Famous trades and blowups, read the way a risk desk reads them. History is this museum’s second dataset (the first is my own tested-and-retired strategies): the trades that made fortunes, and the blowups that made modern risk practice. Each plaque below is dissected for the transferable mechanism, not the story, and usually ends with the rule the pipeline enforces because the episode happened. None of this is advice; every setup here is famous precisely because it was rare.
Two rooms in the Library: Reading Room · Case Studies
The Risk Museum: I test trading-strategy ideas the way a quant desk would, on an engine I built, and I publish the ones that fail my gates. New here? Start at the entrance →
I. The Asymmetry Gallery
bounded loss, unbounded gain, a counterparty who must act
The Tulip Wing shows what crowds do to prices. This gallery shows the opposite: trades that made history not by predicting better, but by finding structural asymmetry. Study the anatomy, not the names.
Soros & Druckenmiller vs the pound
The UK had pegged sterling inside Europe’s exchange-rate mechanism at a level its economy couldn’t support. The peg capped how far the pound could rise; devaluation was unbounded. Reported profit when the peg broke: over a billion dollars.
Anatomy: capped downside (the band), open upside (devaluation), and a forced counterparty. The Bank of England had to defend the peg with finite reserves against the whole market. Not a prediction. A payoff shape.
Burry, Paulson & the CDS trade
Credit default swaps on subprime mortgage bonds cost a small running premium and paid face value on collapse. The bet could be wrong for years at a known, budgeted cost, and right exactly once.
Anatomy: premium-bounded loss, ~10-100x payoff, and a structural mispricing (models assumed house prices don’t fall nationally). The hard part wasn’t the idea. It was carrying the negative carry until being right. Asymmetry pays the patient and breaks the leveraged.
The Volkswagen corner
Porsche quietly accumulated control of VW shares and options until the free float was smaller than the shares short-sellers owed. When disclosed, shorts had to buy stock that barely existed; VW briefly became the world’s most valuable company.
Anatomy: the buyer wasn’t willing but obligated. Covering is mandatory. Forced flows are the strongest force in markets; my own surviving signal (dealer hedging) is this same anatomy in miniature, every day.
Simons & Medallion
The best-returning fund in history reportedly made its fortune with no hero trades at all: thousands of small, fleeting statistical edges, sized tightly, compounding at extraordinary rates for three decades.
Anatomy: IR = IC × √breadth. A tiny edge taken thousands of times beats a large edge taken once, and it never bets the firm. The greatest “big bet” on this wall is the standing refusal to make one. This museum’s pipeline is built in this shape, not Soros’s.
II. The Blowup Wing
the failures that built modern risk practice
Risk management is a science whose experiments are run by other people, at scale, involuntarily. These five episodes are the curriculum: each one broke an assumption the industry didn’t know it was making, and each one left a rule this museum’s machine now enforces on itself.
Long-Term Capital Management
Nobel laureates, genuine edges in convergence trades, levered ~25x and more on the assumption that historical correlations bound the future. One sovereign default later, the positions were right “eventually” and the fund was gone immediately.
Anatomy, inverted: asymmetry flipped. Bounded gains (spreads converge a little), unbounded loss (leverage meets a regime break). Every exhibit above had capped downside; LTCM capped its upside and levered the rest. Markets can stay irrational longer than you can stay solvent. This plaque is why the museum audits regimes.
The Quant Quake
Over three days in August 2007, quantitative equity funds suffered enormous losses while the overall market barely moved. One large fund began liquidating its factor portfolios; every other quant holding the same value-and-momentum book bled with it, and their de-risking fed the spiral. By Friday prices had largely snapped back. Anyone forced out at the bottom locked in the loss permanently.
Anatomy: crowding. Dozens of “independent” models, trained on the same history, had converged on the same positions, so the true exposure was other people’s leverage, which appears on no position sheet. The rule it left behind: crowding is a factor to be measured, not a surprise to be survived. That is layer five of reading a PM’s book, and the reason a 13F ownership-overlap study sits in this museum’s own queue.
Volmageddon
XIV, a note that profited from calm markets by shorting volatility futures, had years of smooth gains, until a single afternoon spike in the VIX forced its own end-of-day rebalancing to buy volatility back into a thin market, amplifying the very move that was killing it. It lost roughly 95% in a day and was terminated.
Anatomy: the strategy was large relative to the market it traded, so its mechanical response to a loss made the loss bigger (reflexivity). Note what did not fail: the premise that selling insurance earns a premium remained true; the structure died. The rule: know whether your size moves your own trigger, and never confuse the survival of an idea with the survival of its wrapper. Every exhibit in this museum declares its capacity assumption for the same reason.
GameStop
Short interest in one stock exceeded its available float. Retail buyers, coordinating in the open, bought stock and call options; dealers hedging those options bought more stock; shorts covering bought more still. Funds on the short side lost billions being, arguably, right about the valuation.
Anatomy: forced flows again. Shorts and option dealers were both obligated buyers, and obligation beats opinion at any horizon. Being right is not a risk plan when the position’s loss is unbounded and the counterparty must act. The rule: short-side positions are sized by the exit, not the conviction. Measuring obligated dealer flow daily is the anatomy of this museum’s own surviving signal.
Barings
Barings was the archetype of the merchant bank, a house whose real asset was its name and whose word moved capital on a phone call. In 1995 a single trader in Singapore ran unauthorised futures positions while also controlling the back office that was supposed to check him. Losses hidden in an error account reached roughly £827 million, more than the bank’s entire capital, and the 233-year-old firm was sold for one pound.
Anatomy: the maker and the checker were the same person. Reputation is an asset that compounds over centuries through discipline and is destroyed in weeks by its absence, so a long record is an argument for controls, never a substitute for them. The rule this museum enforces because of it: the agent that proposes an idea may never judge it, and no single unsupervised process may put the whole enterprise at risk.
Archegos
A single family office built concentrated positions in a handful of stocks through total-return swaps spread across several investment banks. No bank could see the aggregate; each saw an exposure that looked survivable alone. When the stocks fell, simultaneous margin calls forced block liquidations and roughly $10 billion of combined losses across the street.
Anatomy: aggregation failure. Risk that isn’t summed doesn’t disappear. It compounds quietly until it is summed by the market, all at once. The rule: one ledger, everything in it. It is why this museum keeps every hypothesis, result, and rejection in a single state store, and why its multiple-testing penalty counts a family’s whole attempt history. Each test looking fine alone is exactly how Archegos looked to each bank.
III. The Antiquities Wing
the oldest instruments in the collection
Three numbers run the machinery of this museum: π, e, and i. Two of them (π and e) were discovered as limits of an honest iteration; the third (i) was discovered by trusting an honest calculation straight through impossible territory. That is also this museum’s entire method, stated twice.
Archimedes bounds π
Archimedes never computed π. He trapped it: a 96-sided polygon inside the circle, another outside, and the honest statement that the truth lives between them (3 10/71 < π < 3 1/7). Each doubling of the polygon’s sides tightened the bounds. He published the interval, not a point.
Where it hangs in this museum: π sits inside the normal density, so every t-statistic and standard error on every plaque here carries it. But the masterpiece is the method: my survivorship-biased Sharpes are declared as upper bounds; my bootstrap publishes the 5th-to-95th interval, never the point; and the research loop is polygon-doubling. Precision is bought by iteration count, not by cleverness per iteration. Grinold’s law says the same thing about breadth: IR = IC × √breadth. Edge grows like the square root of the number of honest, independent attempts.
Jacob Bernoulli finds e
Bernoulli asked what happens to money compounded not yearly but continuously: (1 + 1/n)n does not run to infinity. It converges to 2.71828… The same man proved the Law of Large Numbers: frequencies converge to probabilities, but only with enough independent trials.
Where it hangs in this museum: everywhere money grows. Log returns are continuous compounding; the museum’s benchmark line ($100k in an index fund) is e doing its quiet work over twenty years. The dark twin is volatility drag: compound growth ≈ average return minus half the variance, which is how exhibits with positive average returns still turned $100k into $93k. And his Law of Large Numbers is the license for my 10,000-draw bootstrap, with a plaque-sized warning he would have endorsed: it holds for independent draws, and market crises are precisely when independence fails.
Bombelli trusts the algebra
Cardano’s formula for the cubic worked beautifully until, on equations with perfectly real answers, its middle steps demanded the square root of a negative number. Cardano called such quantities “as subtle as they are useless.” Rafael Bombelli did the braver thing: he trusted the algebra over his intuition, carried the impossible numbers through the calculation by the ordinary rules, and watched the imaginary parts cancel, leaving the true, real answer standing. Centuries later the meaning arrived: i is rotation (multiplication that turns instead of stretches), and in 2021, experiments published in Nature showed quantum mechanics doesn’t merely use complex numbers; it requires them.
Where it hangs in this museum: my options laboratory prices through the Heston model’s characteristic function. The code multiplies by i and takes the real part at the end. Bombelli’s maneuver, executed on market data: pass honestly through the complex plane, return with a real, checkable price. Engineering adds the darker lesson: a system’s stability is read off pole positions in the complex plane, and when a pole crosses the axis, feedback amplifies until the system destroys itself. The Volmageddon plaque in my Blowup Wing is exactly that crossing, in market clothing. And the method lesson is the museum’s own creed: when the machinery’s intermediate steps look impossible, the discipline is to keep following the rules. The honest calculation lands somewhere real.