Elkhart, Kansas holds about 2,000 people, and until three years ago Shan Hanes was one of the more admired ones. He ran Heartland Tri-State Bank, a $139 million agricultural lender in the far southwest corner of the state. He mowed neighbors' yards. He worked the chain gang at high school football games. His shareholders were farmers and teachers who had parked generations of savings in that bank because Hanes was the guy you wanted running it.

Between May and July of 2023, he wired $47.1 million of the bank's money to strangers he had never met.

11 transfers in eight weeks. He also took $40,000 from his church, $10,000 from the local investment club, and $60,000 out of his daughter's college fund. Heartland ran dry, regulators shut it down, and roughly 30 local shareholders watched their stock go to zero. Hanes sits in Leavenworth today serving 24 years, the longest white-collar sentence ever handed down in Kansas.

Nobody threatened him. The people on the other end were running something far cheaper than a threat. Every time Hanes asked for his money back, they told him it was sitting right there, and that releasing it required one more transfer.

He believed that eleven times.

I got handed the same argument last week in an in an investor meeting. It took me a day to work out why my hands went cold.

One sentence this week. It is the single most profitable sentence in modern fraud, worth billions a year, and it is also passed around as career wisdom by serious people who mean every word of it. Same argument, same machinery underneath. Two very different rooms, and only one of them ends in an indictment. Let me show you.

The Man Who Could Not Stop

What happened to Hanes has an ugly name in my business: pig butchering. The name comes from the operators themselves, who talk about fattening a hog before slaughter, and if you have never heard the term, understand that it describes a con with a schedule.

It starts as a wrong number, or a friendly stranger on LinkedIn, or a woman on a dating app. No pitch. Weeks of ordinary conversation, sometimes months. Somewhere in there a casual mention of a crypto platform the new friend has been doing well on. A small deposit goes in and a beautiful number comes back. The victim withdraws that first small profit easily, on purpose, because the operator wants him to learn that withdrawals work. Then the deposits get bigger.

The balance on the screen is a picture. It is a web page the operator controls, drawing whatever number keeps the man hooked. His actual money left for a private wallet the second it arrived.

So the crisis for the crew is never the pitch. It is the day the victim clicks withdraw for real money, and every crew has a script waiting for exactly that moment. The mechanical version is a fee. Forensic examiners who have taken these platforms apart describe a blocked transaction followed by a notice from "Support" or a "Compliance Officer" demanding a verification tax or a security deposit, usually 20% to 30% of the balance, before the funds can be released. Pay it and the denial comes back wearing a different hat. There is always one more form.

That is what Hanes was doing. He was not investing. By the end he was paying ransom on money that had never existed, wiring bank funds to unlock a number on a screen, and each wire made the last one harder to walk away from.

The volume tells you how well the script works. The FBI logged more than 1,000,000 complaints in 2025 and $20.877 billion in reported losses. Investment fraud was the largest category by a wide margin at roughly $8.6 billion, with cryptocurrency involved in about 72% of it. Average reported loss on the crypto side ran $62,604, and 18,589 people reported losing more than $100,000 each. Nobody wires six figures on impulse. Those are positions people held through doubt, on purpose, because holding felt like discipline.

Then the phone rings again. Fake law firms, fake recovery agents, fake government officials, one crew impersonating FBI complaint-center staff directly. Another 10,500 complaints last year and $1.4 billion gone. The recovery pitch is the same sentence run a third time. You quit too early. Stay in a little longer and it comes back.

The Same Words, a Better Room

Now the part that put me in a bad mood for a day. An investor said this to me last week, kindly, in a room where everyone was well dressed and nobody was committing a crime:

I wrote it down because it sounded true and because I am polite. Read it again knowing what the compliance officer told Hanes.

Both arguments do the identical piece of work. They take a question about evidence, which is should I still believe the number on this screen, and quietly swap it for a question about character, which is am I the kind of man who folds. Once that swap lands, you cannot be wrong anymore. You can only be weak. And a man who cannot be wrong has no mechanism left for stopping.

Notice what is missing from both. Neither one names a single condition that would ever mean stop. There is no number, no date, no observation that counts as an answer. The advice is unfalsifiable by construction, which is not a flaw in it, but a feature. It is the whole reason the sentence survives contact with bad news.

Hanes was not a stupid man. He ran a bank. He had spent his career being the person who spots this, and the script beat him anyway, because it does not attack your intelligence. It attacks the one instrument that would have saved you.

What the Math Actually Says

People invoke game theory when they hand you that advice, and game theory does not say what they think it says.

Patience does get rewarded in repeated games. That much is real. Weight the future heavily enough and cooperative strategies become sustainable, which is the finding Robert Axelrod made famous with his tournaments in the 1980s. Stop caring about later and defection turns rational rather than cowardly.

But every one of those models carries a second requirement, and the career-advice version deletes it. You have to be able to see the other player's moves. Take away observation and the whole structure collapses, because you can no longer tell a partner who is building from a partner who is defecting. Patience without a scoreboard is not a long game. It is a bet placed by a man who decided the scoreboard was a weakness.

The two mechanisms actually running inside that advice are not game theory at all, and they have better names. Loss aversion, from Kahneman and Tversky, is the finding that losses land roughly twice as heavy as equivalent gains. That is the threat-detection line, and it has real experimental weight behind it. Hyperbolic discounting is the other one. We savage the near term and barely discount the far term, which is why the same man sincerely plans to save money and then sincerely does not.

Which flips the conclusion completely. If patience only pays when you can observe, then the founder's job is not endurance. It is building the cheapest test that could prove him wrong and running it before the runway makes the answer academic. The advice recommends the opposite. Take the position, hold it, do not flinch. That is a formula for maximizing your time-to-discovery of a bad thesis, and for a man with 14 months of cash there is no more expensive mistake available.

The post-mortems back it. CB Insights went through 431 venture-backed companies that shut down after 2023. Ran out of capital shows up in about 70% of them, which reads like a cause of death and is really just the coroner's note. Underneath it sit poor product-market fit at 43%, bad timing at 29%, and unsustainable unit economics at 19%. Every one of those is an observation failure. Every one is something the market was saying out loud while somebody held the position and called it conviction.

And the claim that constant winners rarely have better information is survivorship bias in a good jacket. Plenty of people took the identical position with identical nerve and simply lost. They do not get interviewed. Harvard's Shikhar Ghosh, working from about 2,000 companies funded between 2004 and 2010, found roughly 75% of venture-backed firms never return cash to investors, with 30% to 40% liquidating to zero. His line about it has stuck with me for years: venture capitalists bury their dead very quietly. That research is more than a decade old and I could not find a fresher dataset measuring the same thing as cleanly, which is itself the point. The silence he described is exactly why nobody has built one.

Whose Variance Is It

A man holding 30 positions and a man holding 1 are not playing the same game. They are not even playing games with the same shape.

Venture returns follow a power law, so the investor's optimal move is for every company in his portfolio to take the highest-variance swing available. One outcome pays for the other 29. A portfolio of careful, moderate, survivable companies returns him nothing at all, and he knows it.

The founder has no portfolio. He has one position, one runway, one life, and a personal guarantee sitting in a drawer somewhere.

So advice optimized for the first payoff structure, delivered to a man standing in the second, moves variance from the party who can absorb it to the party who cannot. It does not require bad faith. He was describing, accurately, what the winners in his portfolio looked like from the outside. He left out the 29 who looked identical right up until they did not, because survivorship never feels like a bias when you have been in the room with the survivors.

Which leaves one structural feature shared by both rooms, and it is the only one that matters. The person telling you to hold is not the person who eats the loss if you are wrong.

The operator running the platform is committing a felony and the investor across the table is not. I want to be exact about that, because the difference is real and it is the entire difference. One version draws an indictment and a forfeiture action. The other draws a board seat and an anecdote about conviction. Same sentence, same mechanism, same risk rolling downhill onto the same guy. Only the intent differs, and intent is invisible to the man holding the bag.

Field Manual

Six habits. They work the same whether the position is a startup, a stock, or a stranger's trading platform.