Mathematician Edward O. Thorp used probability to find an edge in blackjack and later applied related risk controls to securities, helping connect casino mathematics with the development of market-neutral investing.
In a 2003 essay, Thorp said he began examining blackjack in 1958 after questioning analyses that assumed each round began with a freshly shuffled deck. He argued that cards already played changed the makeup of the remaining deck and a player's expected return.
Thorp estimated that a one-deck model involved more than 33 million possible card compositions. He wrote that an IBM 704 produced the calculations in about two hours of computer time. The Proceedings of the National Academy of Sciences published his paper “A Favorable Strategy for Twenty-One” on Jan. 15, 1961. His book Beat the Dealer followed in 1962, according to UC Irvine records.
Thorp later recalled winning $11,000 during 20 hours of full-scale betting on a 1961 casino trip, using a $10,000 bankroll. He said card counting gave a small group of skilled players an advantage, while many other players still had none.
Thorp summarized his approach to betting in a 2003 essay: “The magnitude of your bets should increase with expectation and decrease with risk.” He carried a related focus on probability and risk into securities markets, where he and stockbroker Jay Regan launched Convertible Hedge Associates in 1969. UC Irvine's archive records that the firm took the name Princeton Newport Partners in 1974. Thorp described it as a market-neutral derivatives fund, with convertible-bond hedging among its core strategies.
In a 2017 interview on The Meb Faber Show, Thorp recalled that the fund had three losing months out of 230, each down 1% or less, and annualized returns of a little more than 19% before fees. Those were his recollections of the fund's performance.
Thorp contrasted blackjack, where he could calculate odds from the cards, with markets, where estimates relied on a single historical record. He also said competition could shrink an investment edge.
