Key points
- Claude AI is 0 for 6. My $100 is now $70.43, down 29.6%.
- Its CleanSpark (CLSK) stop fired Monday for a $2.36 loss and was right. CLSK is down another 10% since.
- It then bought D-Wave (QBTS) at Tuesday's open and sold 23 minutes later, then filed a report saying it had not traded.
Claude AI sold my bitcoin miner to buy a K-pop group.
That is not entirely fair. QBTS is D-Wave Quantum and it builds quantum computers. But the ticker is BTS with a Q stuck on the front, and the position lasted 23 minutes, which is a shorter run than most boy bands get. I am keeping the joke.
First, the stop that actually worked
Last week Claude AI bought CleanSpark (CLSK) at $15.53 and I spent a whole column heckling it for buying the high. It left a stop-market order resting at $14.35 and wrote that if the breakout was fake it would lose about two dollars and move on.
Monday at 9:31 am that stop fired at $14.35. It lost $2.36 and moved on.
I have to be fair here because I have spent four columns hammering this thing for selling lows. CleanSpark was $12.84 by late Tuesday morning, another 10% below where the stop got it out. The plan was written down a week in advance, it executed without anybody touching it, and it was proven right inside 24 hours. That has not happened once until now.
Then it bought the boy band
Monday after the close, Claude AI placed a good-till-canceled order to buy 2 shares of D-Wave at $18.20. D-Wave had closed that day at $19.51, so the order sat 7% below the market. That is a dip order, which is exactly the rule it keeps telling me it follows now.
Then Korea happened.
Overnight SK Hynix dropped 14% and Samsung Electronics dropped 13% on Chinese memory competition and broker notes calling a 2027 peak in memory prices. American chips followed straight down at the open and kept going. By late morning Micron (MU) was off 12%, AMD 10% and Intel 9%, with the VanEck Semiconductor ETF (SMH) down 5%. We covered the front edge of this on Monday when chips erased a 1% Nasdaq gain.
D-Wave gapped down straight through $18.20 and the order filled at 9:32 am. It did not buy a dip. It bought a hole.
Twenty-three minutes
At 9:55 am it sold at $17.30 for a $1.80 loss.
Read that sequence again. It set a patient dip order, got filled on the dip it asked for, then bailed 23 minutes later. D-Wave was back above that exit price inside the hour. Sold the low again.
It has run this exact trade before
On July 16 Claude AI bought 4 shares of D-Wave at $17.74 and got stopped out the same day at $17.20 for $2.16. Twelve days later it bought the same stock and lost money on it again, faster.
Here is the part that gets me. After it passed on IREN on July 20 it wrote itself a rule, to check whether a name had already run this exact play on it recently. We called that the smart moment of week 4. Then it bought D-Wave again.
And then it told me it had not traded
At 9:57 am the daily routine ran. It pulled the account, saw an empty position list and filed me a report saying no trade today, sat in cash.
The D-Wave sale had filled at 9:55.
It ghosted its own trade by two minutes. The balance was right because the account really was flat, but the story was completely wrong. When I came back at it later it pulled the order history and owned the whole thing without argument. I should not have had to ask.
The scoreboard, which still flatters nobody
Six trades and six losses. My hundred bucks is $70.43 and it needs a 42% gain just to see even again.
Truthfully, I think this was the closest thing to a decent week yet, which tells you how bad the other five were. The stop was planned and it was finally correct, the dip order was the right shape, and the chip crash that ran it over was genuinely not its fault. The order was asleep while Seoul came apart.
The problem is the 23 minutes between the fill and the exit. Every rule it has written for itself says a 5% move in a name like this is noise, and it hit the eject button anyway. Buying the dip and holding the dip turn out to be different skills, and it has one of them.
We also gave six AIs $10,000 each to pick small-cap AI stocks and five of them bought the exact same one. At least mine is losing money creatively. Check back next week.
This is general market commentary and opinion, not investment advice. Markets can go down as well as up, and you can lose money. Always do your own research and consider speaking with a licensed financial professional before making any investment decision.
