FIN 352
FIN 352  ·  California State University, Northridge

Investment Management

How assets are priced, how risk gets paid, and how to think clearly when real money and real people are involved.

An undergraduate course by Siqi Wei, Ph.D.
Department of Finance, Financial Planning, and Insurance
David Nazarian College of Business and Economics

Live · Figure 00 portfolios

Every dot is a portfolio. Mix five asset classes at random and the cloud draws its own boundary: the efficient frontier. The red line from the risk-free rate touches it once, at the tangency portfolio. Most of this course leads up to that picture.

OverviewWhat this course is for

A framework that outlasts the market you learn it in

Products change, tickers disappear, and interest rates move in directions no one predicted. The ideas in this course are the ones that still hold after all of that.

The course gives students the theory and the quantitative tools to understand how financial instruments are priced and how they are used in investment decisions. It is rigorous and critical about investing rather than a tour of current practice.

Students learn to value an asset from forecasts of its cash flows and the risks those cash flows carry. Common stocks get the most attention. Fixed-income securities and options each get their own full treatment. Along the way the course builds the economic and statistical models that connect risk to expected return, from Markowitz's mean–variance analysis to the CAPM and multifactor models. It then asks two questions the models leave out: who actually holds the shares, and what happens when the people trading them are not perfectly rational.

  1. Read the investment environment. Tell real assets from financial ones, map the asset classes, and explain how an order becomes a trade.
  2. Value a company from the top down. Move from the macroeconomy to the industry to the firm, and turn financial statements into a price.
  3. Price and hedge a bond. Link price to yield, read the term structure, and measure interest-rate risk with duration and convexity.
  4. Think in contingent claims. Draw the payoff of any option position and explain what drives its premium.
  5. Build the optimal portfolio. Quantify risk, diversify it, allocate capital, and price the risk that is left with the CAPM and factor models.
  6. Account for people. Explain how institutional owners shape firms and markets, and how predictable biases leave marks on prices.
Six parts, one argumentCourse map · click a part to jump
The sequence moves from the market's plumbing to the valuation of each claim, then to the portfolio that combines those claims, and ends with the people who hold them.
I
Part one · Chapters 1–4

The investment environment, its tools, and its machinery

Before pricing anything, we need to know what is being traded, where it trades, who stands in the middle, and what a trade costs.

Real vs. financial assetsAsset classesMarket indexesOrder booksMargin & short salesFunds & ETFs
Chapter 1The investment environment

Paper claims on real things

A society's wealth is its factories, land, software and know-how. A stock or a bond is a claim on the income those real assets produce, and financial markets decide who holds which claim.

Investment management is usually taught as two decisions made in order. Asset allocation decides how much goes into broad classes such as stocks, bonds and cash. Security selection then picks the specific holdings inside each class. Most of the long-run difference between two investors comes from the first decision, even though most financial news covers the second.

One dollar, invested at the end of 1927Real data · interactive
Growth of $1 on a log scale, 1928–2025. Hover to read any year. Stocks end far ahead, but the path runs through 1931 (−44%), 1974 and 2008. That is the risk–return trade-off. Source: Damodaran Online annual returns (S&P 500 with dividends, 3-month T-bill, 10-year T-bond, Baa corporate bond).
What the ninety-eight years say
AssetArithmetic meanGeometric meanStd. deviationWorst year$1 became
$589BJan 27, 2025
One day, one company, one question about AI spending

After the Chinese startup DeepSeek said its R1 model cost a small fraction of what U.S. labs spend to train, Nvidia fell nearly 17% and lost about $589 billion of market value, the largest one-day loss for any U.S. company. The Nasdaq fell about 3% while the Dow rose. That split is a clear example of an industry-specific shock rather than a market-wide one, a distinction that becomes the core of Part V.

Source: Yahoo Finance, Jan 27, 2025
Households
Supply the savings

They buy the claims directly or, far more often, through funds and retirement plans.

Firms & governments
Demand the capital

They issue stocks and bonds to pay for real assets and public spending.

Intermediaries
Stand in the middle

Banks, investment companies, insurers and pension funds pool money, spread risk and lower costs.

Chapter 2Asset classes & financial instruments

A field guide to what gets traded

Every instrument sits somewhere on a short list: the money market, the bond market, the equity market, and the derivatives built on top of them. Indexes summarize each of them in one number.

The taxonomyHover a branch
Maturity and priority of claim organize the whole map: short-term, low-risk debt on the left, then long-term debt, then the residual claim, then contracts whose value depends on all of these.

Why a 3.5% muni can beat a 5% corporate

Interest on most municipal bonds is exempt from federal income tax. To compare the two fairly, convert the muni yield into the taxable yield that would leave the investor with the same after-tax income.

Equivalent taxable yieldrtaxable = rmuni1 − tt is the investor's marginal tax rate. At the break-even rate t* = 1 − rmuni/rtaxable, the investor is indifferent between the two bonds.
Equivalent taxable yield across tax brackets. The curve rises faster as the bracket rises, which is why high-income investors are the natural buyers of municipal debt.

Two ways to average a market

The Dow Jones Industrial Average adds up 30 share prices. The S&P 500 weights each firm by its market value. Move the prices below and watch the two methods disagree on what "the market" did.

Start: ABC $25 with 20M shares; XYZ $100 with 1M shares.

The textbook two-stock example. A 20% gain in the large, cheap stock barely moves a price-weighted index, while a 10% loss in the small, expensive stock drags it down. A value-weighted index follows the money instead.
Chapters 3 & 4How securities trade · investment companies

From an order to a price

A market is a list of standing promises: buy up to here, sell down to there. Liquidity is how much of that list you can use before the price moves against you.

A limit order book, runningSimulation
Bids (teal) wait below the price, asks (red) wait above. New limit orders add depth; market orders take it away. Send a large market order yourself and watch it walk up the book. The gap between the best bid and best ask is what an impatient trader pays for immediacy.
Four orders every investor should be able to place
OrderWhat it saysWhat it guaranteesThe risk
MarketTrade now at the best available priceExecutionThe price, in a thin or fast market
Limit buy / sellTrade only at this price or betterPriceIt may never fill
Stop-lossBecome a market sell if the price falls to the stopA triggerGaps: the fill can be far below the stop
Stop-buyBecome a market buy if the price rises to the stopA triggerBuying at the top of a squeeze

Leverage cuts both ways: margin and short sales

Buying on margin means borrowing part of the purchase price from the broker. The investor's equity is the asset value minus the loan, and when equity falls below the maintenance requirement, the broker issues a margin call. A short sale runs the same logic in reverse: the loss is unlimited because the price has no ceiling.

1,000 shares bought (or shorted) at $100. The line is the margin ratio, equity divided by the position's value. Below the maintenance line (shaded), the broker calls. With 60% initial and 30% maintenance margin, a long position is called at $57.14.
January 2021: a short squeeze in daily dataReal data · GME
GameStop closing price (line, split-adjusted) and share volume in millions (bars), Dec 2020 – Feb 2021. Short interest above the entire free float, coordinated retail buying and call-option hedging drove the price up more than tenfold in two weeks. Shorts covering their positions added buying pressure. Source: Yahoo Finance.

Investment companies: what a fee really costs

Mutual funds and ETFs pool savings, diversify them and handle the administration. The price is an expense ratio, sometimes plus a load. The fee is quoted as a small percentage of assets, but it is charged every year, so its cost compounds.

$10,000 over 30 years in two funds with identical gross returns. The index fund charges 0.03%. The shaded gap is money that went to the fund sponsor instead of the investor.
Pooled vehicles, side by side
VehiclePriced atTradesDistinctive feature
Open-end mutual fundNAV, once a dayWith the fund itselfShares created and redeemed on demand
Closed-end fundMarket priceOn an exchangeOften trades at a discount to NAV
Exchange-traded fundMarket price ≈ NAVIntraday, on an exchangeIn-kind creation keeps price near NAV; tax efficient
Unit investment trustNAVRedeemed with sponsorFixed, unmanaged portfolio
Hedge fundPeriodic NAVRestrictedLightly regulated; leverage, shorting, performance fees
II
Part two · Chapters 17–19

Security analysis: from the economy to one share price

Fundamental analysis runs from the top down. Start with the economy, narrow to the industry, then read the firm's own statements and turn everything into a value per share.

Macro & the business cycleIndustry analysisDividend discount modelsP/E & PVGOFree cash flowDuPont & ratiosTextual analysis
Chapter 17Macroeconomic & industry analysis

The economy sets the tide

A firm's earnings depend on GDP growth, interest rates, inflation, exchange rates and policy long before they depend on its managers. An analyst has to be able to read the economy.

The top-down funnelFramework
The most famous leading indicatorReal data · FRED
10-year Treasury yield minus the 3-month bill rate, monthly, 1982 – Aug 2026; NBER recessions shaded. The spread dropped below zero before the 2001, 2008 and 2020 recessions, and came within 0.13 points of zero before the 1990 recession. The 2022–24 inversion was the deepest in this sample, yet no NBER-dated recession had followed it by the snapshot date. An indicator is evidence, not a guarantee.
Where the cycle isStylized · animated
Cyclical or defensive? Measure it
FirmTypeBeta

Betas on the S&P 500 ETF (SPY), 60 monthly returns to Aug 2026. Firms whose sales track the cycle carry higher market sensitivity.

Reading the cycle: the indicator families
FamilyMovesExamples
LeadingBefore the economyYield-curve spread, building permits, initial jobless claims, new orders, stock prices, consumer expectations
CoincidentWith the economyNonfarm payrolls, industrial production, personal income, manufacturing and trade sales
LaggingAfter the economyDuration of unemployment, inventories-to-sales, prime rate, unit labor costs, CPI for services
Chapter 18Equity valuation models

A share is worth the cash it will pay

Intrinsic value is the present value of all future dividends, discounted at the return investors require. The rest of the chapter is about making that infinite sum usable.

Constant-growth DDM (Gordon)V0 = D1k − gD1 is next year's dividend, k the required return, and g the perpetual growth rate. The formula works only when g < k, and it becomes very sensitive as g approaches k.
Value as a function of growth. Near g = k the curve turns almost vertical. Small disagreements about long-run growth produce large disagreements about price, and that is the main reason analysts disagree.

Growth is not the same as value

Split a price into the value of the current earnings stream, if all of it were paid out, plus the present value of growth opportunities. Reinvesting creates value only when the firm earns more on new capital than its investors require.

Earnings and growth opportunitiesP0 = E1k + PVGO,    g = b × ROEb is the plowback (retention) ratio. With ROE = k, PVGO is exactly zero no matter how much the firm reinvests.

E₁ = $5.00, k = 12.5%.

Price versus plowback, for the chosen ROE. Above k, every retained dollar adds value. Below k, the firm destroys value by growing, and the market penalizes it for reinvesting.
How much of today's price is growth?
FirmPVGO as a share of priceShare

The course's own calculation from the September 2, 2026 snapshot: E₁ = analysts' forward EPS, k from the CAPM with a 5% equity premium, PVGO = P₀ − E₁/k. A negative value means the market prices the firm below the value of its current earnings stream.

When growth is temporary: the two-stage model

Few firms can grow faster than the economy forever. A multistage model forecasts dividends year by year while growth is unusual, then switches to the Gordon formula once the firm matures. The classroom case is Better Mousetraps: a new product lifts growth to 20% for four years, then competitors catch up and growth falls to 5%. The last dividend was $1.00 and investors require 10%.

Two-stage dividend discount modelV0 = Σt=1..T D0(1+g1)t(1+k)t + PT(1+k)T,   PT = DT+1k − g2Stage one is valued dividend by dividend. Stage two collapses into a single terminal price PT, which is then discounted back like any other cash flow.
Present value of each year's dividend (teal) and of the terminal price (red). With the textbook inputs the stock is worth $34.74, and about 86% of that comes from the terminal value. This is the usual result in practice: the stable-growth assumptions carry most of the valuation.

Valuing the firm, not just the dividend

Many firms pay little or nothing in dividends and return cash through buybacks instead. S&P 500 companies spent a record $942.5 billion on buybacks in 2024, against $629.6 billion in dividends. Free-cash-flow models avoid the problem by valuing the cash the business generates, whatever the firm later does with it.

Free cash flow to the firmFCFF = EBIT(1 − t) + Dep − CapEx − ΔNWCDiscount at the WACC to get firm value; subtract debt to get equity.
Free cash flow to equityFCFE = FCFF − Interest(1 − t) + ΔDebtDiscount at the cost of equity k to get equity value directly.
Reverse DCF: what is Apple's price assuming?Course calculation · interactive
One-stage FCFF value per share against the market price. Inputs: fiscal 2025 FCFF of $98.8 billion (operating cash flow $111.5B minus capital spending $12.7B, from EDGAR), WACC ≈ 9.9%, and a $325 share price on the snapshot date. At 4% growth the model gives about $116. To justify $325, cash flow would have to grow about 7.7% a year forever. Reversing a DCF this way turns a price into a statement about the future that you can then judge.
+20%Feb 2, 2024
A first dividend as a signal

Meta announced its first-ever dividend, $0.50 a quarter, together with a $50 billion buyback. The stock jumped about 20% the next day. The dividend itself was small relative to the share price; the market was reacting to what it said about management's confidence in future cash flow, which is Miller and Modigliani's information effect.

Source: CNBC, Feb 2, 2024
Where the P/E ratio comes fromP0E1 = 1 − bk − ROE × bA high multiple is justified only by high ROE on reinvested earnings or by low risk (a low k). The PEG ratio, P/E divided by growth, is a rough shortcut for the same idea.
Three families of valuation
ApproachCore questionToolsWeak spot
Discounted cash flowWhat will it pay, and how risky is that?DDM, two- and three-stage models, FCFF/FCFETerminal value dominates
Relative (multiples)What do similar firms sell for?P/E, P/B, P/S, EV/EBITDA, PEGInherits the peers' mispricing
Contingent claimWhat is the flexibility worth?Option pricing on real assetsInputs rarely observable
Chapter 19Financial statement analysis

Taking ROE apart

Two firms can report the same return on equity for opposite reasons. The DuPont decomposition shows which lever (margin, turnover or leverage) is producing it.

Three statements, one firmHow they connect
Five-factor DuPontROE = Net incomePretax income × Pretax incomeEBIT × EBITSales × SalesAssets × AssetsEquityTax burden × interest burden × operating margin × asset turnover × leverage.
Illustrative business models. A grocer earns thin margins on very fast turnover. Software does the opposite. A bank reaches its ROE mostly through leverage. They can arrive at similar ROEs with very different risks.
EYOct 30, 2024
When the auditor walks away

After a short-seller report in August 2024, Super Micro Computer delayed its annual 10-K. In October its auditor, Ernst & Young, resigned, saying it could no longer rely on management's representations. Ratios can only be as good as the statements they come from, which is why earnings quality is part of the analysis.

Company 8-K filing via SEC EDGAR; widely reported by major financial media

Reading the words: textual analysis

A 10-K contains a few pages of numbers and tens of thousands of words. The words carry information the ratios miss: tone, hedging, what management chose to change, and what it left unsaid. Textual analysis turns that language into measurements that can be tested against returns, earnings and corporate outcomes, and it is now a standard part of finance research.

From a filing to a testable signalResearch pipeline
One passage, two dictionariesInteractive
negativepositiveuncertaintystruck = counted as negative, but not negative in a financial context
An illustrative MD&A-style passage scored with a few entries from each word list. A general-purpose dictionary reads tax, cost, capital, board, foreign, liabilities, vice and crude as negative, so an ordinary paragraph looks gloomy. Loughran and McDonald (2011) found that almost three-quarters of the negative words in the Harvard dictionary are usually not negative in financial text, and they built finance-specific lists in response.
What researchers measure in financial text
MeasureThe ideaWhat the evidence shows
ToneShare of negative and positive wordsFinance-specific negative words in a 10-K relate to filing-date returns, volume and later volatility. Loughran & McDonald (2011)
Media pessimismNegative words in a daily market columnHigh pessimism predicts downward pressure on prices, followed by a reversal. Tetlock (2007)
ReadabilityFog index, length, file sizeFirms with lower earnings file harder-to-read annual reports (Li, 2008). File size is a simple, robust readability proxy (Loughran & McDonald, 2014).
Changes from last yearSimilarity between consecutive 10-KsFirms that change their language substantially later underperform, and prices react slowly. Cohen, Malloy & Nguyen (2020)
Tone managementTone beyond what fundamentals justifyUnusually positive tone in earnings releases predicts weaker future earnings. Huang, Teoh & Zhang (2014)
Deception cuesWord choice on conference callsLinguistic models flag later-restated quarters better than chance. Larcker & Zakolyukina (2012)
Product textSimilarity of business descriptionsText-based industry networks capture competitors that standard codes miss. Hoberg & Phillips (2016)
Political riskPolitical topics in earnings-call transcriptsFirm-level political risk varies widely within industries and affects hiring and investment. Hassan et al. (2019)
Writing for machinesDisclosure when algorithms read filingsAs machine downloads rise, firms avoid words that dictionaries score as negative. Cao, Jiang, Yang & Zhang (2023)
Man + machineAI forecasts compared with analystsAn AI model beats most analysts, and combining the two does better than either alone. Cao, Jiang, Wang & Yang (2024)
2 paperswithdrawn 2025
A lesson about large language models and evidence

Two widely shared working papers claimed that ChatGPT-style models could read financial statements and filings better than people. Both were withdrawn by their own authors in 2025 after replication attempts failed. The methods are promising, but a working paper is not a result until it survives replication and peer review. That standard applies to AI research as much as to anything else in this course.

arXiv:2407.17866 (withdrawn Feb 20, 2025) · arXiv:2306.10224 (withdrawn Oct 9, 2025)
III
Part three · Chapters 14 & 16

Fixed income and the price of time

A bond is the cleanest valuation problem in finance: the cash flows are written in the contract. The difficulty is that the discount rate keeps changing, and this part is about measuring and managing that risk.

Bond pricingYield to maturityTerm structureSpot & forward ratesDefault riskDurationConvexityImmunizationAsset–liability management
Chapter 14Bond prices & yields

Price and yield move in opposite directions

A bond's price is the present value of its coupons and its principal. When market yields rise, those fixed payments are worth less today, and the price falls.

Bond pricePB = Σt=1..T C(1+y)t + Par(1+y)TWith semiannual coupons, use half the coupon, half the yield and twice the number of periods.
Bond lab: price, duration and convexityInteractive
The curved line is the true price; the straight line is the duration estimate. Duration is the slope of the price–yield curve. Convexity is its curvature, the reason the true price always sits above the tangent. The default yield, 4.68%, is the August 2026 average 10-year Treasury yield.
Twenty-five years of the Treasury yield curveReal data · press play
U.S. Treasury constant-maturity yields, 1 month to 30 years, monthly averages. Watch the curve flatten in 2006, fall to near zero after 2008 and again in 2020, then rise sharply and invert in 2022–23. Faded lines show the previous twelve months. Source: FRED (DGS1MO … DGS30).
Pull to parComputed
Three 10-year, 8% coupon bonds priced at constant yields of 6%, 8% and 10%. Premium bonds decline and discount bonds rise until all three are worth par at maturity. The price change comes from time passing, not from any change in market yields.
Chapter 16Managing bond portfolios · asset–liability management

Interest-rate risk, measured

Duration turns a bond's maturity, coupon and yield into one number that answers a practical question: how much will this position lose if rates rise by one percentage point?

Duration ruleΔPP ≈ −D* × Δy + ½ × Convexity × (Δy)2,   D* = D1 + yMacaulay duration D is the weighted-average time until the cash flows arrive; modified duration D* converts it into price sensitivity.
What drives duration
Holding the rest fixed, if…DurationWhy
Maturity risesRises (usually)Cash flows arrive later
Coupon risesFallsMore value arrives early
Yield risesFallsDistant cash flows are discounted more heavily
It is a zero-coupon bond= maturityThere is only one cash flow
It is a perpetuity(1+y)/yFinite even though maturity is infinite

Asset–liability management: duration on both sides of the balance sheet

Banks, insurers and pension funds all hold assets to meet liabilities, and both sides are exposed to interest rates. A bank funds long-term loans with deposits that can leave at any time. A pension fund owes benefits decades from now, so its liabilities have a longer duration than almost any bond it can buy. What matters is the mismatch: how far the two sides move apart when rates change.

Duration gapGap = DA − LA DL,    ΔNW ≈ −A × Gap × ΔyA positive gap means net worth falls when rates rise, the classic bank position. A negative gap means it falls when rates fall, the classic pension-fund position. A zero gap immunizes net worth against a parallel shift.
Balance-sheet stress testInteractive
Net worth after a parallel rate shock, for $100 of assets. The SVB-like preset uses the 6.2-year duration of Silicon Valley Bank's held-to-maturity portfolio reported in the Federal Reserve's April 2023 review; the deposit duration is illustrative. A 300-basis-point rise wipes out its $10 of net worth. The pension preset shows the reverse problem: long liabilities make falling rates the danger.

Immunization funds a dated obligation with a portfolio whose duration equals the horizon. If rates rise, the bond loses value but its coupons are reinvested at higher rates. If rates fall, the reverse happens. With duration matched to the horizon, the two effects offset. The textbook case: a 5-year obligation of $14,693.28 funded with $10,000 of 8%, six-year bonds, whose duration is 5 years.

Immunization: horizon value after an immediate rate changeComputed
Value at year 5 of $10,000 invested at 8% in three ways, if yields jump right after purchase and stay there. Short bonds that must be rolled over fall short when rates drop. Long bonds fall short when rates rise. The duration-matched six-year bond meets the obligation in every scenario, and convexity leaves it slightly above. Durations drift as time passes, so the portfolio has to be rebalanced to stay immunized.
+130bpSep 2022
When the hedge itself becomes the risk: UK pension funds

UK defined-benefit pension funds used leveraged liability-driven investment (LDI) to match their long-duration liabilities. After the September 2022 fiscal announcement, 30-year gilt yields rose about 130 basis points in three trading days. Collateral calls forced funds to sell gilts, which pushed yields higher still. The Bank of England bought £19.3 billion of gilts between September 28 and October 14 to stop the spiral. The hedge was right in principle, but the leverage and collateral needed to hold it created a liquidity problem.

Bank of England, Financial stability buy/sell tools: a gilt market case study, Quarterly Bulletin 2023
$1.8BMar 8, 2023
A duration mismatch becomes a bank run

Silicon Valley Bank had invested deposits in long-dated Treasuries and mortgage-backed securities just before the fastest rate increases in four decades. When it sold about $21 billion of securities at a $1.8 billion loss and announced a capital raise, depositors withdrew funds, and regulators closed the bank two days later. It is a real case of the asset–liability problem that immunization is meant to solve.

Passive
Indexing

Match a bond index's duration and sector weights; accept the market's pricing.

Passive
Immunization

Set asset duration equal to liability duration so price risk and reinvestment risk offset each other.

Active
Swaps & rate bets

Substitution, intermarket spread, rate anticipation and riding the yield curve.

IV
Part four · Chapters 20–21

Contingent claims: paying for the right, not the obligation

An option splits a payoff at a threshold. That single idea covers portfolio insurance, executive pay, corporate debt and the market's fear gauge.

Calls & putsMoneynessPayoff & profitStrategiesPut–call parityBlack–Scholes intuitionVIXOptions in ALM
Chapters 20 & 21Options markets · option valuation

Build a position, see its payoff

Every option strategy is a combination of four basic shapes. Choose one below, change the inputs, and compare the value at expiration with the Black–Scholes value today.

Dashed: payoff at expiration. Solid red: profit after premiums. Thin teal: position value today. Premiums come from Black–Scholes with the stock at $100. Raise volatility and the teal curve moves away from the payoff: time value is what an option buyer pays for uncertainty.
Put–call parityC + K(1+rf)T = S0 + PA call plus a bond pays exactly what the stock plus a put pays, in every state. If the prices differ, there is an arbitrage.
The market's fear gaugeReal data · FRED VIXCLS
CBOE Volatility Index, monthly average, 1990 – Aug 2026. VIX is the 30-day volatility implied by S&P 500 option prices, in other words the market price of insurance. It spikes when investors pay up for puts. Daily closes peaked at 82.7 in March 2020 and 52.3 on April 8, 2025.

Options inside the balance sheet

Asset–liability managers deal with options even when they never trade one. Many bonds, mortgages and deposit accounts give one side the right to change the cash flows, and that right is an option. A homeowner can refinance when rates fall. A depositor can withdraw when rates rise. A policyholder can surrender a contract. Each of these moves cash flows in the direction that hurts the institution, which is why duration alone underestimates the risk.

Negative convexity: a bond with a call option attachedIllustrative · computed
Callable bond = straight bond − call option held by the issuer. This is a 20-year, 7% bond; the call is valued with a Black–Scholes-style formula on the straight bond's price over a five-year horizon, a teaching simplification. As yields fall, the issuer's option gains value and the callable bond's price flattens near the call price. Mortgage-backed securities show the same compression because homeowners refinance.
Embedded options in asset–liability management, and how institutions hedge them
Where the option sitsWho holds itWhen it hurts the institutionTypical hedge
Callable bonds, mortgages, MBSIssuer or homeownerRates fall: assets prepay and must be reinvested at lower yieldsReceiver swaptions, interest-rate floors, callable funding
DepositsDepositorRates rise: funds leave or demand higher ratesPayer swaptions, interest-rate caps, shorter asset duration
Floating-rate loans with capsBorrowerRates rise above the cap and asset income stops risingBuy caps to offset
Insurance and annuity guaranteesPolicyholderRates or markets fall below the guaranteed levelFloors, equity puts, dynamic hedging
Surrender and withdrawal rightsPolicyholderRates rise and policyholders move money elsewhereLiquidity buffers, surrender charges, caps
What moves an option's price
If this rises…CallPut
Stock price S↑↓
Strike K↓↑
Volatility σ↑↑
Time to expiration T↑↑ (usually)
Interest rate r↑↓
Dividend payouts↓↑
V
Part five · Chapters 5–9

Risk, return, and the price of bearing it

Why does anyone earn more than the T-bill rate? Because they hold risk that cannot be diversified away. This part measures risk, removes the part that can be diversified, and prices the part that remains.

HPR & averagesReal vs. nominalRisk premiumSharpe ratioCapital allocationDiversificationMarkowitzIndex modelFama–FrenchCAPM
Chapters 5 & 6Risk, return & capital allocation

Ninety-eight years in one picture

Stock returns are not a single number. They are a distribution with a center, a spread and occasional very bad years, and an investor has to decide how much of that distribution to hold.

Every year of the S&P 500, 1928–2025Real data · hover a dot
Annual total returns sorted into 5-point bins; each dot is one year. The center is near 11%, and roughly one year in four is negative. The left tail is longer than the right. Source: Damodaran Online.
Fisher equationrreal ≈ rnominal − iAn investor is paid in purchasing power, not dollars. When inflation exceeds the interest rate, a "safe" deposit loses real value.
When cash lost to inflationReal data · FRED
Effective federal funds rate versus 12-month CPI inflation, 1960 – Jul 2026. Red shading marks negative real policy rates: the 1970s, the long post-2008 period and 2021–22. Teal shading marks positive real rates.

How much risk should you take?

Optimal position in the risky portfolioy* = E(rP) − rfA σP2A measures risk aversion. From the utility function U = E(r) − ½Aσ², the investor moves along the capital allocation line until the indifference curve just touches it.
The capital allocation line and the investor's best point on it. The risk-free rate is the August 2026 three-month bill (3.72%). Past the risky portfolio, the line represents borrowing to invest more than 100%.
Chapter 7Optimal risky portfolios

The only free lunch in finance

Combine assets that do not move together and portfolio risk falls faster than expected return. Diversification removes firm-specific risk. Market risk remains.

Risk falls, then stopsInteractive
Equally weighted portfolios of N stocks, each with σ = 45%. The floor is √ρ × σ, which is market risk.
Two assets, every mixInteractive
Stocks: E(r) 10%, σ 18%. Bonds: E(r) 5%, σ 9%. Dot: minimum-variance mix. Red line: CAL through the tangency portfolio.
"Diversification is both observed and sensible; a rule of behavior which does not imply the superiority of diversification must be rejected …"Harry Markowitz, Portfolio Selection, 1952
Chapters 8 & 9Index models · the CAPM

Only beta gets paid

Regress a stock's return on the market's. The slope is beta, the stock's exposure to risk that cannot be diversified away. The CAPM says beta, and only beta, earns a premium.

Security characteristic lineRi,t = αi + βi RM,t + ei,tTotal risk = β²σ²M (systematic) + σ²(e) (firm-specific). R² is the systematic share.

60 monthly returns, Sep 2021 – Aug 2026, against the S&P 500 ETF (SPY). Prices from Yahoo Finance, adjusted for dividends and splits.

Each dot is one month. Compare Coca-Cola's flat, tight cloud with the steep, scattered cloud of a semiconductor stock. The line's slope is beta, and the scatter around it is diversifiable risk.
Capital asset pricing modelE(ri) = rf + βi [E(rM) − rf]
The security market line, using measured betas and rf = 3.72%. Turn on the overlay to see each stock's realized average return over the same five years. Realized returns scatter widely and the fitted line is flatter than the CAPM predicts, a pattern documented since Black, Jensen and Scholes (1972) and exploited by "betting against beta" (Frazzini & Pedersen, 2014).
Factor 1 · MKT
Market

The excess return of the value-weighted market. The CAPM's single factor.

Factor 2 · SMB
Small minus big

Small firms minus large firms. A size premium that compensates for risks the market beta does not capture.

Factor 3 · HML
High minus low

High book-to-market (value) minus low (growth). Cheap, out-of-favor firms historically earned more.

Fama–French three-factor modelRi − rf = αi + βiMKT + siSMB + hiHML + ei
VI
Part six · Institutions, behavior, and new questions

Markets are made of people

Most shares are owned by institutions and traded by humans. Both facts help explain which models work and where they fail.

Institutional ownershipVoice & exitInvestment horizonLocal informationProspect theoryBiasesLimits to arbitrageTechnical analysis
Special topicThe role of institutional investors

Who owns the firm matters

Pension funds, mutual funds, insurers, endowments and hedge funds now own most U.S. equity. They can influence management by engaging with it (voice) or by selling (exit), and their incentives depend on how long they plan to hold.

Two instruments of influenceFramework

A trading rate is a horizon

A fund's turnover ratio measures how much of its portfolio it replaces. Each replacement is one sale plus one purchase, so the fraction of positions that actually change hands is half the turnover. That fraction converts directly into an average holding period.

Holding period = 12 ÷ (4 × turnover ÷ 2) months. A quarterly turnover of 0.25 means one-eighth of positions are replaced each quarter, so a typical position lasts two years. Index funds sit at the far left of this curve; high-frequency and short-term trading funds sit at the far right. The churn-rate measure follows Gaspar, Massa & Matos (2005).

Not all institutions are alike

Bushee (1998) sorted institutions by two habits: how often they trade and how concentrated their portfolios are. The result is still the standard way to separate owners who monitor from owners who mostly trade. Click a region to see how each type behaves and what the evidence says.

The owner map: turnover × concentrationInteractive · click a region
A conceptual map, not a data plot. Dedicated, quasi-indexer and transient are the three groups in Bushee (1998). The fourth corner, concentrated and fast-moving, is where activist hedge funds usually operate; it is added here for contrast and is not part of Bushee's scheme.

Common ownership: when every rival has the same owners

The three largest index-fund sponsors, BlackRock, Vanguard and State Street, together form the largest shareholder in most large U.S. companies. Fichtner, Heemskerk & Garcia-Bernardo (2017) put the figure at 88% of S&P 500 firms. If the same funds own every airline, do the airlines compete less hard? The question is one of the most contested in finance.

One owner set, four competitorsDiagram · hover an owner
The argument
Owners of the whole industry prefer softer competition

Azar, Schmalz & Tecu (2018) estimate that airline ticket prices are higher on routes where rival carriers share more of the same owners.

The reply
The link disappears under other measures

Dennis, Gerardi & Schenone (2022) find that the result is driven by how market share enters the ownership measure, not by ownership itself. The debate continues in finance, economics and antitrust policy.

Five decades of evidence

A short reading path through the research that shapes how we think about institutional owners, from monitoring and activism to price pressure and herding.

Landmark studies on institutional investorsInteractive timeline · click a year
Special topic · Chapter 12Behavioral finance

To understand the market, understand ourselves

The models assume rational investors, but decisions are made by people. Behavioral finance studies the predictable ways people depart from the models and asks why arbitrage does not always correct the resulting prices.

Try it before you read the answerExperiment

Choice 1: which would you take?

Choice 2: and now?

Answer both choices to see what most people pick.
The prospect-theory value functionInteractive
Value is measured relative to a reference point, not total wealth. The function is concave for gains (risk-averse), convex for losses (risk-seeking) and steeper for losses. The defaults are Tversky & Kahneman's (1992) estimates.
11.4%vs. 17.9%
Trading is hazardous to your wealth

Across 66,465 brokerage households in 1991–96, the most active traders earned 11.4% a year net of costs, compared with 17.9% for the market. Overconfidence is expensive.

Barber & Odean (2000), Journal of Finance
−$63Mar 2, 2000
The market could not subtract

After the Palm IPO, 3Com's own shares were priced below the value of the Palm shares it still owned, implying a negative value for the rest of 3Com. Short-sale constraints kept arbitrageurs from closing the gap.

Lamont & Thaler (2003), Journal of Political Economy
Why a mispricing can surviveMispricing = biased investors + limits to arbitrageFundamental risk, noise-trader risk and implementation costs mean a correct trade can still lose money before the price corrects (Shleifer & Vishny, 1997).
The canonIdeas this course is built on

Nine decades of research, in order

Every formula in the course comes from an argument someone first had to make. Students meet the originals, and many of their authors later received the Nobel Prize.