Risk and Return — The Four Faces You Need to Name

There is no such thing as a risk-adjusted return. There is only return adjusted for which kind of risk you bore. The history of finance is the history of clever people who measured one face of risk, optimised for it, and were destroyed by another. Sharpe ratios, value-at-risk, beta — they all collapse a fundamentally multi-dimensional thing into one number, and the dimensions you ignore are the ones that take you out.

1 · The four faces of risk

Risk is not one thing. It is at least four things, and they live in different dimensions. Treating them as one number is the original sin of retail finance.

• Volatility. How spread-out the returns are around their mean. Lives in spreadsheets. Measured in standard deviations. Symmetric — a 10% gain and a 10% loss contribute equally.
• Drawdown. How far below the running peak you sit. Lives in 2 a.m. insomnia. Asymmetric — you cannot eat upside.
• Illiquidity. The gap between when you want your money and when you can get it. Lives in PDFs marked “subject to fund manager discretion.”
• Ruin. The probability you go to zero and cannot recover. Lives in obituaries. Different in kind from the others — a 60% drawdown that you can wait out is not ruin; a 60% drawdown into which you are leveraged 3:1 is.

These are not synonyms. A high-volatility asset can have low ruin risk (a diversified equity index over thirty years). A low-volatility asset can be ruinous (a leveraged short-volatility fund: smooth daily returns until the day it goes to zero). A high-Sharpe strategy can have catastrophic illiquidity (LTCM had a Sharpe near 4 right up until it didn’t). The face of risk you measure is not, in general, the face that kills you.

2 · The graveyard of one-number thinkers

Three short obituaries. Each one was an investor or fund that optimised for one face of risk and was killed by another. None of these are obscure — they are the most-studied failures of the last thirty years, and the lesson is the same lesson every time.

LTCM, 1998 — killed by illiquidity, not volatility

Long-Term Capital Management ran a fixed-income arbitrage book with a realised Sharpe near 4 for four straight years. The strategy traded the spreads between near-identical bonds — almost no day-to-day volatility, because the bonds moved together. The failure mode was that the spreads were illiquid: when Russia defaulted in August 1998, every counterparty in the trade wanted out at once, and there was no one on the other side. LTCM’s volatility number never warned them. Its illiquidity exposure did the killing — and the firm lost $\$4.6\text{B}$ in four months and was wound down.

Short-volatility funds, 2018 — killed by drawdown, not vol

A class of products — XIV most famously — sold short-dated volatility for years, posting smooth single-digit annual gains and a beautiful Sharpe. The volatility metric stayed low because volatility itself was low. On 5 February 2018, the VIX rose from 17 to 50 in a single day, and XIV lost roughly 96% of its value overnight. The product was terminated. Half a decade of “smooth returns” was wiped out in a single drawdown event that the volatility metric could not, by construction, see in advance.

FTX customers, 2022 — killed by ruin, not anything visible

Customers who held assets on FTX experienced no volatility, no drawdown, no illiquidity — until the day the company failed and 100% of their balance disappeared. Counterparty risk is a pure ruin event: invisible until it isn’t. No portfolio metric computed from price history can tell you that the custodian holding your assets is solvent. The asset returned its expected return every single day, and then it returned negative one hundred percent.

Three different funds, three different risk types, three different graveyards. In each case the optimisation target — the one number the manager was being graded on — was not the dimension that did the killing. That is not a coincidence. That is the structure of how finance fails.

3 · The math, with its limitations marked

The standard machinery is still useful — provided you remember which face it is measuring and which it is silent on. Three formulas, three distinct purposes, none of them universal.

Expected return: the planning number

Expected return tells you what to plan around. It is silent on the path you will take to get there. A stock that went up 40% last year did not have a 40% expected return going in. It had whatever expected return the market priced it at — probably much lower — and you observed one draw from the distribution.

Volatility: the spreadsheet measure

Volatility is useful because it is simple, symmetric, and lets you compare very different investments on the same scale. The S&P 500 sits around 15–18% in most regimes. A single stock often has 25–40%. Cryptocurrencies can exceed 80%. Treasury bills are effectively zero. None of those numbers tell you whether the asset can go to zero, whether your money is locked up for a year, or whether the spreads will widen exactly when you need them tight.

Drawdown: the human measure

The S&P 500 has experienced ~50% drawdowns in 1929, 1973, 2000–02, and 2007–09. Every generation gets one. In 2008 the index fell ~57% peak-to-trough; a portfolio worth $\$1,000,000$ went to $\$430,000$ and didn’t see the prior peak again for four years. Anyone who could not tolerate that — emotionally or financially — sold at the bottom and locked in the loss. Volatility lives in spreadsheets. Drawdown lives in bedrooms at 2 a.m.

4 · The Sharpe trap

Sharpe is the most cited number in finance and the most dangerous one. It compresses excess return and volatility into a ratio, then ranks managers by it. The compression silently assumes that volatility is the only risk that matters. It is not.

• Sharpe is blind to ruin. A strategy that pays 1% a month for sixty months and then loses everything in month sixty-one has a great Sharpe up to month sixty. (See: every short-vol fund.)
• Sharpe is blind to illiquidity. A strategy that smoothly marks itself to model can post a high Sharpe entirely because of stale prices. (See: many private credit funds, 2021–2024.)
• Sharpe punishes upside volatility. A strategy whose worst month is +5% and whose best month is +30% has the same Sharpe as one whose worst month is −12% and whose best is +13%. The investor experiences these two histories very differently.

Sharpe is a useful summary among strategies that share the same risk type. It is a misleading comparison across strategies whose risks live in different dimensions. The right question is never “which has the highest Sharpe.” It is “which is being honest about the risk it is paying me to bear.”

5 · Diversification — and where it stops working

Diversification is the only free lunch finance serves. Combining assets whose returns do not move perfectly together reduces the portfolio’s volatility more than it reduces its expected return. For two assets with equal volatility $\sigma$, equal weight, and correlation $\rho$:

When $\rho =0$, portfolio volatility drops 29%; when $\rho =-1$, it goes to zero. Harry Markowitz won a Nobel Prize for this in 1952. The catch is that diversification is, itself, a defence against one face — volatility — and is silent on the other three. It does not protect you from ruin (the FTX customer was diversified across five tokens on FTX), nor from illiquidity (the gates close on every fund at once), nor from drawdown during a correlation breakdown.

Correlations between risky assets are not stable; they rise sharply during crises. A portfolio of equities, credit, and real estate looked diversified in calm 2007 markets and fell together in 2008. The academic term is correlation breakdown. The practitioner term is “the diversification you paid for vanishes exactly when you needed it.” The only assets reliably uncorrelated to equity in a crash are government bonds (sometimes) and cash (always) — and you pay for that insurance in foregone return during every non-crisis year.

6 · The naming test

Before any investment, write down — in one specific sentence each — the answers to these four questions. If you can answer all four, you are investing. If you can only answer one or two with specifics, you are speculating, however good the Sharpe on the part you can answer.

• What volatility am I taking? Is the day-to-day move inside what I will not panic about? “A bit” is not an answer. “Up to 4% in a single day” is.
• What drawdown am I taking? What is the worst plausible peak-to-trough this asset has historically endured? Can I survive that without forced selling? Plan for the historical worst, not the recent average.
• What illiquidity am I taking? If I need this money in six months, can I get it without a forced-sale discount? What does the small print say about gates, lock-ups, notice periods, settlement delays?
• What ruin risk am I taking? What chain of events would take this asset to zero permanently? Counterparty failure? Leverage call? Concentration? Fraud? Name the chain. If you cannot, assume one exists and you haven’t found it yet.

7 · The upshot

There is no risk-adjusted return. There is only return adjusted for which kind of risk you bore. Volatility is what spreadsheets fear. Drawdown is what humans fear. Illiquidity is what fund gates exist to ration. Ruin is what you actually need to avoid. They are four different things, four different graveyards, and four different prices.

Diversify against volatility. Size positions for the drawdown you can survive. Read the lock-up clause before you wire the money. And — always, always — name the failure chain that takes you to zero, before you take the position. Doing those four things, deliberately, is the entire framework. The rest is implementation.