Somewhere in New Jersey, there is a building full of computers whose entire job is to sit a few feet closer to the stock exchange than yours.
Not a few miles, but a few feet.
Because at the speeds these machines trade, distance is measured in the time it takes light to cross a room.
The firms that own them pay a fortune for that room…
In fact, it is some of the most expensive real estate in finance, and almost nobody who works there ever sets foot inside.
I thought about that warehouse this morning, when reading about a hairstylist who now runs what he calls a “hedge fund” from his kitchen table.
He’s not alone because brokerages like Robinhood and Webull have started letting ordinary investors hand the wheel to an AI agent.
The agent watches the market and clicks buy and sell on its own.
You can tell the agent what to watch for, like earnings announcements or unusual options activity…
One 19-year-old in the article even managed to turn $3,000 into $8,000 in a few months, including one Micron options trade that returned over 500%.
The idea is intoxicating… you’ve got an emotionless machine trading faster than you ever could.
But that leaves one teeny tiny but very crucial detail out…
This AI agent sends its orders over the same public internet you use to stream Netflix… from a laptop hundreds of miles from that New Jersey warehouse.
Seeing as light travels about a foot every billionth of a second… with the professionals having spent decades and untold millions buying back those billionths.
By the time your machine decides to pounce, the fast money has already been and gone.
So the one thing the AI is sold on… speed… is the one race you’re never winning…
But the good news is, you never needed to be in that race…
Warren Buffett didn’t get rich from quick execution… he got rich by being patient and following rules he set in calm moments rather than panicked ones.
That is where AI tools actually earn their keep.
They are a superb research assistant and a tireless enforcer of a plan.
But while a machine can help you hold that discipline, it should never hold the trigger.
Because when it holds the trigger, two things happen.
Researchers at the National Bureau of Economic Research found these models pile into the same crowded trades and still fail to beat a simple index fund.
And in 1987, when the professionals' machines all decided to sell at once, they set a record single day decline at 22%.
So let AI do all the reading it wants… just never let it do your deciding.
Oliver
P.S. The investors in that article said their agents had not gone rogue "yet."
Yet is doing a lot of work in that sentence, the 1987 funds didn’t think theirs would either.