What if the hardest part of trading isn’t finding good setups, it’s staying in them?

Most traders blow up their accounts the same way: they find a winner, get nervous watching it run, take profits too early, then watch it 3× from the sidelines. Or they panic on a headline, cut a position at a loss, and it reverses the next week.

The problem isn’t intelligence.
It’s emotion.

So we ran an experiment:

What happens when you remove the human?

We gave an AI $100,000 in June 2025, let it build a portfolio in one week, and then made ourselves a hard rule:

Zero intervention.

No trimming winners.
No panic selling.
No “hedging just in case.”

We locked our hands and watched.

Four months later? $165,000.

But here’s what matters more than that number:

If we’d just bought sector ETFs with the same allocations, we’d be sitting at $125K.

The AI picked individual stocks inside those sectors that crushed the benchmarks by 40% — not because it “called the market,” but because it stayed disciplined in places where humans usually crack.

This isn’t a backtest.
This isn’t cherry-picked.

This is what happened when we let structure beat emotion, in real time, with real trades.

The Experiment Rules (This Is the Whole Point)

• Starting Capital: $100,000 (June 2025)
• Portfolio Built By: Stonki (June 5–12, 2025)
• Human Involvement After: 0%
• Current Value: $165,000 (Oct 24, 2025)
• Total Gain: +65% in ~4 months
• Benchmark (Sector ETF basket): +25%
• Alpha (Outperformance): +40%

Here’s what that means in dollars:

If we’d just bought the sector ETFs in the same weights Stonki used, we’d be sitting at ~$125K today.

Instead, we’re at ~$165K.

That $40K difference is what I call the “AI premium” — and it’s the only thing worth talking about.

The rest of this article breaks down exactly how that happened, name by name.

Step 1: How We Built It

When we launched Stonki’s beta on Discord, we had one question:

Can an AI designed to think like a disciplined trader actually beat a human reacting to headlines?

So we gave Stonki a simple brief:

• Build a diversified, long-term portfolio
• Show us your entries, risks, and reasoning
• After that, we’re hands-off

In one week (June 5–12), Stonki selected nine positions:

AMD QCOM STNG HOOD BABA APLD ASTS MSTY TLT

This wasn’t random stock picking.

For each name, Stonki delivered a full memo:

• Price context
• Catalysts
• Risk parameters
• Position sizing

It doesn’t just suggest tickers — it thinks in portfolio construction.

Final Allocation Logic

• Semis: AMD, QCOM
• Energy / Shipping: STNG (cash-flow play)
• Fintech: HOOD (high beta)
• AI Infrastructure: APLD (compute narrative)
• China Recovery: BABA
• Space / Comms: ASTS
• Tactical Yield: MSTY (BTC-linked income)
• Defensive Hedge: TLT (23% allocation in bonds)

Not “YOLO into AI hype.”

More like: build something that survives.

Step 2: The Results

Raw P/L as of October 24, 2025

Eight out of nine names were green.
Several were massively green.

The one loser — MSTY — is actually important.

Stonki warned us in advance exactly why it might underperform, even while including it. That matters, because it shows this wasn’t blind faith.

It was a calculated risk.

But here’s the real question every trader asks:

“Okay, but isn’t that just because semis and AI pumped this year?”

Fair.

Let’s do the grown-up comparison.

Step 3: AI vs. Sector ETFs

(Where the 40% Alpha Comes From)

Stonki didn’t just pick stocks — it picked them inside specific sector buckets.

Here’s how we allocated across sectors (using ETFs as proxies):

• SMH (Semis): 43%
• XLE (Energy): 4%
• XLF (Financials): 10%
• XLC (Communications): 3%
• XLK (Tech): 10%
• KWEB (China Internet): 3%
• MSTY (Tactical Yield): 3%
• TLT (Bonds): 23%

Nearly a quarter of the portfolio sat in 20+ year U.S. Treasury bonds.

Now compare Stonki’s picks to the ETFs:

The Killer Insight

Same starting $100K.
Same sector weights.
Same four-month window.

• ETFs: ~$125K
• Stonki picks: ~$165K

+ $40K

That’s the 40% alpha.

Stonki’s edge wasn’t knowing AI would run — everyone knew that.

The edge was:

• Picking very specific names
• With very specific catalysts
• And letting them run without flinching

Step 4: How It Actually Thought

(This Part Made Me Take It Seriously)

Here’s what convinced me this wasn’t luck.

Let’s look at three trades and the actual reasoning Stonki laid out before the positions went on.

BABA (June 10, 2025)

Entry: $120.88

What Stonki Saw:

Stonki’s Plan

Entry: Above $122 with volume confirmation
Allocation: 70% stock / 30% July $125 calls
Risk: Stop on daily close below $119.50
Targets: $125 / $127.50 / $134

Catalysts Flagged

• Rumored Apple–Alibaba AI partnership (80% confidence) → transformative for BABA Cloud
• Easing U.S.–China tensions
• Potential Beijing stimulus package

That’s not “LLM vibes.”

That’s structure, flow, macro, and narrative converging.

What Actually Happened

STNG (June 6, 2025)

Entry: $40.29

What Stonki Saw

Classic value thesis — the kind most day traders overlook because it’s not “sexy.”

What Actually Happened

AMD (June 10, 2025)

Entry: $121.73

What Stonki Saw

If you’re an options trader, you know exactly what that line means: stop hunting tops, just sit in the obvious secular bid.

What Actually Happened

MSTY (The One That “Lost”)

Entry: ~$21.20

MSTY is a covered-call ETF tied to MSTR / Bitcoin.

Stonki flagged it as a “conditional yes” — good for yield during sideways crypto action, but with capped upside if BTC rips.

Stonki’s Warning (verbatim)

“Capped upside in bull markets: if BTC/MSTR rip, MSTY underperforms because calls get exercised away. High correlation risk. Use as income during chop, rotate out if BTC breaks trend.”

Then Bitcoin ripped.

The position shows -45.69% on paper.

But here’s the important part: MSTY is a dividend-focused strategy. The stock price “loss” was largely paid out as dividends throughout the holding period — that’s how covered-call ETFs work.

The value didn’t disappear.
It was distributed.

This position was actually inspired by a Reddit strategy that involved buying monthly put protection. But because we committed to zero active management after the initial build, we didn’t execute the hedge side.

So while the raw P/L shows red, the real return including dividend income would look very different.

Still, Stonki’s core prediction held:

If BTC rips, MSTY underperforms in capital appreciation.

It called the exact risk that materialized.

For me, that’s critical.

I don’t want “our AI is magic.”
I want:

“Our AI told us what it believed, why, what could go wrong — and then reality lined up with that model, including the risk case.”

That’s a trading partner.
Not a black box.

Step 5: Why This Actually Matters If You Trade

Here’s what I personally took from this whole experiment.

Humans Chase Certainty. AI Chases Structure.

Every trader I know has done the:

“This is up 20%, I should lock it in before it dies.”

Stonki didn’t.

It stayed in HOOD (+90.16%).
It stayed in APLD (+145.62%).

It didn’t scalp winners just to feel safe.

Consistency Beats Cleverness

Was APLD an obvious “AI infra + crypto compute” narrative?

Yes.

Did most people actually size it properly and sit on it?

No.

They traded around it 30 times.

The alpha was in holding.

Hedging Matters

23% of the portfolio sat in TLT.

Stonki did that on purpose.

It wasn’t trying to be a hero.
It was building something that survives, not something that screenshots well.

What Stonki Actually Is (And Isn’t)

This isn’t “we’ll trade for you.”

Stonki isn’t a broker.
It doesn’t press buttons for you.

What it does is the part most traders are worst at:

• Builds the plan
• Shows the thesis
• Sizes the risk
• Tracks the catalysts
• Refuses to panic

You’re still the one executing.

But you’re no longer improvising emotionally.

If you’re a trader — especially in stocks and options — you already know:

The hard part isn’t finding trades. It’s sticking to them.

This experiment was just proof of that, in numbers.

What’s Next

We’re taking everything that worked in this challenge and baking it into the Stonki Web App:

• ✅ Generate full trade Recipes (what to buy, why, size, risk, catalysts)
• ✅ Track performance vs. the right benchmark (not just “up or down,” but “did it beat its sector ETF?”)
• ✅ Get reminded when the thesis is still intact — so you don’t bail just because Twitter is loud that day

The biggest thing I want you to take from this isn’t:

“Wow, AI made 65%.”

It’s this:

This approach was boring.
It was slow.
It was calm.
It was rules-based.

And it was survivable.

That matters more than any one name in the portfolio.

Because at the end of the day, Stonki didn’t outsmart the market.

It just stayed rational in places where humans usually don’t.

And that, for me, is the edge.

Final Notes

We’ve taken everything we learned — the data, the discipline, the clarity — and built it into a platform that helps you trade with structure, not emotion.

If you made it this far, thank you for reading.

This experiment started as a small idea in a Discord chat and turned into one of the most insightful projects we’ve ever done.

Watching traders react to what we’ve built, learning from your feedback, and building alongside this community is what keeps us going.

Whether you’ve been with us since beta or just discovered Stonki today — thank you for being here.

We can’t wait to show you what’s next.

Data Note

All numbers as of October 24, 2025.
This was a $100,000 paper trading account managed by Stonki’s AI from June 5–12, 2025.
No human changes were made after the initial allocation.

Disclaimer

This article is for educational purposes only and does not constitute financial advice.
Trading and investing involve risks. Past performance is not indicative of future results.
Please conduct your own research or consult a financial advisor before making investment decisions.

Portions of this article were generated with AI assistance and reviewed for accuracy.