Orinfor

finance, decoded | markets · money · incentives

[Orinfor @orinforxb]

In 2018 Warren Buffett explained to 40,000 people how $10,000 turns into $51 million.

Then he named the one profession it would have wiped out, and the room laughed.

The math took him nine minutes. Put $10,000 into American business on March 11, 1942, the week the United States was losing the war on both oceans. Never look at a quote again. Never take another piece of advice from anyone.

You finish with $51 million.

Do the same thing with gold and you finish with $400,000. Same money. Same 76 years. For every dollar the first person made, the second made less than a penny.

Then Buffett looked up from his notes and added this:

"There's one problem. Your friendly stockbroker would have starved to death."

Everybody laughed. It is not a joke. It is the entire mechanism, said out loud, by the one man in the building with no reason to hide it.

Nobody in that industry has to lie to you. The broker is paid per transaction, so he needs you to move. The index fund that beat him earns a few basis points and cannot make a dollar more by touching your account. Two people, same market, opposite instructions written directly into how they eat.

One of them is structurally required to leave you alone. The other one is structurally required to call you.

Forty-nine minutes later in this same session, someone asks Buffett about a bank that had fired 5,300 employees. His answer, about a company he was the largest shareholder of:

"Wells Fargo is a company that proved the efficacy of incentives. It's just that they had the wrong incentives."

Not a scandal. A structure. It worked perfectly, on the wrong target, for years, and it was printed in the annual report as a strength.

The article above is that structure taken all the way apart. Every salary, fee, commission and bonus that has ever existed is one of three things. Only one of them pays more when you win.

You are sitting inside three or four of them right now, and you have probably never been told which.

[Orinfor @orinforxb]

https://t.co/2b3SoHyQDc

[Orinfor @orinforxb]

https://t.co/2b3SoHyQDc

[Orinfor @orinforxb]

In 1999 the New York Mets tried to pay a player $5.9 million to go away.

He said no.

Bobby Bonilla was 36 and finished. The team wanted him off the roster and offered to write the check that afternoon. He asked for something else instead. Don't pay me now. Pay me later, at 8%, starting in 2011.

The Mets agreed in a room where everyone thought they had won.

Every July 1st since 2011, they mail him $1.19 million. They will keep mailing it until 2035. The $5.9 million they wanted to hand over in cash turned into roughly $30 million.

Nobody negotiated a raise. The number was already sitting there.

Jonathan Gruber spends an entire lecture on the reason in MIT's introductory economics course. A dollar tomorrow is worth less than a dollar today, so every promise of money has to be dragged back to the present before it can be compared to anything. Money in different years isn't the same substance. Adding it up is like weighing a pound of apples, a pound of steak and a pound of gold and announcing you have three pounds.

Which means every income has two numbers. The amount, and the when.

Wages arrive now and stop the moment the work stops. Capital arrives later and multiplies while it waits. Arbitrage arrives instantly and dies on contact. Insurance collects first and pays out years afterward.

Four ways to make money. Four positions on a calendar. They were never four sizes of the same thing.

Bonilla wasn't a financier. He was a ballplayer at the end of his career who ran one line of arithmetic the front office was too relieved to run.

The people who get rich rarely negotiate a better number. They negotiate a better date.

[Orinfor @orinforxb]

@cryp_cryp Thanks mate. Honestly he could've asked for 9-10% and they'd still have signed

[Orinfor @orinforxb]

In 1999 the New York Mets tried to pay a player $5.9 million to go away.

He said no.

Bobby Bonilla was 36 and finished. The team wanted him off the roster and offered to write the check that afternoon. He asked for something else instead. Don't pay me now. Pay me later, at 8%, starting in 2011.

The Mets agreed in a room where everyone thought they had won.

Every July 1st since 2011, they mail him $1.19 million. They will keep mailing it until 2035. The $5.9 million they wanted to hand over in cash turned into roughly $30 million.

Nobody negotiated a raise. The number was already sitting there.

Jonathan Gruber spends an entire lecture on the reason in MIT's introductory economics course. A dollar tomorrow is worth less than a dollar today, so every promise of money has to be dragged back to the present before it can be compared to anything. Money in different years isn't the same substance. Adding it up is like weighing a pound of apples, a pound of steak and a pound of gold and announcing you have three pounds.

Which means every income has two numbers. The amount, and the when.

Wages arrive now and stop the moment the work stops. Capital arrives later and multiplies while it waits. Arbitrage arrives instantly and dies on contact. Insurance collects first and pays out years afterward.

Four ways to make money. Four positions on a calendar. They were never four sizes of the same thing.

Bonilla wasn't a financier. He was a ballplayer at the end of his career who ran one line of arithmetic the front office was too relieved to run.

The people who get rich rarely negotiate a better number. They negotiate a better date.

[Lumen @lumenxbt]

https://t.co/3hWKMIBJcw

[Orinfor @orinforxb]

@mathhattan Exactly. They weren't deferring the money, they were investing it. In a Ponzi scheme

[Orinfor @orinforxb]

@BonBona9z Gambling is betting everything. Investing is betting a quarter

[Orinfor @orinforxb]

A coin. Heads, your money grows 50%. Tails, it falls 40%.

The odds are in your favour. Every textbook says take that bet.

Take it a hundred times and you end with almost nothing. Not probably. With near certainty.

Average across ten thousand people who each play once: +5% per round.
Follow one person who plays ten thousand rounds: −5% per round.

Same coin. Same payouts. Opposite sign.

Start with $1,000. The first calculation says you're holding $131,500 after a hundred rounds. The second says about five dollars.

Both numbers are correct. They answer different questions.

Ole Peters, Santa Fe Institute and the London Mathematical Laboratory, put that coin in front of a Gresham College audience in London in November 2012 and pointed out that economics has been quoting the first number since the 1650s.

The name for the gap is non-ergodicity. The average across possible worlds stops matching what happens in one life the moment gains and losses multiply instead of add.

Which is every business, every portfolio, every leveraged position ever taken.

Then he changes one thing. Not the odds. Not the payouts. The size of the bet.

Stake everything each round and the growth rate is −5%.
Stake a quarter of what you have and it turns positive.

Same coin. The only variable that ever mattered was how much.

Ruin isn't a bad outcome in that game. It's an exit. There is no round after zero, and the average never has to notice you left.

The people who compound for thirty years aren't the ones with the best odds. They're the ones who sized the bet so there's always a next round.

[Lumen @lumenxbt]

https://t.co/3hWKMIBJcw

[Orinfor @orinforxb]

Eighty-nine years of being right ended in a single quarter and cost $182 billion. The cause was one word in one formula, and it was never on the balance sheet.

The formula is the fourth way to make money, written out:

σ = √(p(1−p)/n)

Pool n risks and uncertainty collapses at 1/√n. At a 1% event rate, a hundred policies gives ±1% precision. A million gives ±0.01%. Nobody at the company knows which customer crashes. They know what fraction will, to two decimals, and they sell the difference.

Fear in, money out. That gap is the entire product.

The word is independent. The risks in the pool have to be unrelated to each other, and correlation doesn't weaken that formula — it deletes it. Perfectly correlated, a million policies carry exactly the uncertainty of one policy. Not worse. Identical.

AIG is what that looks like from the inside.

1919, Shanghai. Cornelius Vander Starr opens an insurance shop, an American selling policies to Chinese customers, which nobody was doing. He runs it 49 years, gets out ahead of Mao, rebuilds in New York. Hank Greenberg takes over and turns it into the largest insurer on earth.

2008. The federal government commits $182 billion to keep it upright.

Robert Shiller walks through the whole arc at Yale on January 31, 2011 — lecture five of twenty-three — and the cause he names is not fraud and not incompetence. It's the independence assumption. AIG's book was thousands of separate-looking contracts that were, underneath, one bet on American house prices.

He titled the lecture "Opportunities and Vulnerabilities." Both halves came out of the same line of chalk.

Which makes the fourth engine a narrower job than it looks. Nobody gets paid for absorbing risk. They get paid for being right about how unrelated their risks are — a variable no auditor checks, that no spreadsheet contains, and that stays invisible until the one week it's the only thing that matters.

Worth asking of any guarantee, warranty, retainer or fixed price:

If this goes wrong for one client, does it go wrong for all of them in the same week?

If yes, that isn't a portfolio. It's one bet, sold many times.

[Superior @andreysuperior]

https://t.co/rUN1cPDY65

[Orinfor @orinforxb]

Professional venture capitalists lose money on 65% of their investments.

That is not the failure case. That is the business model.

Correlation Ventures ran every US venture financing between 2004 and 2013 that ended in a shutdown, an acquisition or an IPO. 21,640 of them. The full distribution:

64.8% returned less than 1x — the money did not come back
25.3% returned 1–5x
5.9% returned 5–10x
2.5% returned 10–20x
1.1% returned 20–50x
0.4% returned more than 50x

Read the first line and the last line together. Two out of three investments lose money. Four in a thousand return more than fifty times.

Now the arithmetic that makes it work. Put $1 into each of 250 companies at those odds. About 162 return roughly nothing. But the one company in the 50x bucket returns at least $50 on its own — a single name covering a third of the wreckage before you count anything in between.

The most you can lose on any one bet is what you put in. The most you can make has no ceiling. That asymmetry is the whole game, and it's why a 65% loss rate is a functioning industry rather than a scandal.

It also kills the obvious lesson. These weren't amateurs guessing. Every one of those 21,640 financings was chosen by professionals after months of diligence — and they still lost money on two thirds.

Worth keeping — the three conditions that make a losing rate survivable:

1. Bounded downside. You can lose 1x and never more. No leverage in a tail-heavy game, ever.
2. Unbounded upside. If the best realistic case is 3x, the losses eat you alive. A 50x has to be possible.
3. Enough attempts. One bet at these odds is gambling. Two hundred is a portfolio.

Miss any one of the three and the same distribution that builds fortunes will quietly take everything you have.

[Tigerflow @tigerfl0w]

https://t.co/2JQpx1CJeg

[Orinfor @orinforxb]

97% of people do not live in a power law.

They live in an exponential distribution — mathematically the same one that describes energy in a gas.

This is the part that keeps getting skipped. When physicists actually fit income data instead of asserting a shape, US incomes split cleanly into two populations that obey two different laws:

The bottom 97–99% follow an exponential Boltzmann-Gibbs distribution. Yakovenko's group calls this the "thermal" class.

The top 1–3% follow a Pareto power law. The "superthermal" class.

And there is no clearly defined middle class in between. Not as a political statement — as a fitting result. The data show two regimes and a handover, not a continuum.

Why an exponential? Because money is conserved in a transaction. What leaves one account arrives in another. Energy behaves the same way in a collision, and when you run the statistical mechanics on a conserved quantity being randomly exchanged, you get the exponential distribution every time. It shows up in income data across 67 countries.

Now the part that reorganises everything.

The thermal bulk is essentially stationary over time. It doesn't move. What changed over the decades is that the tail detached and swelled — the top's share of income went from 4% in 1983 to 20% in 2000, then fell back with the 2001 crash, because the tail tracks the stock market and the bulk tracks wages.

So inequality didn't grow by the bottom sinking. It grew by the top separating.

Worth keeping — the only question that matters here:

"Does my money grow by addition, or by multiplication?"

Wages, fees and salaries are additive exchange in a conserved pool. That mathematically produces an exponential, no matter how large the number is.

Ownership — equity, royalties, capital — grows as a percentage of itself. That is the only process that produces a tail.

You don't move up inside the exponential. You leave it or you don't.

[Tigerflow @tigerfl0w]

https://t.co/2JQpx1CJeg

[Orinfor @orinforxb]

@senmarketjp Exactly. No grain is greedy

[Orinfor @orinforxb]

Billionaire wealth, earthquake magnitudes, and grains of sand falling onto a pile all follow the same curve.

Only one of those three involves greed.

That single fact breaks the story almost everyone tells about power laws.

The usual explanation is proportional growth — the rich get richer, the popular get more popular, and the gap compounds until a sliver at the top holds most of everything. It's real, it's called preferential attachment, and Barabási and Albert formalised it for networks in 1999.

But it requires somebody to already be ahead. A tectonic plate isn't ahead of anything.

So in 1987, three physicists — Bak, Tang and Wiesenfeld — built the simplest model that could find the other mechanism. Drop grains of sand onto a pile, one at a time. That's the entire model.

The pile grows until it reaches a critical slope. From then on, every new grain can trigger an avalanche, and the avalanche sizes follow a power law. Mostly tiny. Occasionally enormous. No grain is special. No grain has an advantage. The distribution shows up regardless.

They called it self-organized criticality — systems that tune themselves to the edge of collapse and then produce cascades at every scale.

It's why earthquake magnitudes follow the Gutenberg-Richter law. Same story for forest fires, avalanches and power-grid failures. Not accumulation. Structure.

Two mechanisms, and they imply completely different strategies:

PREFERENTIAL ATTACHMENT — advantage goes to whoever arrived first. The lever is timing.
CRITICALITY — advantage goes to whoever stands in a medium that transmits. The lever is position.

Worth keeping — the question that tells them apart:

"If I do something excellent here, can it propagate on its own, or does it stop with the person I hand it to?"

If it stops, you're standing on sand that's too flat, and no amount of quality will manufacture a tail. Being early in a dead medium is still dead.

[Tigerflow @tigerfl0w]

https://t.co/2JQpx1CJeg

[Orinfor @orinforxb]

Buffett paid $8.6 million for a company in 1967. It handed him $19.4 million to invest on day one.

Not profit. Not a loan. Other people's money, legally in his hands, with no repayment date.

The company was National Indemnity. The $19.4 million was float.

Here's the mechanism, and it's the most underrated structure in finance. An insurer collects premiums now and pays claims later — sometimes decades later. In the gap, it's holding a very large pile of somebody else's money. If the underwriting roughly breaks even, that pile costs nothing. If underwriting is profitable, the pile costs less than nothing: you are being paid to hold money you get to invest.

That solves the one problem everybody hits.

The standard advice is: earn money, save the difference, invest the pile, wait thirty years. True, brutal, and slow, because the size of your pile is capped by your salary minus your rent.

Buffett didn't solve that constraint. He walked around it. He bought the machine that produces piles.

Watch it compound:

1967 — $19.4 million of float
2015 — $88 billion
2024 — $171 billion
End of 2025 — roughly $176 billion

His words: insurance is "the engine that has propelled our expansion since 1967."

Sixty years of investing with a permanently growing supply of capital that isn't yours and doesn't need to be paid back on any schedule. That's the trick. It was never the stock picking.

Worth keeping — float exists far below Berkshire's altitude. Look for it anywhere money arrives before the work is delivered:

Annual subscriptions paid upfront
Deposits and retainers
Prepaid packages
Deferred revenue of any kind

Ask one question of your own income: "does the money arrive before or after I deliver?"

Before is float. After is a wage.

[Superior @andreysuperior]

https://t.co/rUN1cPDY65

[Orinfor @orinforxb]

@Profit80lvl True, but the algorithm can't say "I don't like you." It's blind. That’s a whole different level of the game.

[Orinfor @orinforxb]

There are only four ways to make money, and two of them require somebody's permission.

Nobody ever mentions that part.

Labor: someone has to hire you. Capital: someone has to hand you money, or you spend twenty years accumulating it yourself. Both have a gatekeeper. Both existed in Florence in 1400 for exactly that reason — they were the only two that could exist back when distribution wasn't free.

Naval Ravikant cuts it differently, and the cut is sharper.

His version, posted 31 May 2018: "Fortunes require leverage. Business leverage comes from capital, people, and products with no marginal cost of replication (code and media)."

Four forms of leverage. Labor, capital, code, media.

The first two are permissioned. People have to agree to work for you. Someone has to give you the money. Someone can say no, and mostly they do.

Code and media are permissionless. Nobody approves your commit. Nobody signs off on you publishing. You build it once and it runs while you sleep, at zero marginal cost, for one person or ten million, at the same effort.

That's the real reason people get stuck on engine one, and it has nothing to do with discipline or savings rate.

Both engines you were told about have a gatekeeper. Both of the ones that don't were invented after the framework you inherited.

Essentially every new fortune of the last twenty years runs on the permissionless two. Not because those people work harder. Because their output stopped needing anyone's approval to reproduce.

Worth keeping — ask this of every income stream you have:

"Who has to say yes for this to double?"

If the answer is a boss, a client, an investor or a committee, you're on permissioned leverage and your ceiling is somebody else's calendar.

If the answer is nobody, you've found the other kind.

[Superior @andreysuperior]

https://t.co/rUN1cPDY65

[Orinfor @orinforxb]

@marlowfinds The jump isn't earning more per hour. It's making something that gets paid while you sleep.

[Orinfor @orinforxb]

@Profit80lvl Sometimes they're worth it

[Orinfor @orinforxb]

He didn't beat the casino by gambling. He beat it by noticing that a casino is an insurance company.

Ed Thorp, math professor. In 1962 he published Beat the Dealer and proved card counting worked. Casinos changed the rules, added decks, and banned him. The edge closed, the way every edge closes.

So he stopped playing at the table and rebuilt the table.

Think about what a casino actually is. It is not in the gambling business. It absorbs risk it has measured precisely, from thousands of people who have not measured it, and collects a small statistical premium every single time. Volume, a known edge, and enough capital to survive variance.

Fear in, money out. That's not gambling. That's the insurance business wearing carpet and neon.

Thorp took the identical structure to Wall Street. Princeton/Newport sold and hedged the instruments other people were afraid to hold — warrants, convertibles, options. Collect the premium, hedge the exposure, repeat it thousands of times.

Nineteen years of that: 19.1% a year, 227 profitable months out of 230, worst month under 1%.

That is not what a gambler's track record looks like. That is what an insurance company's track record looks like.

Now the uncomfortable version for the rest of us. In nearly every transaction you make, one side is the casino and one side is the player. The $99 laptop warranty. The extended shipping guarantee. The fixed-price quote instead of hourly. The spread on your brokerage trade.

You pay a premium for certainty. Someone with better arithmetic collects it.

Worth keeping — one question, asked before you pay for anything that removes uncertainty:

"Am I paying the premium here, or collecting it?"

If the answer is never "collecting," you are funding other people's engines full time and calling it being careful.

[Superior @andreysuperior]

https://t.co/rUN1cPDY65

[Orinfor @orinforxb]

@marlowfinds Depends which chair you're in.