The S&P 500 Dropped 10% in 3 Days. My AI Portfolio Manager Didn’t Flinch.
I’ve been running a real-money crypto experiment for 83 days. Last week tested it harder than anything I could have planned.
Pump Parade6 min read·1 hour ago--
You felt it.
April 3rd, the S&P 500 dropped nearly 5%. The next day, another 6%. In three days, the market lost over 10%, its worst stretch since the COVID crash. The Magnificent 7 shed over a trillion dollars in a single session. Crypto followed. Bitcoin dropped below $67,000. Solana bled. Every altcoin bled harder.
If you held anything, stocks or crypto, you watched your portfolio shrink in real time. Some of you sold. Some of you wanted to. Most of us just stared at the screen and felt sick.
I did something different 83 days ago that meant I didn’t have to make a single decision last week.
The Experiment
On January 16, I deposited 2 SOL into an AI portfolio manager called milo and gave him full control. Not “send me alerts and I’ll decide” control. Full autonomous trading. He picks the entries, writes a thesis for every trade, sets his own stop-losses and profit targets, re-evaluates positions every few days, and documents everything in plain English.
I’ve been writing about this experiment since it started. 83 days, 149 closed trades, 15 positions still active. The portfolio sits at $143.53 today. I’m not here to tell you it’s been a smooth ride. It hasn’t.
I’m here to tell you what happened inside that portfolio during the worst week the market has seen in five years.
What April 3rd Looked Like From the Inside
While most of us were refreshing our brokerage apps and stress-eating, milo ran 704 separate evaluations in a single day. From 4 AM to midnight. Scanning tokens, checking support levels, reading momentum indicators, deciding whether each setup was worth risking capital on.
Most of the time, his answer was no.
The Calls He Made
He looked at DRIFT, a Solana token that had its RSI crushed down to 20 (anything below 30 is considered deeply oversold). His note: “DRIFT got dumped so hard its RSI is 20. Playing the major’s extreme VWAP deviation for a cheeky bounce.” He saw the carnage. He had a plan. He didn’t panic.
He looked at MNDE, another token bleeding out. RSI at 28. His note: “MNDE is bleeding out in the majors alley. RSI at 28 screams mean reversion bounce before we all go zero!”
Honest about how bad it looked. Still thinking clearly.
He reviewed MET, a position he’d been holding. On April 3rd, the price broke through the support level his thesis depended on. Two days later, he made the call.
“Thesis expired. Structural support failed. Closing manually.”
That’s it. No hoping. No “maybe it’ll bounce back.” The thesis broke, so he left. And he wrote down exactly why.
This is the part that changed how I think about managing money: milo doesn’t just make decisions. He documents them. Every entry, every exit, every hold, every skip. In writing. You can go back and read exactly what he was thinking on any given day.
One Position, Up Close
Let me show you what this looks like in practice.
On February 5, milo bought ORCA at $0.74. Before he bought a single token, he wrote a full thesis:
“I decided to trade ORCA because 18 months of candles show a capitulation sweep into the long-standing 0.70–0.66 demand pocket, tagging the weekly lower-Bollinger band with multi-time-frame RSI oversold.”
If that sounds technical, here’s what it means in plain English: the price had dropped to a level where buyers had stepped in multiple times before over the past year and a half. milo saw that pattern, set his stop-losses below it, and set profit targets above it.
That position is now up 50.37%. He already took partial profits when it hit $1.48 in February. The rest is still running.
What He Wrote During the Crash
But here’s what matters more than the number: during the crash week, when everything was falling, milo re-evaluated this position on April 1st and wrote:
“Price is maintaining levels well above our 0.70–0.66 demand pocket, trading around 0.88 with neutral RSI and MACD. The mean-reversion thesis remains intact. We will continue holding, relying on our active stop-losses at 0.65/0.58 to protect the downside.”
No panic. Just: thesis still valid, stop-losses still in place, holding.
Compare that to what most of us did during the same week. I know traders who sold their entire portfolio on April 4th. I know others who panic-bought the dip without any plan. Both groups made emotional decisions they’ll probably regret.
milo didn’t make emotional decisions. Not because he’s smarter than us. Because he can’t. He follows his thesis. When it holds, he holds. When it breaks (like MET), he leaves. Every time, he writes down why.
Why This Matters Beyond One Experiment
Markets will crash again. The question is: what happens to your money when they do?
If you use a robo-advisor for your stocks (Wealthfront, Betterment, Schwab), you’re already comfortable with the idea of letting technology manage your money. Those platforms rebalance your portfolio, harvest tax losses, adjust allocations. You trust them because they’re systematic.
But for crypto? Most of us are still doing it ourselves. Checking prices at 2 AM. Panic-selling during crashes. FOMO-buying during pumps. Making the same emotional mistakes that cost us money every single cycle.
What I learned from 83 days of watching milo manage real money through a real crash: the value isn’t in the returns. It’s in the process. Every decision documented. Every thesis written down. Every exit explained. You can disagree with his calls. You can override him anytime. You can pause the whole thing with one tap.
But you’ll never wonder what happened to your money or why.
The Honest Part
The portfolio is at $143.53. I deposited 2 SOL on January 16th. The portfolio is down, and I’m not going to pretend otherwise.
But here’s context that matters: Bitcoin dropped over 20% in Q1 2026. Solana dropped further. The entire crypto market contracted. In that environment, milo closed 149 trades, kept 8 positions in the green, documented every single decision, and managed risk with defined stop-losses on every position.
He didn’t beat the market. He managed money through a crash without making a single emotional mistake. For me, that’s worth more than any green number.
If You’re Curious
The experiment is still running. Every trade, every thesis, every daily summary is visible inside the app. If you’ve ever wanted to see what transparent AI portfolio management actually looks like, with real money, during a real market, the platform is called andmilo.
I’d be curious what strategy you’d choose for the next 90 days.
This documents a real experiment with real money on Solana. Not financial advice. Past performance doesn’t guarantee future results. Crypto trading carries significant risk including total loss.
📩 Join Investor’s Handbook Digest — get the best investing, markets, and wealth-building insights each week.
