The Man Who Solved Blackjack and Then Went Very Quiet

There is a particular kind of conversation that happens at the good tables, usually somewhere after the second bottle, and it always arrives at the same place. Somebody says they have a feel for it. A run, a streak, a sense of when the night is going their way. And everyone nods, because it is a nice thing to believe about yourself, and because the alternative is admitting you have no idea what is happening.

I have sat through that conversation more times than I can count. What I have never once heard at one of those tables is the name Edward Thorp, which is a shame, because he is the only person who ever actually settled the argument.

A mathematics lecturer, a borrowed computer, three pages

In the winter of 1961, a young mathematician at MIT published a paper in the Proceedings of the National Academy of Sciences. It ran to three pages. Volume 47, issue one, pages 110 to 112, and it carried the rather plain title A Favourable Strategy for Twenty-One

Three pages. That is the whole thing. You can read it in about eleven minutes and most of it is tables.

What Thorp had done was borrow time on an IBM 704, which in those days meant queueing for the machine and running your job overnight while everybody else slept, and use it to work out something nobody had bothered to establish properly. Blackjack, alone among the games on a casino floor, has a memory. The cards that have gone are gone. The deck that remains is not the deck you started with. And if you keep a rough track of what has left, there are moments, quite ordinary moments in an ordinary shoe, when the remaining cards favour the player rather than the house.

Not always. Not by much. But measurably, and repeatably, and that is a completely different animal from a feeling after the second bottle.

The paper landed at the American Mathematical Society meeting that January and the press picked it up almost immediately, which Thorp had not really expected. A year later he published Beat the Dealer, and the thing went off like a firework. It sold in the hundreds of thousands. Ordinary people, people who had never held a mathematics textbook in their lives, started carrying little strategy cards into casinos.

The bit everybody gets wrong

Here is where the story usually gets told badly.

The popular version says Thorp went to Vegas and won a fortune and the casinos threw him out. The first and third parts are broadly true. The middle part is the part people inflate, and Thorp himself has been fairly blunt about it over the years.

He went out to Nevada in 1961 with backing from a couple of men who had rather more money than he did, and he ran the system live over a few days. He won. He did not win the sort of number that changes a life. What he proved was that the maths held up outside the machine room, which was the entire point of going.

The response from the floor was swift and it was not subtle. Decks got changed. Shuffles came earlier. Rules were quietly rewritten in the middle of the night and then quietly rewritten back when the punters complained. Thorp was recognised, then followed, then barred, and eventually he took to wearing disguises, which is a genuinely funny image for a man whose entire contribution was a set of probability tables.

And then there is the other project, the one that sounds invented and is not. Thorp had been working with Claude Shannon, the man who more or less created information theory, on a small computer you could hide on your body to predict a roulette wheel. It was conceived in the mid-fifties, built in Shannon's basement, and tested in Las Vegas in 1961. Toe switches for input. A tiny speaker in the ear for output. Guinness now recognises it as the first wearable computer, which means the ancestor of the watch on your wrist was built to beat a casino, and the two men who built it walked away from the money.

That last part is the bit I keep coming back to. They had a working edge on a roulette wheel and they went home.

Nineteen years, three losing months

Because what happened next is the actual story, and almost nobody tells it, because it is not photogenic.

Thorp took the same apparatus of thought off the casino floor and pointed it at the stock market, which is a much bigger table with much worse dress sense. In 1969 he set up Princeton Newport Partners with Jay Regan, and it is generally regarded as the first market neutral hedge fund of the kind that everybody now runs.

The numbers on that fund are worth sitting with for a second, because they are not what people expect from a story that starts in a casino. Over nineteen years it compounded at roughly 15.1 percent a year net of all fees. Two hundred and twenty-seven winning months. Three losing ones. None of the three worse than one per cent.

Three bad months in nineteen years is not a hot streak. It is the exact opposite of a hot streak. It is what it looks like when somebody has removed almost all of the luck from the process and replaced it with arithmetic and an enormous amount of extremely boring work.

Thorp shut the fund in 1989 and has spent the decades since being, by every account, quite content and mostly out of view. He is in his nineties now. He still writes. He does not appear to want anything from anybody, which in a world of people announcing thier own success at volume is a kind of luxury all of its own.

What edge actually means, for people who already have some

I think the reason Thorp matters to a readership like this one has very little to do with cards.

Most of the people I know who have done genuinely well have an instinct that runs directly against him. They believe in the read. The gut. The deal that felt right in the room. And sometimes it was right, and they remember those with great clarity, and they do not keep a separate list of the ones that felt equally right and went nowhere.

Thorp's whole life is an argument that the feeling is not the asset. The asset is a small, measurable, provable advantage, applied over and over again with enough discipline that variance stops mattering. That is a deeply unglamorous proposition. It says the interesting part of your success is the least interesting part of your day.

It also says something rather uncomfortable about size. He was one of the early users of what is now called the Kelly criterion, the formula that tells you what fraction of your capital to commit given the size of your edge. And the honest answer that formula keeps returning is: much less than you want to. Bet too big with a real advantage and you still go broke. That applies to a shoe of cards, and it applies to a sales room, and it applies to the third apartment you were thinking about, and the maths does not care which one you are doing.

The game he walked away from is now in everybody's pocket

The strange coda to all of this is what the game itself became.

Thorp's edge worked because a physical deck has a memory and a physical dealer is slow. Modern online blackjack has mostly closed that door. Shoes get reshuffled continuously, the tables run at a pace no counter could track, and the whole thing is streamed from a studio rather than dealt in a room you can walk into. The advantage he found does not really survive the transfer.

What did survive is the arithmetic underneath, and the fact that the game still carries the thinnest house margin on the floor when it is played correctly. Swift Casino runs its live tables through Evolution alongside titles from Playtech, NetEnt and Games Global, low limit seats through to VIP tables, with a blackjack bonus of 100 per cent up to twenty-five pounds plus a hundred free spins on the welcome. That is a UK offer for UK players; it carries wagering conditions, and the terms shift about, so read them on the page rather than taking my word for any of it. Thorp's actual advice to ordinary players, which he has repeated for sixty years and which nobody wants to hear, was to learn basic strategy properly, set a number you are prepared to lose before you sit down, and treat that number as the whole of the entertainment budget. The man who found the edge spent most of his life telling people they did not have one.

The unglamorous half nobody photographs

There is one more piece of Thorp that gets left out, and it is the piece I find most useful.

Princeton Newport did not run on insight. It ran on process. Position records, reconciliations, a back office that knew exactly what was held and why, so that when something moved, nobody was guessing. The edge was in the paper, and the paper was in the filing. Genuinely dull, and completely load-bearing, because an advantage you cannot administer is not an advantage at all.

The same thing is true, at a much smaller scale, of almost every good small business I know. The charter broker. The gallery with two staff. The tailor with a waiting list and no system for the waiting list. All of them are one person's memory away from a bad week. And the reason they resist fixing it is nearly always the pricing, because most help desk software charges per seat, so the founder plus the one assistant plus the part-timer who covers August all get billed as though they were a department. Maxdesk charges per workspace instead of per person, with unlimited agents and unlimited tickets on the free tier, so a three-person operation can put every enquiry into one queue without anybody doing sums about whether the summer help is worth a licence. It is email only; there is no live chat, and the free plan carries Maxdesk branding on outbound mail with a rolling three-month history rather than the full archive. Which is a perfectly reasonable trade for never again sending two different answers to the same client.

Not a thrilling paragraph, I know. Neither was the back office in Newport Beach, and that one produced three losing months in nineteen years.

What he actually proved

The temptation with a story like this is to file it under gambling, which is where it usually gets filed, and that is precisely the mistake.

Thorp did not prove that casinos can be beaten. He proved something considerably more awkward, which is that almost everything we call intuition is noise, that the small number of real edges in the world are boring and narrow and quickly closed, and that the only reliable way to make one pay is to size it correctly and then repeat it untill the story stops being interesting.

He is ninety something years old and he seems perfectly happy. He beat the house, he beat the market, and the thing he chose to do with the rest of it was read, and think, and stay out of the papers. There are worse ways to spend an edge.