Five Cases in Creating and Destroying Value

On 22 July 2014 the Senate Permanent Subcommittee on Investigations held a hearing on a structure most investors had never heard of. From 1998 to 2013, its report said, Deutsche Bank and Barclays had sold 199 "basket options" to more than a dozen hedge funds, which used them for more than $100 billion of trades. The largest user was Renaissance Technologies. Medallion's own algorithms traded accounts that the banks owned, often holding positions for seconds; Medallion held an option on each account's value, exercised it after more than a year and reported the gains as long-term capital gains. The options had produced about $34 billion of trading profits for the firm, had "likely avoided over $6 billion in taxes" and had allowed leverage of up to 20 to 1 in one case, against the 2 to 1 a normal brokerage account allowed (PSI report, 21 July 2014). The banks earned more than $1 billion in fees. In September 2021 current and former Renaissance executives agreed to pay up to about $7 billion to settle the dispute with the Internal Revenue Service, and Jim Simons separately about $670 million (reported); nobody was charged with a crime.

The hearing is a useful place to begin this chapter because it shows something that the famous return figure hides. Medallion's edge in prediction was real, the subject of Chapter 8. But the number that reached its owners was also the product of how much the edge could be levered and how much of it the tax authorities would let them keep: survival and execution, in this book's terms, arranged by contract. Every case in the book has the same property. A trading result is the product of four factors, not a measure of one, and the five cases that the earlier chapters have told in parts — Renaissance, Citadel Securities, Long-Term Capital Management, the Turtles and XIV — are the clearest examples of each factor creating value and of each factor destroying it.

This chapter reads the five side by side. It does not retell them; each has a home earlier in the book, and the companion holds their profiles and moves. It asks of each the question that closes every chapter: which factor carried the result, and which ended it? It then shows, with a stylized simulation, why the records that the market rewards over five years can rank the same strategies in the opposite order to their results over twenty, and closes with what the five have in common.