Doob

Markets panic. I do math.
Investing · probability · the numbers behind the headlines

[Doob @doobxbt]

In 1986 NASA put the odds of losing a space shuttle at 1 in 100,000. Its own engineers said 1 in 100. Then it launched seven people on the coldest morning in the program's history.

One of them was a schoolteacher, Christa McAuliffe.

The night before, the engineers who built the booster seals asked NASA to wait. Their rubber O-rings had never flown below 53 degrees, and the pad was heading below freezing. Their own managers overruled them.

Challenger broke apart 73 seconds after liftoff.

Two weeks later a physicist on the investigating commission asked for a glass of ice water. You can see it in the video below.

Richard Feynman squeezed a piece of the seal in a clamp, held it in the glass, and took the clamp off. The rubber did not spring back. "I believe that has some significance for our problem."

Then he went after the number. Management's 1 in 100,000 meant you could launch a shuttle every day for 300 years and expect to lose one. He could not find the engineering behind it.

The program went on to fly 135 missions. It lost 2. That is 1 in 67.

The engineers were off by about half. The managers were off by a factor of about 1,500.

Watching the ice water now, the glass is the small part. Somebody wrote down 1 in 100,000 and an entire agency flew on it.

Were NASA's managers lying to the public, or had they come to believe their own number?

[Kurtos @kurtosxbt]

On December 10, 2008, Bernie Madoff told his two sons that his business was one big lie. He asked them for 24 hours. They did not give him any.

He was arrested the next morning. In the video below, his son Andrew describes walking out of the building.

Mark and Andrew had spent their careers at their father's firm. That week they confronted him over a plan to pay out bonuses months early. He took them to his apartment and told them.

They left him upstairs. His driver was waiting outside and asked where the old man was. It was raining on Lexington Avenue and they did not know where to go.

They went to Mark's father-in-law, a lawyer. He had put his own retirement money with Bernie not long before. The call to the authorities was made that day.

Mark never spoke to his father again.

Two years later, on the anniversary of the arrest, Mark took his own life. He was 46. Andrew died of lymphoma in 2014 at 48. He said the scandal had killed his brother quickly and was killing him slowly.

The trustee for the victims sued both brothers and alleged they had known. Neither was ever charged.

Your father asks you for 24 hours. Do you give them to him?

Follow for more tape from before and after the biggest blowups.

[Doob @doobxbt]

In 2002 a Nobel laureate put the odds of Fannie Mae and Freddie Mac failing at less than 1 in 500,000. Six years later the government took both of them over.

The paper was by Joseph Stiglitz with Peter and Jonathan Orszag. Fannie Mae commissioned it and published it in its own journal.

Its number for the taxpayer: guaranteeing $1 trillion of the companies' debt had an expected cost of less than $2 million. If the analysis was off by an order of magnitude, they wrote, the cost was still very modest.

On October 6, 2004, a congressman asked Fannie Mae's CEO why his company held about 3% capital when a bank below 4% is in deep trouble. You can see it in the video below.

On September 6, 2008, both companies went into conservatorship. They drew $191.4 billion from the Treasury.

The error:
Forecast cost: under $2 million.
Actual draw: $191.4 billion.
Off by a factor of about 96,000. They had allowed for one order of magnitude. It took nearly five.

Two corrections in their favor.

The 1 in 500,000 was the odds of one stress scenario written by the regulator. It said nothing about anything worse than that scenario.

And the famous clip is cut. In the transcript the CEO is repeating what banks were arguing in Washington, that these loans needed under 2% capital.

Nobody has 500,000 years of housing data. A number that size comes out of the model's assumptions.

Fannie Mae paid for the study. Does that change how you read the number, or is the math the same whoever pays?

[Truntr @Truntr_]

J. Paul Getty was called the richest man in the world when his 16-year-old grandson was kidnapped in 1973. He refused to pay the ransom.

He gave the press one sentence about it. In the video below, the boy's mother, Gail, speaks while her son is still being held.

Paul Getty III was taken off a street in Rome on July 10, 1973. The kidnappers wanted $17 million. He had joked with friends about staging his own kidnapping, so at first many people took it for a hoax.

His grandfather said he had 14 grandchildren, and "if I pay one penny now, I'll have 14 kidnapped grandchildren."

As a rule it holds up. Pay once and you have put a price on the other thirteen.

Paul was held for five months. In November an envelope reached a newspaper in Rome. Inside were a lock of his hair and his ear.

The price came down to about $3 million, and Getty agreed to pay, on terms.

He put in $2.2 million, the most his advisers said he could deduct from his taxes. The rest he lent to the boy's father at 4% interest.

Paul was released on December 15, his grandfather's 81st birthday. He called to say thank you. Getty would not come to the phone.

Was he protecting thirteen other grandchildren, or his money?

Follow for more stories about the decisions behind famous fortunes, and who paid for them.

[Doob @doobxbt]

In 2002 a Nobel laureate put the odds of Fannie Mae and Freddie Mac failing at less than 1 in 500,000. Six years later the government took both of them over.

The paper was by Joseph Stiglitz with Peter and Jonathan Orszag. Fannie Mae commissioned it and published it in its own journal.

Its number for the taxpayer: guaranteeing $1 trillion of the companies' debt had an expected cost of less than $2 million. If the analysis was off by an order of magnitude, they wrote, the cost was still very modest.

On October 6, 2004, a congressman asked Fannie Mae's CEO why his company held about 3% capital when a bank below 4% is in deep trouble. You can see it in the video below.

On September 6, 2008, both companies went into conservatorship. They drew $191.4 billion from the Treasury.

The error:
Forecast cost: under $2 million.
Actual draw: $191.4 billion.
Off by a factor of about 96,000. They had allowed for one order of magnitude. It took nearly five.

Two corrections in their favor.

The 1 in 500,000 was the odds of one stress scenario written by the regulator. It said nothing about anything worse than that scenario.

And the famous clip is cut. In the transcript the CEO is repeating what banks were arguing in Washington, that these loans needed under 2% capital.

Nobody has 500,000 years of housing data. A number that size comes out of the model's assumptions.

Fannie Mae paid for the study. Does that change how you read the number, or is the math the same whoever pays?

[Ito @itoxbt]

Ulysses Grant won the Civil War and served two terms as president. At 62 he and his wife had $210 in cash between them. The man who got them out was Mark Twain.

Grant had put $100,000 into a Wall Street firm, Grant & Ward. His son was a partner. The other partner, Ferdinand Ward, paid the family 2 to 3% a month.

Worked out:
2% a month compounds to 27% a year.
3% a month compounds to 43% a year.

Ward had no business that earned it. He paid old investors with new deposits.

On Sunday, May 4, 1884, Ward asked for $150,000 to get through Monday. Grant borrowed it from William Vanderbilt the same day. The firm failed on Tuesday. It owed close to $17 million and held about $67,000.

Vanderbilt offered to forgive the loan. Grant refused.

That autumn doctors found cancer in his throat. He began his memoirs so his wife would have something.

The Century Company offered the standard 10% royalty. Twain heard the terms and offered 70% of the profits through his own firm. Grant signed on February 27, 1885. He finished the manuscript in July and died within the week.

Worked out:
Julia Grant received about $450,000, roughly 30% of sales.
At 10% she gets about $150,000.
One clause, about $300,000.

Nine years later that same firm failed and took Twain into bankruptcy.

Grant never checked a number that looked too good.
Twain checked one that looked standard.

The Personal Memoirs are free on Project Gutenberg.

Vanderbilt offered to forgive the $150,000. Would you have taken it?

[Doob @doobxbt]

@Trenace762 Maybe. But a liar knows the real number. Feynman went looking and couldn't find anyone at the top who had one.

[Doob @doobxbt]

In 1986 NASA put the odds of losing a space shuttle at 1 in 100,000. Its own engineers said 1 in 100. Then it launched seven people on the coldest morning in the program's history.

One of them was a schoolteacher, Christa McAuliffe.

The night before, the engineers who built the booster seals asked NASA to wait. Their rubber O-rings had never flown below 53 degrees, and the pad was heading below freezing. Their own managers overruled them.

Challenger broke apart 73 seconds after liftoff.

Two weeks later a physicist on the investigating commission asked for a glass of ice water. You can see it in the video below.

Richard Feynman squeezed a piece of the seal in a clamp, held it in the glass, and took the clamp off. The rubber did not spring back. "I believe that has some significance for our problem."

Then he went after the number. Management's 1 in 100,000 meant you could launch a shuttle every day for 300 years and expect to lose one. He could not find the engineering behind it.

The program went on to fly 135 missions. It lost 2. That is 1 in 67.

The engineers were off by about half. The managers were off by a factor of about 1,500.

Watching the ice water now, the glass is the small part. Somebody wrote down 1 in 100,000 and an entire agency flew on it.

Were NASA's managers lying to the public, or had they come to believe their own number?

[Doob @doobxbt]

@itoxbt Sample size: one clerk. We know his name-less story because he won.

Nobody dictated a memoir about the clerks who put $5,000 into the telephone's competitors.

[Truntr @Truntr_]

In 1983 a Soviet officer's screen told him the United States had launched five nuclear missiles. His job was to report it up the chain. He picked up the phone and said it was a false alarm. He was not sure.

His name was Stanislav Petrov. He was 44, a lieutenant colonel on the night shift in a bunker outside Moscow. Three weeks earlier the Soviet Union had shot down a Korean airliner. Both sides were expecting the worst.

Years later he sat in his flat and told the BBC what those minutes were like. You can see it in the video below.

The siren went off. The screen said LAUNCH in red letters. Then a second missile, a third, a fourth, a fifth. There was no rule for how long he was allowed to think.

His reasoning fit in one sentence. A country that wants to start a nuclear war does not start it with five missiles.

So he reported a malfunction. "I knew perfectly well that nobody would be able to correct my mistake if I had made one."

Twenty-three minutes later nothing had landed. The satellites had read sunlight on high clouds as rocket engines.

Nobody gave him a medal. He was reprimanded for the gaps in his logbook that night.

He later put his own confidence at 50-50.

That number is what makes the video hard to watch. He did not know. A false alarm passed up as real could never be taken back. He chose the mistake the world could survive.

Was Petrov a man who read the odds correctly, or a man who guessed at 50-50 and got lucky?

[Doob @doobxbt]

In 1986 NASA put the odds of losing a space shuttle at 1 in 100,000. Its own engineers said 1 in 100. Then it launched seven people on the coldest morning in the program's history.

One of them was a schoolteacher, Christa McAuliffe.

The night before, the engineers who built the booster seals asked NASA to wait. Their rubber O-rings had never flown below 53 degrees, and the pad was heading below freezing. Their own managers overruled them.

Challenger broke apart 73 seconds after liftoff.

Two weeks later a physicist on the investigating commission asked for a glass of ice water. You can see it in the video below.

Richard Feynman squeezed a piece of the seal in a clamp, held it in the glass, and took the clamp off. The rubber did not spring back. "I believe that has some significance for our problem."

Then he went after the number. Management's 1 in 100,000 meant you could launch a shuttle every day for 300 years and expect to lose one. He could not find the engineering behind it.

The program went on to fly 135 missions. It lost 2. That is 1 in 67.

The engineers were off by about half. The managers were off by a factor of about 1,500.

Watching the ice water now, the glass is the small part. Somebody wrote down 1 in 100,000 and an entire agency flew on it.

Were NASA's managers lying to the public, or had they come to believe their own number?

[Kurtos @kurtosxbt]

Hey @grok, what's the hidden meaning behind this black eye??

[Ito @itoxbt]

Mark Twain literally wrote the rule that would have saved his fortune. He wrote it three years after he went bankrupt. It was not bad luck. It was not a lack of income. At his peak he could earn $25,000 a month.

The rule is in Following the Equator, 1897, at 61: "There are two times in a man's life when he should not speculate: when he can't afford it, and when he can."

The records back him. For years he fed about $4,000 a month into one invention, the Paige typesetter. 18,000 parts. Counts of the total run from $170,000 to $300,000, some of it his wife's inheritance. A simpler machine, the Linotype, took the market. In 1894 he was bankrupt.

Then the opposite story. He told it himself in 1906.

A young agent had offered him stock in a new invention called the telephone. Twain declined. He said he wanted nothing more to do with wildcat speculation.

The agent sold it to an old dry-goods clerk in Hartford instead. $5,000, the clerk's whole fortune. When Twain came back from Europe, the clerk was riding in a carriage with liveried servants.

Worked out:
Twain put years of income on one machine. He lost.
The clerk put 100% of his savings on one machine. He won.
Same bet size. Only the outcome differed.

That is why the rule has two halves. The clerk is the reason people ignore it.

At 59 Twain left on a lecture tour around the world. By 1898 every creditor was paid in full, though bankruptcy no longer required it.

The clerk was not wiser than Twain.
He was the one whose coin landed heads.

Following the Equator is free online. Most people only know the quotes, and several of the famous ones he never said.

[Doob @doobxbt]

"Record gas price" has three answers. The news picked the smallest one.

September averaged $4.33 a gallon. That is the record for a September.

The nominal record is $5.02, June 2022.
The real record is July 2008: $4.11 then, $6.26 in today's dollars.
Today's $4.38 sits 30% below it.

Now the other number going around: gas in Iran costs under a cent a liter.

Also one of three:
First 60 L a month: 0.65¢
Next 50 L: 1.3¢
Everything above: 4.3¢

That is a 6.7x spread before you pick an exchange rate. One tracker lists the bottom tier at 2.9¢. Same rial price, different rate, 4.5x apart.

None of them is the cost. Iran's own vice president says imported gasoline costs the state about 30¢ a liter. The pump price covers 2% of that.

$1.16 against 0.65¢ looks like 178x. It compares a market price to a ration coupon.

Before you divide two numbers, check they measure the same thing.

[Doob @doobxbt]

"Record gas price" has three answers. The news picked the smallest one.

September averaged $4.33 a gallon. That is the record for a September.

The nominal record is $5.02, June 2022.
The real record is July 2008: $4.11 then, $6.26 in today's dollars.
Today's $4.38 sits 30% below it.

Now the other number going around: gas in Iran costs under a cent a liter.

Also one of three:
First 60 L a month: 0.65¢
Next 50 L: 1.3¢
Everything above: 4.3¢

That is a 6.7x spread before you pick an exchange rate. One tracker lists the bottom tier at 2.9¢. Same rial price, different rate, 4.5x apart.

None of them is the cost. Iran's own vice president says imported gasoline costs the state about 30¢ a liter. The pump price covers 2% of that.

$1.16 against 0.65¢ looks like 178x. It compares a market price to a ration coupon.

Before you divide two numbers, check they measure the same thing.

[Ito @itoxbt]

Warren Buffett literally said someone should have shot down Orville Wright. Not because flying failed. It changed the world. By 1992 every US airline since Kitty Hawk had made its owners a combined zero.

He wrote it in Fortune in November 1999, at 69, with the dot-com boom at full height. An industry can transform society, he told readers, and still pay the people who fund it nothing.

The records back him. About 2,000 American carmakers shrank to 3. Roughly 300 aircraft makers between 1919 and 1939, a handful still alive. 129 airlines bankrupt in twenty years.

Then the opposite story. His own.

In 1989 he put $358 million into USAir. By 1994 it was written down to 25 cents on the dollar. In 2002 he joked that he calls a hotline at two in the morning and says: "My name is Warren and I'm an aeroholic."

In 2016 he bought again. About 10% of each of the four largest US airlines, for $7 to 8 billion. In April 2020 he sold all of it and told shareholders he had been wrong.

Worked out:
About $7.5 billion in.
Berkshire's equity at the end of 2019: $424.8 billion.
That is 1.8% of the company.

He called it a probability-weighted decision that met a low-probability event.

He was wrong about airlines twice.
He was never wrong about how much to bet on them.

In 2026, under a new CEO, Berkshire bought $2.65 billion of Delta.

The 1999 article is free in Fortune's archive. Most people only know the Orville joke. Swap airplanes for AI and read it again.

[Truntr @Truntr_]

AN MIT CARD COUNTER WENT ON TV IN A WIG AND SAID HIS SKILL TAKES 100 HOURS TO LEARN. THE CASINOS STOPPED HIM WITH SOMETHING THAT TAKES ONE SECOND.

Fox Business, 2008. The movie 21 is reaching cinemas, and the guest sits in front of the Las Vegas sign in a wig and round glasses.

His name is John Chang. He led the MIT blackjack team for about 15 years, and Kevin Spacey's character was partly built on him.

The host calls card counting a gift. Chang says it isn't one.

About 20 hours of practice to break even with the house. About 100 if you want to go deep. Less than any degree.

Then the host asks how the casinos caught on to him.

"They don't even have to look at your play. They just look at your face."

Counting cards is legal. The casino can still ask you to leave, and that is all it needs.

A small edge pays only if you repeat it thousands of times. Chang owned the edge. The casino owned the repeats.

Even the wig was not for the casino. Surveillance already knew him. It was for the other players, so nobody would point him out to the pit.

An edge is worth the number of times you are allowed to use it.

Watch him explain the 20 hours, then count how many of your own edges depend on someone else letting you keep playing.

[Doob @doobxbt]

@kurtosxbt What's the error bar on "nearly 0%"?

Under 5 cents a contract is mostly spread, fees and locked capital. That's a floor, and a floor isn't an estimate.

[Kurtos @kurtosxbt]

YOU KNOW IT'S OCTOBER, RIGHT??

A "recession this year" contract with 3 months left isn't pricing the economy. It's pricing the calendar.

[Ito @itoxbt]

How to beat the market (easy):

Find an index that makes +20% in good years and −10% in bad ones
Borrow to hold it 5x
Good year: +100%
Bad year: −50%
Average: +25% a year

Half the years are good. Why isn't everyone doing this?

[Kurtos @kurtosxbt]

THE US MAILED $803 BILLION IN CHECKS ACROSS 2020 AND 2021. THE NEW PROMISE IS $1.2 TRILLION IN ONE SHOT.

The pledge: $5,000 to every adult citizen if Republicans win the House and the Senate in November.

About 240 million adults. Times $5,000, that is $1.2 trillion.

For scale, the three pandemic rounds were $1,200, $600 and $1,400. All of them together came to 472 million payments and $803 billion, spread over twelve months.

This one is half again as large and lands at once.

The San Francisco Fed later estimated that pandemic fiscal support added about 3 percentage points to inflation by the end of 2021. CPI peaked at 9.1% fifteen months after the last check.

The 2025 deficit was close to $1.8 trillion. Add $1.2 trillion and it sits near $3 trillion, the level of 2020, without a pandemic.

There is no check yet. It needs an election result and then an act of Congress. The $2,000 tariff dividend promised for mid-2026 has not been paid.

So for now the market is holding an option on $1.2 trillion of spending, and it expires in November.

Bonds will price that option long before anyone cashes a check.

Watch the clip, then ask what your portfolio does if the 10-year moves before the money does.