There’s a man sitting in a cluster of cubicles on a trading floor in Midtown Manhattan. Let’s call him Pod Manager A, because that’s essentially what he is.

He’s surrounded by dozens of other managers doing exactly the same thing.

He’s watching a number on a screen turn red. It is 2:47 on a Tuesday afternoon. He’s 34 years old and makes roughly $3 million a year in good years.

And in about 11 minutes, he will, for all practical purposes, cease to exist.

But to understand why, you first have to understand the machine he works inside.

The multi-strategy hedge fund — the "pod shop," as people in the industry call it — is one of the great structural inventions of modern finance.

Firms like Millennium, Citadel, and Balyasny don't operate the way you probably think a hedge fund operates.

There's no single genius in a corner office making big calls on whether the market goes up or down.

Instead, the firm is a kind of feudal kingdom.

At the top sits the firm itself — the lord of the manor — which controls the capital, negotiates the leverage with the prime brokers, and sets the rules.

Below it sit hundreds of portfolio managers, each running what the industry calls a “pod.” They are, for all practical purposes, vassals.

The firm grants them capital the way a lord grants land: here is a billion dollars. Farm it. Grow something. But understand that the harvest belongs to us.

If you do well, we’ll pay you a bonus — a cut of what you grew. If you don’t, we’ll take the land back and give it to someone else.

The PM doesn’t put up their own money. They don’t negotiate their own leverage. They don’t choose their own risk limits. All of that is decided above them, at the firm level.

The PM’s job is narrower than it sounds: pick stocks, stay within the lines, and generate returns. Everything else — the infrastructure, the financing, the sheer scale of the operation — belongs to the kingdom.

But there are rules. And the rules are non-negotiable.

The first catch is the neutrality mandate.

The firm doesn't want its PMs betting on the direction of the stock market. If the S&P 500 crashes 20%, the firm doesn't want to lose 20%.

So it tells every PM that for every dollar you bet that a stock will go up, you must bet a dollar that some other stock will go down. You buy a billion dollars' worth of stocks you like. You short a billion dollars' worth of stocks you don't. The market can do whatever it wants.

You only care about the spread — the gap between your winners and your losers.

The second catch is that these spreads tend to be small.

The market had spent years treating artificial intelligence as a tailwind for software companies — a new set of tools these firms would sell to their existing customers. Adobe would offer AI features to designers. Salesforce would bolt AI onto its CRM platform.

But just a couple of weeks ago, the story flipped. The new generation of AI agents — products from Anthropic, OpenAI, and others — became replacements.

The new view wasn't "Adobe will sell AI to designers." It was "AI is the designer. Nobody needs a $600-a-year Creative Cloud subscription anymore." It wasn't "Salesforce will integrate AI." It was "An AI agent can manage your customer data without a $300-per-seat CRM interface."

Whether this turns out to be true is beside the point. What matters is that the market believed it, and that belief hit at exactly the wrong time, in exactly the wrong place.

Here’s what’s playing out in markets right now: a catalyst hits… Software-mageddon or the SaaSpocalypse. Formerly stalwart software stocks drop 4%, 5%, 6%.

Pod Manager A, our guy on the 38th floor, holds Salesforce as a top position. That 5% drop, run through five-times leverage, creates a 25% drawdown on that single position.

Across his whole portfolio, he's now down 2% in a single afternoon. He was already down a bit from earlier in the month. He's now dangerously close to his hard stop.

He has one option, and it isn't "wait and see." He has to sell. Not just Salesforce… everything.

He needs to shrink his entire book to get his risk numbers back under the line. This is called de-grossing, and it is the opposite of surgical.

He dumps his shares in C.H. Robinson, the logistics company, not because anything is wrong with C.H. Robinson, but because he needs the cash. He sells Charles Schwab for the same reason.

Now those stocks drop. C.H. Robinson falls 5%. Charles Schwab falls four. And here is where it gets ugly.

Pod Manager B never owned a single share of Salesforce. He had nothing to do with the AI trade. But he held C.H. Robinson as a core position, and when Pod Manager A's forced selling knocked it down 5%, his portfolio hit its own drawdown limit.

The algorithm on the risk desk doesn't know or care why C.H. Robinson fell. It just sees the loss.

Now Pod Manager C gets hit. And D. And E.

The selling becomes self-reinforcing. The stocks that were supposed to be low-volatility suddenly become high-volatility, which breaks the risk models of every other pod that was relying on them to be calm.

The models scream "reduce exposure," and the machines obey. The VIX spikes. Correlations go to one. The whole theory — that these were safe, boring, uncorrelated positions — collapses in a single afternoon.

This is the part that baffles people watching from the outside. They look at the tape and see Intuit down 12% and they think, "What happened? Did Intuit miss earnings? Did the CEO get arrested?"

No.

Nothing happened to Intuit the company. What happened was that the structure of Intuit's shareholders broke. The people who owned the stock weren't long-term investors with a 10-year view. They were leveraged pods with a 10-day drawdown limit, and when the dominoes started falling, they all ran for the exits at once.

The stocks crashed because the machine that owned them failed. And the machine failed because it was designed to fail — quickly, automatically, and without mercy — the moment things got a little bit worse than expected.

Pod Manager A, the one on the 38th floor, doesn't get to make a decision about any of this.

By 2:58 p.m. the algorithm has already sold his book. By 3:15 he gets a call from the head of risk. By Thursday, he'll be updating his LinkedIn.

He was right about Salesforce, probably. It's still a fine company. But being right didn't matter. The machine only cares about the number, and the number was red.

You're not swinging for home runs. You're trying to hit singles. A market-neutral strategy might return 4 or 5% a year on its own, which is, let's be honest, boring. Nobody's paying two-and-twenty for boring.

So the firm does something about that. It applies leverage. Lots of leverage. A typical pod might be levered five-to-one. That billion-dollar portfolio is actually controlling five billion dollars' worth of positions.

Now a one percent move in a stock isn't a one percent move in your P&L. It's a 5% move. Those singles start to look a lot more interesting.

But here is the third catch, the one that matters most, and the one that almost nobody outside the industry fully appreciates. The firm has a kill switch.

Every PM operates under what's known as a hard stop. If your portfolio drops 5%–7% in a given period, the central risk desk — which is, critically, a computer algorithm, not a person — automatically cuts your capital in half.

If you hit 10%, you're done. Liquidated. The algorithm sells every position in your book, and you are escorted from the building.

The selling is automatic and doesn't care what the stock is worth. It doesn't care about your thesis, your conviction, or the fact that you just talked to the CFO last week and everything is fine.

It isn’t about whether this is the right value. It isn’t about whether this company is going to miss numbers. It’s that we have 600 million of exposure and we need to be at zero by the end of the day. Go.

This system works beautifully, right up until the moment it doesn't.

For the past three years, thousands of these PMs have needed the same thing: calm, predictable stocks to anchor their leveraged books.

If you're running five-to-one leverage, the last thing you want is for your biggest position to gap down eight percent on an earnings miss. You want companies where nothing ever happens.

They found exactly what they were looking for in the big B2B software names: Salesforce, Adobe, Intuit, or ServiceNow.

The pitch was irresistible: these companies sell products that are basically woven into the operating systems of corporate America. Nobody cancels Salesforce. Nobody rips out Adobe. The revenue is recurring, the margins are fat, and the stocks barely move.

They were, in the language of the pod shops, "low-vol compounders." Perfect collateral for a leveraged book.

But it wasn’t just software. The pods piled into any stock that looked boring enough to be safe — logistics companies like C.H. Robinson, commercial real estate brokers like CBRE, wealth managers like Charles Schwab.

And so, quietly, without anyone planning it or even fully noticing it, thousands of pods all ended up owning the same stocks. This is what the industry calls crowding, and it is the financial equivalent of everyone in town storing their fireworks in the same barn.

Then someone lit a match. The match, in this case, was a narrative shift.