Investing Classics: Options & DerivativesAsymmetry, Volatility, and the Cost of Insurance

Nassim Taleb, Antifragile (2012); "optionality" is the book's central mechanism.

The buyer pays a premium to lock the loss; the seller collects a premium and assumes an obligation with, in theory, no ceiling. That is the whole structure. What matters is that the asymmetry is priced: the premium is precisely the market's quoted price for "bounded loss, unbounded gain," and over time it is enough to offset the asymmetry itself. So to win as a buyer you must be right about direction, magnitude, and timing at once.

After Lehman failed in September 2008, Berkshire put $5 billion into Goldman Sachs perpetual preferred stock (10% annual dividend) and also received warrants to buy $5 billion of common at a $115 strike. The preferred was current cash flow; the warrants were the asymmetry — worthless if Goldman did not survive, fully open on the upside if it did. The two sides settled net in 2013 and Berkshire took roughly 13.1 million shares without paying cash. The precondition: very few parties were willing to put up $5 billion in cash at that moment.

The appeal of asymmetry is systematically overrated. The great majority of out-of-the-money options expire worthless, and the move that "should have paid off" is routinely missed because the expiry was wrong — time sits structurally on the seller's side. By Cboe's own figures, same-day-expiry (0DTE) contracts have recently accounted for roughly half of S&P 500 index option volume: they carry almost no time value and amount to a bet on one day's move.

• What percentage of the underlying position is the premium? Annualized, would I still pay it?
• Does my view include magnitude and timing, or only direction?
• Could trimming the position and holding cash achieve the same goal more cheaply?

Black & Scholes published their model in the Journal of Political Economy in 1973. Thirty-six years later, two veterans of options quant work wrote down a caution about models.

The model splits an option's price into a few observable inputs and one unobservable quantity: future volatility. The observables are the same for everyone, which makes the options market fundamentally a market in volatility — and explains how you can be right on direction and still lose: you bought when implied volatility was high, the event resolved, volatility collapsed, and the Vega loss ate the Delta gain. After the October 1987 crash, implied volatility on low-strike puts became permanently higher than on other contracts. That volatility skew is the market's verdict on the model: the tails are fatter than it assumes.

On February 5, 2018, the VIX closed at 37.32 against the prior day's 17.31 — a one-day rise of about 116%, the largest single-day percentage gain in the index's history. XIV, a short-volatility exchange-traded note issued by Credit Suisse with roughly $1.9 billion in assets, saw its indicative value drop about 96% after the close and was liquidated on February 21. It had nearly doubled during 2017. Its holders did not bet the wrong way; they bet the right way for years and returned the principal in a single day.

Yet "never sell volatility" is equally wrong. Implied volatility runs persistently above realized volatility, and that gap is an insurance-like payment to whoever carries the tail risk. XIV's problem was not direction but structure — daily reset, embedded leverage, and a contractual acceleration once a single-day drop crossed a threshold — which turned a strategy with positive long-run expectancy into one certain to reach zero. Earning the premium requires surviving the spike days.

• Where does implied volatility sit in its own history? Am I paying for an event already priced?
• If direction is right but volatility falls 30%, does the trade still make money?
• Is the equity curve "too smooth"? Smoothness usually means risk was pushed into the tail.

Buffett's 2008 letter, explaining why Berkshire sold very long-dated index puts.

A covered call (holding stock and selling a call against it) cuts off the right tail of the return distribution in exchange for a fixed premium. It is never right for a core compounding asset, because the entire return from compounding lives in that right tail — Cboe's BXM buy-write index has, since 1986, produced annualized returns close to the S&P 500 at visibly lower volatility, yet fell far behind through the strong bull market of the 2010s. A protective put is insurance: clean in logic, expensive in practice, and correctly used for a defined window and a defined event.

Between 2004 and 2008 Berkshire sold European-style index puts with about $37.1 billion in notional value and 15- to 20-year terms, collecting roughly $4.9 billion in premiums. When markets collapsed in 2008, the accounting liability produced billions in unrealized losses on the books. But the contracts carried two decisive terms: European exercise (settlement only at expiry, so interim paper losses created no delivery obligation) and almost no collateral posting. The contracts ran off around 2017, mostly expiring worthless. The protection was not accurate judgment; it was terms that spared the position from having to settle at the worst moment.

The same position under daily mark-to-market collateral terms would not have survived 2008. AIG's Financial Products unit sold credit default swaps on exactly those terms: a ratings downgrade in September 2008 triggered cascading collateral calls, liquidity drained within days, and the US government ultimately committed up to about $182 billion. What killed it was not the terminal judgment but the path — any contract that requires you to post cash on the most frightening days demands it exactly when you are least able to pay.

• Is this holding a core compounding asset or a position I would happily sell at that price? Never write calls on the former.
• What does the insurance cost annually as a share of the portfolio? Would five straight years still be worth it?
• Under the worst path (not the worst outcome), could I be forced to liquidate at the bottom?

Buffett's 2002 letter, written the year after Enron collapsed and six years before the financial crisis.

Note that the man who wrote that has used derivatives throughout (see Card 3). What he objects to is not the tool but the three things it carries at once: leverage hidden inside the notional, counterparty risk, and the closed loop of mark-to-model pricing (with no market price you value it on your own model, and the model is tuned by people with a bonus at stake). So the test has to land on total exposure: selling puts to collect premium — whether it is called "yield enhancement" or "willingness to buy lower" — creates a new downside exposure.

In 1993, MGRM, the US subsidiary of Germany's Metallgesellschaft, signed ten-year fixed-price fuel supply contracts and hedged them by rolling short-dated crude futures. The direction was sound, but the maturities were badly mismatched: oil fell that year, paper losses on the short-dated leg immediately triggered margin calls, while the gains on the long-dated contracts would only arrive over the following decade. Unable to keep funding the calls, the company was forced to liquidate, realizing roughly $1.3 billion in losses. A logically correct hedge became fatal speculation through a mismatch in the timing of cash flows.

Conversely, "using no derivatives" is not a neutral stance either: choosing not to hedge currency or rates is an actively held exposure, one that appears on no statement and is therefore never examined. Damodaran makes the point repeatedly — the real question is not whether to use derivatives but which risks you are paid to bear. For most individual investors, the risks worth hedging can be handled by cutting position size and holding cash. The most common outcome of derivatives in individual hands is trading a position problem for a liquidity problem.

• After this trade, if the underlying falls 30%, do my total assets improve or worsen? Worsen means speculation.
• What multiple of my net worth is the notional exposure? Am I thinking in margin figures instead of exposure figures?
• Does the hedge's cash-flow timing match the timing of the risk it hedges (the MGRM lesson)?