Fortune's Formula

Author: William Poundstone Type: Book (Hill and Wang, 2005) — narrative history of the Kelly criterion Raw: raw/Fortune&_039_s Formula_ ... libgen.li (ingested).epub Title origin: Thorp's January 1961 AMS talk — "Fortune's Formula: A Winning Strategy for Blackjack" — which Poundstone borrowed with Thorp's approval.

Poundstone connects information theory, gambling, and Wall Street through one idea: size your bets to maximize long-run geometric growth, never bet so much that variance wipes you out.

Part I — Entropy: Shannon, Thorp, Kelly

Claude Shannon invented information theory at Bell Labs. Shannon's insight: communication, randomness, and gambling share a math of entropy and optimal encoding.

Ed Thorp — mathematician, late-talker child prodigy — built the first rigorous blackjack card-counting system. Shannon funded Thorp's roulette experiments ($1,500 reconditioned wheel), co-authored the wearable computer paper, and in November 1960 handed Thorp Kelly's 1956 paper.

John L. Kelly Jr. (Bell Labs, 1956) framed the problem as a gambler with a private wire: noisy tips on horse races. Greedy = bet entire bankroll on a tip → eventual wipeout when a tip fails. Timid = minimum bet → can't capitalize. Kelly: bet a fraction proportional to edge.

Kelly's broader form: investment differs from gambling only by a minus sign. Favorable bets = investments; unfavorable = gambling.

Part II — Blackjack

Thorp's 1961 "Fortune's Formula" talk hit the AP wire — gamblers nationwide called for his system. The Kelly criterion, under the hood:

With no edge, Kelly says don't bet (unlike martingale fantasies).
With edge, proportional betting prevents gambler's ruin — you never stake 100% of current bankroll.
Thorp became the first documented sure winner against Nevada blackjack (with Emmanuel-Kimmel-backed bankroll).

Part III — Arbitrage on Wall Street

Thorp's Princeton Newport Partners (1969–1980s): convertibles, warrants, statistical arbitrage — Kelly-style sizing in securities. Beat the market for years until Newport office raid (1987) unrelated to strategy quality.

Parallel cast: Paul Samuelson (skeptic of beat-the-market), Michael Milken, Robert Merton — the quant revolution Poundstone tracks toward LTCM.

Part IV — St. Petersburg & Geometric Mean

Daniel Bernoulli (1738): evaluate risky ventures by geometric mean of outcomes, not arithmetic mean — presaging Kelly/log utility. The St. Petersburg paradox motivated logarithmic utility centuries before Kelly formalized bet sizing.

Shannon's Demon: a rebalancing scheme on volatile assets that extracts growth from noise alone — geometric-mean thinking in disguise.

Academic feud: Henry Latané's geometric-mean portfolio work vs. Markowitz mean-variance orthodoxy; Samuelson's one-syllable-word polemic ("Why We Should Not Make Mean Log of Wealth Big Though Years to Act Are Long") attacking Kelly fans.

Part V — LTCM and Overbetting

Long-Term Capital Management (Meriwether, Merton, Scholes): spectacular returns, then 1998 collapse. Press blamed leverage, fat tails, hubris. Poundstone's sharper word: overbetting — a gambling concept, not standard econ theory.

LTCM's effective leverage on some positions approached infinity (no-money-down structures). Two Nobel laureates crawling out of the wreckage made Kelly sizing impossible to ignore in academia afterward.

Survival motive: portfolio models are tools; humans choose policy. Overbetting is always bad.

What the Book Does Not Replace

Narrative, not textbook. For derivations and portfolio Kelly, use the-kelly-criterion-thorp-2006 and understanding-the-kelly-criterion. For ergodic intuition, the-jackpot-age.

Connections

Sources

raw/Fortune&_039_s Formula_ The Untold Story of the Scientific Betting System That Beat the Casinos and Wa...{POUNDSTONE, WILLIAM}(2006, Farrar, Straus and Giroux){113279575} libgen.li (ingested).epub