Triam

the stories money tells

[Triam @triiiam]

https://t.co/4a0fNKJZDq

[Triam @triiiam]

@velesxbt thats insane lmao

[Triam @triiiam]

@Frandeeer edge that survives contact is the only edge, nice post

[Triam @triiiam]

A group of MIT students beat the casinos out of millions with nothing but arithmetic. No system, no hunches, no psychic streak. They just counted the deck better than the professionals whose entire job was to stop them, and the house only won back by changing the rules.

The lesson underneath it is the one nobody selling a newsletter wants you to hear. The amateurs with the math beat the professionals with the credentials.

That is easy to accept about a blackjack table. It is harder to accept about the market, where the people who sell you forecasts are quietly betting you will never check whether they are any good.

So I checked. 1,197 live prediction markets. A mob of anonymous gamblers, pricing with real money on the line, forecast reality more accurately than the experts you pay to do it. Same lesson, no casino required.

My 1,197 markets and the code are in the article. Run it yourself.

[Triam @triiiam]

@carbonyxxx exactly, most people just never bother to learn it

[Triam @triiiam]

A group of MIT students beat the casinos out of millions with nothing but arithmetic. No system, no hunches, no psychic streak. They just counted the deck better than the professionals whose entire job was to stop them, and the house only won back by changing the rules.

The lesson underneath it is the one nobody selling a newsletter wants you to hear. The amateurs with the math beat the professionals with the credentials.

That is easy to accept about a blackjack table. It is harder to accept about the market, where the people who sell you forecasts are quietly betting you will never check whether they are any good.

So I checked. 1,197 live prediction markets. A mob of anonymous gamblers, pricing with real money on the line, forecast reality more accurately than the experts you pay to do it. Same lesson, no casino required.

My 1,197 markets and the code are in the article. Run it yourself.

[Triam @triiiam]

https://t.co/4a0fNKJZDq

[Triam @triiiam]

@localminimaa the sandwich was the memo lmao

[Triam @triiiam]

@caspr_exe saving this for later, great post

[Triam @triiiam]

@Rossst_03 Feynman never disappoints, absolute legend

[Triam @triiiam]

@carbonyxxx Society of Mind is a classic, glad someone remembers it

[Triam @triiiam]

@VoltexGar Tao is a genuine legend

[Triam @triiiam]

The most famous statistician in England went to a county fair in 1906 to prove ordinary people are stupid.

He collected 787 guesses of an ox's weight from butchers, clerks and farmers. The ox weighed 1,198 pounds. The crowd said 1,207. Nine pounds off, closer than the cattle experts standing in the same tent. He was 84, it broke everything he believed, and he published it anyway.

His name was Francis Galton. For a century his ox has been a cute footnote in a statistics textbook.

I stopped treating it as a footnote. I re-ran his experiment on 1,197 real prediction markets, real money instead of sixpence cards. The crowd is still within about two points of reality. In the messy middle, where the experts earn their fees, it lands within half a point.

It has exactly one blind spot, and it is the mirror image of the flaw a century of racetrack studies promised I would find: this crowd undercharges for lightning. The rare event pays out almost twice as often as its price. Then I ran the machinery that kills findings like that. It executed all twenty tests but one, which froze on the exact line between a discovery and a coincidence.

The full test, the numbers, and the one crack still breathing are in the article. The code fits on one screen. Run it yourself.

[Triam @triiiam]

https://t.co/4a0fNKJZDq

[Triam @triiiam]

@0xRokko The "die inside" line isn't the warning. It's the retention metric

[Triam @triiiam]

@caspr_exe Forward deployed engineers are the tell

[Triam @triiiam]

@0chob The warning was the calmest part

[Triam @triiiam]

Nobody typed the order. Nobody clicked buy. And in less time than it takes to eat lunch, a firm that moved a huge share of America's stock trades bled out on the exchange floor while its own engineers watched, unable to reach the switch fast enough.

Forty-five minutes. Around 440 million dollars. The morning of August 1, 2012.

The firm was Knight Capital, one of the largest market makers on the planet, the quiet plumbing under everyday trades. That morning it pushed new software onto eight servers. A technician missed one. On that single server, a new flag switched on a piece of old code asleep since 2003, a function called Power Peg that was never built to run loose.

It woke and did exactly what it was written to do. Buy high, sell low, again, again, faster than any human could read the screen. Millions of orders poured into 154 stocks. Its own engineers understood the problem within minutes. That was the horror. Understanding changed nothing. Their attempt to fix it, rolling the code back off the servers, only spread the bad instruction wider.

By the time they killed it, the loss had swallowed the company's entire capital and then some. Days later Knight survived only on an emergency 400 million dollar rescue that handed most of the firm to outsiders. Within a year the name was gone, folded into a merger.

No villain. No hack. No greed. Just an instruction slightly wrong, running at a speed the human hand was never built to interrupt.

The post above is right that the danger was never malice. A system does not need to hate you to end you. It only needs a goal a little off from yours and the speed to reach the bottom before you reach the switch. Knight built a machine that obeyed perfectly, and perfect obedience to a flawed order was the whole catastrophe.

We keep bracing for the machine that turns on us. The one that ruins us will simply do as it is told.

[Casper @caspr_exe]

https://t.co/kMbJeWstjh

[Triam @triiiam]

@Di_Krass_ never knew the lottery arbitrage part

[Triam @triiiam]

@caspr_exe nobody asked about our friends, wild

[Triam @triiiam]

It took one 28-year-old three years to destroy a bank that had survived 233 years, the wars against Napoleon, and every panic in between. When it was over, the whole thing was sold for a single pound.

For most of those three years he was the firm's golden boy. London watched the profits pour out of its small Singapore office and asked nothing, because nobody interrogates the man who is printing money. What no one upstairs could see was a second set of books. Every winning trade was booked to the official account. Every loss was buried in a hidden one he had numbered 88888, where it sat and grew in the dark.

The trader was Nick Leeson, and for a while he was genuinely good. He was right often enough to stay a star. He just never let anyone see the trades that went the other way. Then the Kobe earthquake sent the Nikkei into freefall, and the losses he had hidden for years detonated in weeks. The final hole came to 827 million pounds, more than the entire bank was worth. The reason, he admitted later, was smaller than any theory written about him: he was too afraid to confess one mistake, so he hid it until it was too big to hide.

The post above is right that a high win rate is not safety. Leeson is that lesson wearing a suit. Barings did not collapse because he lost too often. It collapsed because it counted his wins and never once sized his worst loss. A track record tells you how good someone looks. It says nothing about what a single position they refuse to close can cost.

Being right is the cheap part. Surviving the trade you will not admit is the whole game.

[Frander @Frandeeer]

https://t.co/b81hwcYS19