Every share, bond, and loan on earth is a promise about the future, and finance is the discipline of deciding what such promises are worth today. Three questions organize the whole field. What is time worth? What is the price of risk? And how much of what we know is already in the price? This long read walks through the machinery that modern finance built to answer them, from compound interest to the efficient frontier, with the curves and the formulas that made the subject a science.
Money’s First Axiom: Time Carries a Price
Strip any financial instrument down to its frame and you find the same skeleton underneath. A share of stock, a Treasury bond, a mortgage, an annuity, a lottery ticket: each is a claim on cash that has not arrived yet. Finance, at bottom, is the business of pricing such claims, and before it can say anything intelligent about risk it must answer a stranger question first. What is time itself worth? A dollar in your hand and a dollar promised for next year are not the same object, and the entire apparatus of modern markets rests on measuring exactly how different they are.
The intuition is ancient. A dollar today beats a dollar tomorrow for three reasons that reinforce one another:
- It can be put to work. Today’s dollar can be invested and start earning at once; the promised dollar sits idle until it arrives.
- It is certain. The future dollar depends on someone honoring a promise. Hold that thought; it becomes the subject of everything that follows.
- It will buy less. Inflation quietly shrinks what a future dollar can purchase, year after year, whether or not anyone announces it.
From this asymmetry comes the twin machinery of future value and present value. Future value asks what today’s money grows into if invested at a given rate of return. Present value runs the film backward: it asks what a promised future sum is worth right now, in hand. The two are mirror images, and the mirror is the interest rate.
Compound Interest, Undersold
Money grows the way it does because interest earns interest. That sounds like a bookkeeping footnote; it is actually the difference between a straight line and a curve that eventually goes vertical. At a return of roughly 7 percent a year, money doubles about once a decade, a shortcut captured by the rule of 72 (divide 72 by the rate to approximate the doubling time). One doubling is pleasant. But doublings stack. Four decades at that pace turns one dollar into roughly sixteen, and the final decade alone adds more than the first three combined. Exponential processes are back-loaded, which is why they bore the human eye for years and then astonish it.
Einstein is forever quoted calling compound interest the eighth wonder of the world. He almost certainly never said it, and the cliche, if anything, undersells the phenomenon by dressing arithmetic up as folk wisdom. The honest version is less quotable and more unsettling: modest, patient returns beat spectacular, interrupted ones, and time in the market matters more than nearly anything an investor can do with talent. Benjamin Franklin understood this in 1790, when he left roughly a thousand pounds each to Boston and Philadelphia with instructions to let the money compound for two centuries. By 1990 the funds had grown to millions of dollars, a bequest that did nothing clever except refuse to stop.
Discounting is the same engine thrown into reverse. To find what a future payment is worth today, you divide it by one plus the discount rate, raised to the number of years you must wait. The discount rate is best understood as an exchange rate between the present and the future, and small changes in it move distant values violently. A hundred dollars due in thirty years is worth roughly 55 dollars today when discounted at 2 percent, but only about 23 dollars at 5 percent. Same promise, same calendar, less than half the price. The exponent does that work, silently.
Inflation’s Silent Tax and the Anchor Rate
Every quoted return, meanwhile, comes in two flavors. The nominal return is the number printed on the statement; the real return is what survives after inflation, approximately the nominal rate minus the inflation rate. Inflation is the silent tax: never legislated, never itemized, collected from every holder of money without a single form to sign. At a modest 3 percent a year, purchasing power halves in roughly a quarter century. American savers learned the harsher version in the 1970s, when double-digit inflation meant that even respectable nominal yields left them poorer in real terms with each passing year.
All of this converges on one number: the risk-free rate, in practice the yield on short-term U.S. Treasury bills, loans to the federal government so brief and so secure that default and rate risk all but vanish. Its power comes from opportunity cost. Whatever else you do with money, you could have held Treasury bills instead, so their yield is the hurdle every other use of capital must clear. The risk-free rate is the pure price of time, and every other rate in the economy is built on top of it, premium by premium. When the anchor moves, everything chained to it moves too.
The year 2022 supplied the demonstration. With inflation surging, the Federal Reserve and its peers raised rates at the fastest pace in roughly four decades, taking short-term rates from near zero to above 4 percent within about a year. Mechanically, a higher discount rate shrinks present values, and the damage grows with distance: cash flows far in the future sit under a larger exponent. Assets whose value lives in the distant future, what practitioners call long-duration assets, fell hardest. Funds holding long-term Treasury bonds, the safest credits on earth, lost on the order of 30 percent. Growth stocks, whose earnings were projected years out, behaved like long bonds in disguise; the Nasdaq fell by roughly a third. Austria’s famous century bond, maturing in 2120, lost roughly three quarters of its peak market value. No borrower had become less trustworthy. Only the price of time had changed, and it repriced the world.
Time, then, has a posted price, refreshed daily in the Treasury market. But the certainty of a T-bill is the exception, not the rule. Most promises arrive wrapped in doubt: will the borrower pay at all, on schedule, in full? Pricing that doubt is where finance leaves pure arithmetic behind, and its oldest laboratory is the bond market.
| $10,000 invested at | 10 years | 30 years | 50 years |
|---|---|---|---|
| 4% per year | $14,802 | $32,434 | $71,067 |
| 7% per year | $19,672 | $76,123 | $294,570 |
| 10% per year | $25,937 | $174,494 | $1,173,909 |
Bonds: Where Time Meets Arithmetic
Equities get the headlines, but bonds get the money. The global bond market is, by most estimates, larger than all the world’s stock markets combined, with outstanding debt well over one hundred trillion dollars, and it is arguably the most information-dense market on earth. Every price in it encodes a forecast: of inflation, of central bank policy, of the odds that a borrower makes good on a promise. Traders like to say the bond market is smarter than the stock market; it is certainly more explicit.
A bond is the simplest financial promise there is. A borrower, whether a government or a corporation, agrees to pay the holder a fixed series of interest payments, the coupons, and then to return the principal, the face value, at maturity. That is the whole contract. Because the cash flows are spelled out in advance, the price follows directly from the logic already built: a bond is worth the present value of its promised payments, each discounted back to today at the going rate of interest. Nothing about it is mysterious: the time value of money, applied line by line.
Price and Yield, Bolted to a Seesaw
From that one fact flows the most important mechanical relationship in finance: when yields rise, bond prices fall, and when yields fall, bond prices rise. Always. That is not a tendency; it is arithmetic. An existing bond’s coupons are fixed the day it is issued; if market rates climb, newly issued bonds offer fatter coupons, and the old bond can compete only if its price drops far enough that a buyer, paying less for the same stream of cash, earns the new, higher rate. Price and yield are two descriptions of the same object, and they can only move in opposite directions.
How far the price moves is a question of duration, which measures the sensitivity of a bond’s price to a change in yield. As a rough rule, a duration of seven means a loss on the order of seven percent when yields rise by one percentage point. Two features stretch duration out. The first is maturity: the further away a cash flow sits, the harder a change in the discount rate hits its present value. The second is the coupon: the lower it is, the more of the bond’s value rides on the distant repayment of principal, and the more the instrument behaves like one big far-off payment. A thirty-year bond issued with a tiny coupon is, in effect, a long lever arm bolted to interest rates.
Duration is only a first approximation, because the price-yield relationship is not a straight line but a curve, and the curvature, called convexity, bends in the holder’s favor. When yields rise, prices fall by a bit less than duration alone would predict; when yields fall, prices rise by a bit more. For ordinary bonds it is a small, quiet gift: padding on the way down, spring on the way up.
Reading the Curve
Line up the yields on government bonds from the shortest maturity to the longest and you get the yield curve, the term structure of interest rates, and probably the most watched chart in macroeconomics. Its normal shape slopes upward: lenders demand extra yield for locking money away longer, a compensation known as the term premium, and expectations of growth tend to mean expectations of higher rates ahead. A flat curve reads as hesitation, a market unsure which way policy is heading. An inverted curve, with short-term yields above long-term ones, is the strange one, with a famous reputation. Inversion says, in effect, that investors expect the central bank to be cutting rates before long, and central banks cut rates when economies weaken. Historically, in the United States, inversions have preceded recessions with striking regularity.
That reputation was tested recently. From 2022 into 2023, as the Federal Reserve raised short-term rates rapidly to fight inflation, the US curve inverted deeply and stayed inverted for an unusually long stretch, by some measures the deepest inversion in roughly four decades. A recession did not arrive on the timetable the signal’s history suggested, and economists spent the episode debating whether the indicator had misfired or was simply early rather than wrong. The honest reading is that an inverted curve is a warning, not a verdict: it reveals what the market fears, not what the future holds.
Pricing a Promise: Credit Spreads
So far the promises have been assumed good. When they might not be, the market charges for the doubt. A corporate bond yields more than a Treasury of the same maturity, and the gap, the credit spread, is the going price of default risk. Rating agencies sort borrowers into investment grade, the sturdier credits, and high yield, or junk, less politely, and the spreads behave accordingly. In calm markets, investment-grade spreads have tended to sit on the order of one to two percentage points over Treasuries, while high-yield spreads have run several times that; in true crises, they have blown out above ten percentage points, as in 2008. The spread is a fear gauge with a coupon attached: it widens when investors doubt and compresses when they relax.
All of this arithmetic converged, brutally, in 2022. A decade of near-zero rates had left bonds with historically low coupons, historically high durations, and almost no income cushion to absorb losses. When inflation returned and yields surged, the seesaw did exactly what the formula said it would. Broad US investment-grade bond indexes lost on the order of thirteen percent that year, their worst showing in the modern history of such benchmarks, and long-maturity Treasuries fared far worse. The lesson was not that bonds are secretly dangerous; it was that starting yields matter, and that even the safest borrower on earth cannot shield a lender from discounting. What failed that year was not any single asset but the assumption that bonds would always cushion stocks, which raises a deeper question: how do assets move together, and what protection can combining them buy?
| Treasury (illustrative) | Approx. duration | Price move if yields rise 1 point |
|---|---|---|
| 2-year note | about 1.9 | roughly −2% |
| 10-year note | about 8.5 | roughly −8.5% |
| 30-year bond | about 19 | roughly −19% |
Risk, Return, and the Only Free Lunch
Interest rates put a price on time. But time is only half of what an investor sells when capital changes hands. The other half is certainty. A Treasury bill promises a fixed sum on a fixed date; a share of stock promises whatever is left after every other claimant has been paid, which may be a fortune or nothing. The distance between those promises has a price of its own, and pricing it requires, first, a way to measure risk.
Volatility, the Workhorse
Finance settled on a deceptively simple gauge: dispersion. Take an asset’s returns over many periods, ask how widely they scatter around their average, and summarize the scatter in a single number, the standard deviation. An asset whose annual returns cluster between 3 and 5 percent is calm; one that swings between minus 30 and plus 40 percent is not, even if both average out to the same figure. Volatility became the workhorse measure of risk not because it is perfect but because it is tractable. It can be estimated from data, compared across assets, and, crucially, combined across a portfolio using ordinary statistics.
Its imperfections are well cataloged. Market returns do not follow the tidy bell curve the number implicitly assumes. Extreme moves arrive far more often than a normal distribution predicts, the famous fat tails: the roughly 20 percent one-day fall in US stocks in October 1987 was, on textbook assumptions, an event that should not have happened once in the lifetime of the universe. Volatility also treats upside and downside symmetrically, though no investor complains about upward surprises. And it says little about drawdowns, the peak-to-trough collapses that actually end investing careers: American stocks lost on the order of 80 percent of their value from 1929 to 1932, and roughly half from late 2007 to early 2009. Standard deviation is a thermometer, not a diagnosis. Still, as a first approximation it works, and it makes the rest of portfolio theory possible.
Climbing the Ladder of Risk and Return
Measured this way, the long American record arranges itself into a strikingly consistent hierarchy. Over roughly the past century, Treasury bills, the closest thing to a riskless asset, have returned something on the order of 3 percent a year, barely ahead of inflation over the long run. Government bonds, which add interest rate risk, have delivered roughly 5 percent, with annual swings in the high single digits. Large-company stocks, which add the full uncertainty of enterprise, have returned roughly 10 percent a year before inflation, or something like 6 to 7 percent after it, with an annual standard deviation of very roughly 20 percent. The precise figures shift with the start date and the dataset, but the ordering never does: more risk, more long-run reward.
The gap between what stocks have earned and what bills have earned, historically on the order of 4 to 6 percentage points a year, is the equity risk premium. It is best read as a wage: compensation paid to whoever agrees to hold the asset that falls hardest at the worst moments, and to keep holding it through the drawdowns. Compounded over decades, that wage is enormous, which is precisely why the volatility that earns it cannot be wished away. If stocks were reliably safe, they could not stay cheap enough to return more than bonds.
Diversification, the Only Free Lunch
Nearly everywhere in markets, more return means more risk. There is one celebrated exception, and it rests on a piece of arithmetic. A portfolio’s expected return is simply the weighted average of its holdings’ expected returns. Its volatility, however, is not. Unless the holdings move in perfect lockstep, some of one asset’s bad days land on another’s good days, and the offsets cancel. Combine imperfectly correlated assets and the portfolio’s swings shrink below the weighted average of the individual swings, while the expected return does not shrink at all. Risk falls; return, in expectation, does not. That asymmetry is why diversification is routinely called the only free lunch in investing.
The lunch has a limit, and the limit is instructive. Some risk is idiosyncratic, specific to a single company: a failed drug trial, a fraud, a factory fire. Spread money across enough names and those shocks wash out. Classic studies from the late 1960s onward found that a randomly assembled portfolio of roughly 20 to 30 stocks eliminates most single-name volatility; additional holdings help only at the margin. What remains is systematic risk, the tide that moves every boat at once: recessions, rate shocks, panics. No amount of diversification removes it, because it is common to everything. The distinction carries a sharp implication for pricing. Markets should pay a premium only for systematic risk, since idiosyncratic risk can be shed for free, and no one gets paid for bearing what costs nothing to avoid.
60/40 and Its Stress Test
The classic embodiment of the free lunch is the 60/40 portfolio: roughly 60 percent stocks for growth, 40 percent bonds for ballast. For much of recent history the two moved loosely or even in opposite directions, since the recessions that crush equities usually pull interest rates down and push bond prices up. Then came 2022. Inflation surged, central banks raised rates aggressively, and both legs buckled together: US stocks fell somewhere in the neighborhood of a fifth, while even high-quality bonds lost on the order of 10 to 15 percent, handing the 60/40 mix one of its worst years in decades. The episode did not repeal diversification; it exposed the fine print. Correlations are not constants, and inflation is the one enemy that attacks nominal bonds and equity valuations at the same time.
Diversification, then, tells an investor that mixing assets is wise. It does not say how much of each to hold, or which combination wrings the most return from a given tolerance for risk. Someone had to turn the free lunch into mathematics, and in 1952 a young graduate student in Chicago did exactly that.
| Correlation between the two assets | Portfolio volatility (50/50 mix of two 20%-volatility assets) |
|---|---|
| +1.0 (move together perfectly) | 20.0% |
| +0.5 | 17.3% |
| 0.0 (independent) | 14.1% |
| −0.5 | 10.0% |
| −1.0 (perfect hedge) | 0.0% |
Markowitz and the Geometry of Portfolios
In the spring of 1952, the Journal of Finance published a short article called “Portfolio Selection” by Harry Markowitz, a University of Chicago doctoral student then in his mid-twenties. It contained almost nothing that Wall Street considered wisdom. There were no stock tips, no verdicts on railroads or steel companies. Instead there were statements about means, variances, and covariances. The paper landed quietly, then spent the following decades reorganizing the profession, because it changed the question investors were asking. The old question was: which securities are good? The new question was: how do securities behave together?
The shift sounds modest and is anything but. Before Markowitz, the craft of investing was security selection, and the implicit theory held that a portfolio was simply a collection of individually good ideas. Markowitz pointed out that this treats a portfolio like a bag of marbles when it is really a chemical compound. The risk an asset adds to your holdings is not its own volatility but its covariance with everything else you own. A violently volatile gold miner can reduce the risk of a stock portfolio if it tends to rise when the rest falls. A placid utility can add risk if it sinks in exactly the recessions that sink your other holdings. Judged in isolation, the miner looks reckless and the utility looks safe. Judged in context, the ranking can flip.
Tracing the Efficient Frontier
Markowitz’s real contribution was to turn that observation into geometry. Summarize any portfolio with just two numbers: its expected return and its volatility. Plot risk on the horizontal axis and expected return on the vertical, and every possible portfolio, every conceivable blend of assets, becomes a single point on the map. Together those points form a dense cloud known as the feasible set.
Most of that cloud is dead weight. For nearly any portfolio inside it, another exists that offers more expected return at the same risk, or the same return at less risk. No sane investor wants a dominated point, so attention collapses onto the cloud’s upper-left boundary: the efficient frontier. Northwest is the compass heading of every rational investor, because north means more return and west means less risk. Portfolios on the frontier cannot be improved in one dimension without paying in the other. At the frontier’s leftmost tip sits the minimum-variance portfolio, the one combination of risky assets with the lowest volatility achievable, the closest a fully invested portfolio can come to standing still. From there the frontier bends up and to the right, and choosing a spot along it becomes a matter of temperament: how much extra turbulence will you accept for extra expected return?
One Line to Rule Them All
Then comes the twist that made the theory operational. Add a risk-free asset, something like a Treasury bill, and the geometry simplifies dramatically. Mixing cash with any risky portfolio traces a straight line on the map. Anchor that line at the risk-free rate on the vertical axis and pivot it upward until it just grazes the curved frontier. The line is the capital market line, and the point where it kisses the frontier is the tangency portfolio. Every combination on that line beats every portfolio on the old frontier except the tangency point itself.
The implication, often called two-fund separation, is startling. In this idealized world, every investor should hold the same portfolio of risky assets, the tangency portfolio, and differ only in how it is mixed with cash. The cautious retiree keeps most of her money in Treasury bills and a sliver in the tangency portfolio. The aggressive investor borrows and holds more than one hundred percent of his wealth in it. Risk appetite determines the dose, not the medicine. Portfolio choice splits cleanly into two decisions: what the best bundle of risky assets is, and how much of it you can stomach.
The slope of that line has become finance’s standard scorecard. It measures how much expected return a portfolio delivers above the risk-free rate for each unit of volatility, and it is called the Sharpe ratio, after William Sharpe. Long-run diversified stock portfolios have historically earned Sharpe ratios on the order of a few tenths. A manager reporting a ratio near two, year after year, deserves suspicion rather than applause; Bernard Madoff’s impossibly smooth returns implied exactly that kind of number.
Where the Map Misleads
In practice, the elegant machine has sharp edges. The optimizer needs estimates of expected returns, volatilities, and correlations, and expected returns in particular are notoriously noisy. Small changes in the inputs can produce wildly different “optimal” portfolios, and the mathematics leans hardest on the numbers most likely to be wrong, piling money into whatever asset a flattering estimate favors. Practitioners call this garbage in, gospel out: the computer launders shaky guesses into precise-looking allocations. Modern users tame it with constraints, conservative inputs, and heavy doses of humility.
The deeper caveat is that correlations are not loyal. They are measured mostly in calm markets, and they tend to climb toward one in a panic, exactly when diversification is supposed to earn its keep. In the crash of October 1987 and again in the autumn of 2008, assets that had seemed pleasantly unrelated fell together, with high-grade government bonds among the few reliable exceptions. Diversification, it turns out, resembles an insurance policy whose coverage shrinks as the floodwater rises. It remains the best free lunch in finance; it is just not an unlimited buffet.
Markowitz shared the Nobel Prize in economics in 1990, nearly four decades after his article appeared. He liked to admit that, when choosing his own retirement allocation years earlier, he skipped the optimizer and split his contributions roughly evenly between stocks and bonds to minimize future regret. Even the architect of portfolio theory was human. His successors were more literal-minded. William Sharpe, among others, asked what would happen to prices if everyone actually followed Markowitz’s advice, and the answer became the most famous model in finance.
CAPM: Hanging a Price Tag on Every Risk
Markowitz had answered a private question: given the menu of risky assets, how should one investor combine them? He left a public question hanging. If every investor followed his advice, what would happen to prices? The answer arrived in the mid-1960s from William Sharpe, a young economist who had sought out Markowitz for guidance on his dissertation, and independently from Harvard’s John Lintner, with the Norwegian economist Jan Mossin close behind and Jack Treynor circulating similar ideas in unpublished notes. The result was the Capital Asset Pricing Model, and it did something no theory of markets had managed before: it attached an explicit price tag to risk itself.
The logic is disarmingly compact. Suppose all investors see the same expected returns, the same volatilities, the same correlations, and can borrow and lend at the same risk-free rate. Then Markowitz’s mathematics hands each of them the same answer: hold the tangency portfolio, the one best bundle of risky assets, and adjust for temperament only by mixing it with cash or leverage. But here is the twist. If everyone wants the same risky portfolio, then in equilibrium that portfolio must be the market itself, every asset held in proportion to its total value. It cannot be otherwise: prices adjust until the securities investors collectively wish to hold are exactly the securities that exist. The tangency portfolio, that abstract object on Markowitz’s frontier, turns out to be the entire market wearing a mathematical disguise.
Beta, or the Only Risk That Pays
From that single equilibrium fact flows a radical conclusion. The risk that determines an asset’s expected return is not its standalone volatility. It is the asset’s covariance with the market portfolio, distilled into a single number: beta. A stock with a beta of 1 tends to move in step with the market. A beta of 2 amplifies each market swing; a beta of 0.5 dampens it. Everything else about the stock’s turbulence, the lawsuits, the failed product launches, the eccentric founder, is idiosyncratic noise that a diversified investor has already washed out. And the market pays no premium for risk you could have eliminated for free. In the model’s stern accounting, a wildly volatile stock whose fortunes are uncorrelated with everything else is, for pricing purposes, almost riskless.
The pricing relation that results is the security market line. In prose: an asset’s expected return equals the risk-free rate plus the asset’s beta multiplied by the market risk premium, the extra return investors demand for holding the market rather than cash. Expected return rises linearly with beta, and with nothing else. Double the beta, double the compensation above the risk-free floor. It is a supply-and-demand curve for courage.
Beta is not merely a theorist’s toy; it maps recognizably onto the business landscape. Regulated utilities, whose customers keep the lights on in booms and busts alike, tend to carry betas well below 1, often on the order of 0.5. Consumer staples behave similarly. At the other end sit businesses that live and die with the cycle: semiconductor makers, luxury-goods houses, airlines, and much of the technology sector, where betas above 1 are common. The pattern is intuitive once seen. Defensive cash flows barely flinch when the economy stumbles; cyclical ones swing harder than the market that prices them.
Boardroom Workhorse: CAPM in Practice
That practicality explains why CAPM escaped the journals and colonized the corporate world. A company weighing a new factory needs a hurdle rate, a minimum return that justifies tying up shareholders’ capital, and CAPM offers a defensible recipe: estimate the project’s beta, look up the risk-free rate, add a beta-scaled slice of the market premium, and you have a cost of equity. Surveys of chief financial officers have long found it the most widely used method for exactly this purpose. Utility regulators lean on the same machinery when they set the allowed returns that pipeline and power companies may earn, turning a piece of equilibrium theory into rate-case arithmetic argued before public commissions.
The empirical record, however, has been less obliging than the boardroom embrace suggests. Almost as soon as researchers could test the model properly, they found that the line relating beta to average returns, while broadly upward sloping over the long run, is flatter than the theory predicts: low-beta stocks have historically earned more than the model says they should, and high-beta stocks less. Then came the anomalies. Small stocks appeared to outperform beyond what their betas justified. So did cheap stocks trading at low multiples of book value, and stocks that had recently been rising, the momentum effect. Even the low-volatility finding hardened into an anomaly of its own, the boring stocks quietly beating the exciting ones. By the early 1990s Eugene Fama and Kenneth French had gathered the evidence into a multifactor alternative, adding size and value factors to the market factor, and later work extended the list. Where CAPM saw one priced risk, its successors saw several.
Yet the model refused to die, because its failures never dislodged its language. Beta remains the standard shorthand for market sensitivity; alpha, the return left over after beta has taken its share, remains the industry’s definition of skill, and every fund manager who claims to deliver it is measured against a CAPM-shaped ruler. The core insight, that only undiversifiable risk should command a premium, survives every empirical bruise. When Sharpe shared the 1990 Nobel Prize with Markowitz and Miller, the committee was honoring not a formula that fit the data perfectly but a way of thinking that had reorganized finance around it.
Still, the whole edifice rests on a quiet assumption. For beta to be the right measure and the market portfolio the right benchmark, prices must already embody what investors collectively know. Whether they actually do is not a footnote. It is the next battleground.
Efficient Markets and Their Discontents
Every idea in this story so far, from discounting to diversification to beta, points toward one unsettling question: if risk is priced this rationally, can anyone beat the market at all? In 1970, Eugene Fama of the University of Chicago distilled two decades of research into a sweeping claim. In an efficient market, he argued, prices fully reflect all available information. New information moves prices, but because news is by definition unpredictable, price changes are too. The best forecast of tomorrow’s price is today’s price plus a modest reward for bearing risk.
Fama sorted the claim into three forms, each ruling out a different way of getting rich. The weak form says past prices contain no exploitable patterns, which dooms technical analysis: the chart patterns, the moving averages, the folklore of reading the tape. The semi-strong form says prices already incorporate all public information, which dooms ordinary stock picking: by the time an earnings report or a merger announcement reaches your screen, the price has moved. The strong form says even private information is already in the price, which would make insider trading unprofitable. Almost no one believes the strong form; insiders demonstrably profit from what they know, which is why the law forbids them to trade on it. But the weak and semi-strong forms held up disturbingly well against decades of evidence.
Sharpe’s Arithmetic That Cannot Lose
In 1991 William Sharpe, whom we last met building the CAPM, published a short essay whose logic is closer to accounting than to economics. The market’s return is, by definition, the weighted average return of everyone who holds the market. Index investors, who simply hold everything, earn that average before costs, so active investors as a group must earn it too, because together the two groups are the market. Subtract fees, and the conclusion is airtight: after costs, the average actively managed dollar must underperform the average passively managed dollar. Not usually. Always, as a matter of arithmetic. Some managers beat the market in any given year, but for every winner there is a loser on the other side of the trade, and both are paying fees.
John Bogle turned this arithmetic into a business. In 1976 his young firm, Vanguard, launched the first index mutual fund for the general public. Wall Street derided it as “Bogle’s folly,” and the launch raised only a small fraction of its target. The folly compounded quietly for decades. Low cost, it turned out, was the one advantage that never faded. Today trillions of dollars sit in index funds and exchange-traded funds, and passive vehicles account for something on the order of half of American fund assets, one of the great migrations of capital in financial history.
Behavioral Counterattack
Yet the hypothesis has a soft underbelly, and Fama himself identified it. Any test of market efficiency is really a joint test: you must first assume a model of what returns should be, such as the CAPM, before you can declare a price wrong. A strategy that earns suspiciously high returns always admits two explanations. Either the market is inefficient, or your pricing model has mismeasured risk. The data alone cannot tell you which. This joint hypothesis problem means efficiency can never be cleanly proven or refuted, which is either a profound insight or a convenient escape hatch, depending on whom you ask.
The psychologists pressed hardest. Beginning in the 1970s, Daniel Kahneman and Amos Tversky documented how systematically human judgment departs from the cool rationality that efficiency assumes. People are overconfident: most of us believe we are above-average drivers, and traders bring the same conviction to their portfolios, trading too often and paying for it. People are loss averse: losing a dollar hurts roughly twice as much as gaining one feels good, which helps explain why investors cling to losers and sell winners too soon. And people herd, finding comfort in the crowd precisely when the crowd is most dangerous.
History kept supplying exhibits. On October 19, 1987, American stocks fell by roughly a fifth in a single day, with no news remotely proportionate to the move. In the late 1990s the Nasdaq roughly quintupled on the promise of the internet, then surrendered most of those gains within a few years. In 2008, securities the models had rated nearly riskless collapsed and carried the banking system to the brink. And in early 2021, shares of GameStop, a struggling mall retailer, rose many times over within weeks as retail traders coordinated on social media, then fell most of the way back. Each episode can be rationalized after the fact. Together, they strain the picture of a market that always prices risk correctly.
The institutions delivered their verdict with a wink. In 2013 the Nobel committee awarded the economics prize jointly to Eugene Fama, who taught us that prices are unpredictable, and Robert Shiller, who taught us that markets swing to irrational extremes, along with Lars Peter Hansen, who built the tools to test both. Honoring the thesis and the antithesis in the same ceremony was the academy’s way of conceding that each man is, in his fashion, right.
Lessons That Survive
The pragmatic synthesis runs roughly as follows. Markets are not perfectly rational, but they are ferociously competitive, and those are not the same thing. Prices can be wrong, yet knowing when they are wrong, and staying solvent while you wait to be proven right, is brutally hard. Sharpe’s arithmetic grinds on regardless: costs are the one variable an investor controls completely, and over decades a difference of one percentage point in annual fees compounds into a startling share of a lifetime’s wealth. The humility lesson is not that nobody ever beats the market. It is that you should not bet your retirement on being the exception.
The price of risk is real, observable, and largely beyond any one investor’s control; what remains in your control is how much risk you buy, how cheaply you buy it, and how honestly you judge your capacity to hold on when it misbehaves. Markets will keep stumbling and theorists will keep arguing, and through it all the patient, diversified, low-cost investor will keep collecting the one premium a century of finance suggests is durably on offer: the reward for bearing risk you can afford to take.