Infor The: Dan Tu D B A L L Reviews The Life of One of Quantitative
Infor The: Dan Tu D B A L L Reviews The Life of One of Quantitative
Infor The: Dan Tu D B A L L Reviews The Life of One of Quantitative
T
Ed Thorp cracked blackjack, used the first wearable he year is 1938. The place, about 45
miles out of Chicago. On the steps of a
computer to beat roulette, started the world’s first market a boy of not quite six faces off
against a perplexed looking local man
quantitative hedge fund, anticipated the Black- who holds a heavy tome belonging to
the kid, and studies it with some skepti-
Scholes formulae by five years, and has maintained cism. “Egbert 802 to 839,” the boy
begins, quietly, and in a considered
consistently excellent returns through nearly forty tone somewhat beyond his years he con-
tinues, “Ethelwulf 839 to 857, Ethelbald
years in hedged portfolios and derivatives. 857 to 860, Ethelbert 860 to 866,
Ethelred I 866 to 871, Alfred the Great
Dan Tudball reviews the life of one of quantitative 871 to 901, Edward I 901 to 924,
finance’s great heroes, and speaks to the Ethelstan 924 to 940, Edmund I 940 to
946, Edred 946 to 955, Edwig 955 to 959,
In for the
24 Wilmott magazine
Wilmott magazine 25
ED THORP
1014, Canute the Great 1016 to 1035, shopping expedition the young Thorp’s physics, electronics, astronomy and but I remember seeing pictures of
Harold 1036 to 1039, Harthacnute 1039 parents and friends were sat down try- mathematics. Being most interested in homeless people in the newspapers, tat-
to 1042, Edward III 1042 to 1066, ing to induce the child to speak - still a chemistry he sat for the All-Southern- tered clothes and that sort of thing. It’s
Harold...” The man’s face sets in disbe- popular pastime in the Thorp house- California high school chemistry test, something that people of that era
lief as the boy continues his litany; the hold. Some people stepped out of the despite being a few years younger than remember very vividly. Saving every-
book he refers to is A Child’s History Of elevator and someone asked, “Where’s other students sitting for the exam. He thing, And that had an impact on me, I
England by Charles Dickens - every the man gone?” Thorp recounts the came fourth in that part of the state, was very frugal for the first twenty-five
entry in the chronology of monarchs moment down the line from Newport and was very proud of that result but he years of my life and this allowed me to
recited from this boy’s memory to per- Beach “Oh, he’s gone to buy a shirt,” so recalls that the reason why he had only make it through university on very lim-
fection. The boy’s father is equally everybody’s eyes popped out and the achieved that position rather than com- ited means.” With money in short sup-
taken aback. Thorp Sr had long known next question was “‘Where has the ing first was down to a new section ply, and with a burgeoning interest in
his son was a prodigy, but this display is woman gone?’ and I answered ‘Oh, requiring slide rules. Ed only had a 10- expensive experimentation, Ed would
shocking. Only moments before the She’s gone to the bathroom to do pee- cent slide rule, which was “a piece of deliver newspapers at three in the
morning in order to fund his science.
A child with a preternatural gift for reflection, and When speaking to Thorp about
these formative years a vivid picture is
26 Wilmott magazine
ple. Until I was nine I believed that
everything I saw in print was true I
found it impossible to believe other-
wise. Until, that is, I saw two newspa-
pers with conflicting information and
that particular naivety disappeared
rather rapidly after that.”
Wilmott magazine 27
ED THORP
I
Vivian and had achieved his Ph.D at n spring 1955 Ed Thorp was in his second year of graduate experiments, to try to predict the moment when the spinning
UCLA, work on roulette had resulted physics at UCLA. At tea time one Sunday he got to chatting ball would fall into the waiting pockets. From their model they
from some idle banter with friends on with colleagues about how to make ‘easy money’. The conver- were able to predict any single number with a standard devia-
how to make easy money, back in 1955. sation turned to gambling, and roulette in particular. Was it possi- tion of 10 pockets. This converts to a 44 per cent edge on a bet
This time round it was a trip to Las Vegas ble to predict, at least with some exploitable degree of accuracy, on a single number. Betting on a specific octant gave them a
for a cheap, non-gambling, vacation the outcome of a spin of the wheel? Some of his colleagues, the 43 per cent advantage.
that got Ed thinking. At the time the pre- ones in the know, were certain that the roulette wheels were man- It is one thing to win on paper, or in the comfort of a base-
vailing assumption was that none of the ufactured so precisely that there were no imperfections that could ment. It is quite another to win inside a noisy casino.
major gambling games allowed for sys- be discerned, never mind exploited. But Ed’s counter to that was From November 1960 until June 1961 Ed and Shannon
tems. The accepted thought was that simple, if the wheels are so perfect you should be able to predict, designed and built the world’s first wearable computer. The
because most games depended upon using simple Newtonian principles, the path of the ball and its twelve transistors, cigarette-pack sized computer was fed data
independent trials processes – i.e every final resting place. by switches operated by their big toes. One switch initialized
spin or dice roll was unaffected by those Ed got to work in the late 1950s, playing around with a the computer and the other was for timing the rotation of the
that preceded it – then there was no way cheap miniature roulette wheel, filming and timing the revolu- ball and rotor. The computer predictions were heard by the
a mathematical system would allow you tions. He met up with Claude Shannon, the father of informa- computer wearer as one of eight tones via an earpiece. (Ed and
to numerically track outcomes and rea- tion theory in 1959, originally to discuss his blackjack results, Shannon decided that the best bet was on octants rather than
sonably predict future outcomes. Unless but the conversation soon turned to other games and roulette single numbers since the father of information theory knew
you used rigged dice or had some infor- in particular. Shannon was fascinated. Shortly afterwards they that faced with n options individuals take a time a+b ln(n) to
mation on the croupier, you may as well met up at Shannon’s house, the basement of which was make a decision.)
bring along a rabbit’s foot as a calcula- packed with mechanical and engineering gadgets, the perfect This computer was tested out in Las Vegas in the summer
tor. Ed had previously concluded (in playground for further roulette experiments. of 1961. But for problems with broken wires and earpieces
1955) that roulette was an exception to Ed and Shannon together took the roulette analysis to falling out, the trip was a success. Similar systems were later
this rule, because he wasn’t using a greater heights, investing $1,500 in a full-size professional built for the Wheel of Fortune which had an even greater edge,
numerical system and instead relied on wheel. They calibrated a simple mathematical model to the an outstanding 200 per cent.
the physical properties of the mecha-
nism. ‘Fortune’s Formula’ at the American
Prior to his trip to Vegas Thorp had
been given a paper, published in the
The important observation he Mathematical Association. After the
trip to Nevada Ed had tested some of his
Journal of the American Statistical
Association written by US Army mathe-
took with him was that he had own theories on MIT’s own IBM 704
mainframe computer (a far cry from 10-
maticians (Roger R. Baldwin, Wilbur E.
Cantey, Herbert Maisel, and James P.
been losing at a far slower rate cent slide rules!) and duplicated in a few
hours what would have taken over
McDermott) on basic strategy in the
game of blackjack. The contention of the than others on the table 10,000 man years of labor on a hand
held calculator. It was these findings
paper was that the house edge on black- that he presented in Washington. At the
jack could fall as low as 0.0062 (some- end of the presentation all of Ed’s
what later corrected by them to .0032), losing at a far slower rate than others on Independent trials processes were not a mimeographed copies of his report
Ed made himself a little reference card the table and the realization that he factor in this game and thus, Ed rea- were snapped up as the 300 or more
to take to the table, purchased ten could modify the methodology. soned, all you needed was a decent fre- mathematicians in the room rushed
bucks’ worth of chips and prepared to Like roulette, Blackjack was in fact quency of favorable situations and the podium. When Thorp had arrived
test the methodology. Once at the table also an exception to the rule that gam- adjustments in the betting spread in in Washington he was already aware
he played the game for about twenty bling games couldn’t be beaten by fair order to get the edge. Ed, whether he that the media had whipped up a small
minutes – never having played it before, means. At that time, when a card was realized it or not was on the edge of storm in advance. An AP reporter had
and this being the first time he’d set foot dealt it was put aside, thus shifting the something himself. been leafing through the Association’s
in a Casino – eventually losing the ten composition of the now depleted deck In the fall of 1961 Ed was CLE Moore abstracts prior to the meeting and
dollars, but the important observation in a set manner, a manner that would instructor at MIT and went to called Ed, this resulted in a story in the
he took with him was that he had been either favor the player or the casino. Washington to present a paper entitled Boston Globe the next day. Ed recalls that
28 Wilmott magazine
ED THORP
Wilmott magazine 29