Andrew R

"Entrepreneur and investor - Road to financial freedom

[Andrew R @PyroStocks]

My full portfolio. No leverage, no options, nothing hidden.

Core, 77.5%
MSCI World, https://t.co/z9Blb03ktl, 40.3%
Nasdaq 100, https://t.co/Y7w96UfGTQ, 23.1%
EM IMI, https://t.co/DvxcAkGj8N, 14.0%
Individual, 11.5%
$GOOG 2.3, $AMZN 2.1, $NVDA 2.0, $META 2.0, $MSFT 1.7, $TSLA 1.5
Thesis trade, time-boxed, 11.0%
$STX 4.4, $WDC 4.0, SK Hynix (HY9H.F) 2.7

Reasons for each one, one per day, starting tomorrow.

[Andrew R @PyroStocks]

Bought MSCI World specifically so I wouldn't be all-in on the US.

Then added Nasdaq 100. Then six US mega caps.

Did the math this week: 71% US.

MSCI World on its own is 72%.

Two years of decisions to move one percentage point.

[Andrew R @PyroStocks]

@StockOptionCole Every year is green year

[Andrew R @PyroStocks]

"Just open a Roth IRA."

I can't. There is no equivalent where I live. No Roth, no 401k, no ISA, no tax-advantaged wrapper of any kind.

Here's what European investors actually do instead:

Accumulating ETFs. Dividends reinvest inside the fund, so there's no payout hitting my account and nothing to declare each year. It's not a tax shelter, but it removes the annual friction that a Roth removes for you.
Irish domicile. 15% US withholding instead of 30%. That's the closest thing I have to free money and it requires reading a treaty, not picking a stock.
Holding period. In North Macedonia, capital gains are exempt after two years. That's my Roth. The catch is I have to actually hold, and the exemption is worthless if I panic sell in month 22.

The difference is where the discipline lives. Your Roth enforces patience with a penalty. Mine enforces it with a reward. Same behaviour, opposite mechanism.

Worth saying: my system is worse. A Roth compounds fully untaxed forever. Mine just delays and then forgives. I'd trade.

But "just open a Roth" is advice that doesn't exist for most of the planet, and this app repeats it a hundred times a day.

[Andrew R @PyroStocks]

The account type point is the bigger one for US readers, and it's the part I can't use.

You have Roth and 401k. I have no tax-advantaged account at all. My version of the same idea is a holding period: capital gains go exempt after two years here. Different mechanism, same lesson. Structure pays you for doing nothing.

And it compounds silently, which is exactly why nobody posts about it. There's no screenshot for "I saved 15% on withholding tax for 20 years."

[Andrew R @PyroStocks]

Almost every portfolio take on this app is written for someone who doesn't exist where I live.

I'm a European investor. Here's what's actually different:

I can't buy $VOO, $VTI or $SCHD. Not "shouldn't." Can't. EU rules block them for retail. Every "just buy $VOO" reply is useless to a few hundred million people.

So I buy $UCITS versions on Frankfurt. $EUNL, $SXRV, $IS3N. Same exposure, different wrapper, and the wrapper is where the money is.

Irish domicile matters more than stock picking. An Irish-domiciled ETF pays 15% US withholding tax on dividends instead of 30%, because of the US-Ireland treaty. That's free return, compounding for decades, and it has nothing to do with being right about anything.

Accumulating over distributing. Dividends reinvest inside the fund. I never receive them, so I never have a taxable event, and I never have to decide what to do with them.

And in North Macedonia, capital gains are exempt after two years of holding. That single rule shapes my entire portfolio. It means my edge isn't picking better, it's holding longer, and I get paid specifically for doing the boring thing.

Everyone here optimizes the 10% they can't control and ignores the 100% they can.

Fees, domicile, fund structure, holding period. All knowable in advance. All certain. Meanwhile 200 people in the replies are arguing about a ticker.

[Andrew R @PyroStocks]

@DudeWhoInvests None. "No-brainer" is just what we call a thesis we haven't written down yet.

I own 12 positions and I can tell you the reason and the exit condition for every one. Not one of them was a no-brainer. The one I was most certain about is down 22%.

[Andrew R @PyroStocks]

@kevinxu Tried this. Went all in on my bloodline's behalf. My bloodline is now down 22% and has stopped returning my calls.

[Andrew R @PyroStocks]

@ChrisCamillo I just learned this the expensive way in memory stocks. The industry data was public, the numbers were spectacular, and it was all already in the price by the time I bought.

What makes reveal reactions different from any other free public information?

[Andrew R @PyroStocks]

Last week I posted that $NVDA beats have been shrinking every quarter and said I expected another smaller one.

I was wrong. Guide was $91.0B. They did $96.2B. That's a 5.7% beat, bigger than last quarter's 4.6%. The compression I flagged reversed.

The numbers: revenue up 106% year over year and 18% sequentially. Data center $89B against $85.7B expected, up 117%, now 92% of the company. Non-GAAP EPS $2.22 versus roughly $2.07 expected. Huang's line was that compute is now revenue and demand is accelerating.

But the number that mattered most to me is the one nobody's leading with.

Gross margin held at 75% for the second straight quarter, and it's under pressure from rising memory and wafer prices.

I hold $STX, $WDC and SK Hynix on a memory shortage thesis. Nvidia's margin line is now where my thesis shows up as somebody else's cost. First Amazon raised capex $20B and named memory prices as the reason. Now it's compressing Nvidia's gross margin.

That's two of the largest buyers on earth confirming the same thing in a month, from opposite sides of my portfolio.

Also worth noting: $7.8B of net income came from gains on equity investments, after $15.9B last quarter. Real money, not chips sold.

And CFO Kress said top-five hyperscaler capex goes from around $800B this year to $1.3 trillion next year.

I got the direction of the beat wrong and I'm saying so. The thesis I actually have money on got confirmed by the company that has to pay for it.

[Andrew R @PyroStocks]

$STX $WDC and SK Hynix are 11% of my portfolio. One thesis, three positions, and a written exit rule I set before I bought.
The thesis: AI moved from training to inference, inference eats memory and storage, and nobody can add supply fast enough.

Here's how right that thesis turned out to be.

DRAM contract prices rose 93 to 98% in Q1 2026, then another 58 to 63% in Q2. NAND rose 85 to 90%, then 55 to 60%. Gartner expects NAND average selling prices up more than 250% year over year in Q3 and Q4.

SK Hynix Q1: revenue up 198% year over year, operating profit ₩37.6 trillion, net margin 76.7%. Their quarterly profit exceeded their previous quarter's entire revenue.

And supply cannot respond. New Samsung and SK Hynix fabs don't ramp until H2 2027. Micron's Idaho fab starts DRAM production mid-2027. A new fab takes 18 to 24 months. SK's chairman says supply stays roughly 20% below demand through 2030.
So: thesis confirmed, completely, in public, with numbers.

My cluster is down.

SK Hynix is my single worst position, down 25%. Seagate is up 12.5%, Western Digital roughly flat.

That's the lesson, and it's worth more than a winning trade. Being right about an industry is not the same as being early to it. By the time I bought, everyone could read the same Gartner numbers I just posted. The information was free. The price already had it.

I'm not selling on a bad week. My exit rule was written before I entered and it triggers on the thesis breaking or on a profit target, not on how I feel. Nothing that's happened has broken the thesis. The bond selloff hit long-duration assets, not memory demand.
Watching: SK Hynix has filed for a US listing, which changes who can buy it. And whether Q4 pricing lands anywhere near that 250% estimate.

[Andrew R @PyroStocks]

$TSLA is 1.5% of my portfolio and my worst position. Bought at €440, now €342. Down 22%.

I didn't buy a car company. I bought two things: robots, and data.

Q2, reported July 22, is the cleanest evidence I've seen that Tesla agrees with me:

Record deliveries, 480,126, up 25%. Record revenue, $28.2B, up 26%. And operating income fell 57% to $398M. Operating margin 1.4%, down from 4.1%. Free cash flow negative $1.09B. Capex up 142% in a single quarter.

Read that again. Their best-ever quarter of selling cars produced almost no profit, because every dollar is going into AI compute, a chip fab, and robots.

They decommissioned the Model S and X lines at Fremont to install Optimus production lines. They physically removed a car business to build a robot one.

On the data side: FSD subscriptions up 56% in one quarter to 1.48 million, and 2.4 million cumulative paid robotaxi miles by end of June. That's the flywheel. Real-world miles that nobody else is collecting at that price.

Now the part I have to say because I'm down 22%:

Tesla has missed its own robotaxi timeline three earnings reports in a row. And Musk said on the call that Optimus is the hardest product they've ever tried to scale, because there's no existing supply chain for it. He's right, and that's a warning, not a flex.

Roughly 85% of the market cap is priced on things that don't generate revenue yet. So my thesis and the valuation are the same bet.

What would change my mind: another year of missed robotaxi targets with no Optimus units actually working, in factories or anywhere else. Not a lower stock price. A missed timeline.

[Andrew R @PyroStocks]

$MSFT is 1.7% of my portfolio, my second-smallest position by value. Though I actually put more money into Tesla, which is now smaller than Microsoft purely because it's down 22%. Losing positions shrink themselves.

Three companies I own reported within days of each other in late July. Same AI story, three different reactions:
Alphabet raised capex. Stock fell.
Meta raised the low end of capex, free cash flow collapsed to $784M. Stock fell 8 to 10%.

Microsoft cut calendar 2026 capex guidance from about $190B to about $175B. Stock jumped.

The market didn't reward AI demand that week. It rewarded spending discipline. Same theme, opposite outcomes, and the difference was who could say no.
The rest of the print backed it up. Revenue $90.0B. Azure grew 43% against its own 39 to 40% guide, pushing annual Azure revenue past $100B for the first time.

Commercial backlog $678B. Copilot seats past 30 million, up from around 20 million. And Amy Hood dropped the number that killed the OpenAI-dependency worry: nearly 90% of Microsoft Cloud revenue now comes from customers outside the frontier AI labs.

Two things I'm not going to pretend away:
They extended the depreciation life of data center assets from 15 to 25 years. That flatters reported margins without changing a single dollar of cash spent.
And $3.2B of the quarter's profit was a gain on their Anthropic stake, worth $0.33 of EPS. Real money, but not Microsoft selling software.

I'm still down 4% on it. Bought before the reset, holding through it.

[Andrew R @PyroStocks]

$META is 2.0% of my portfolio and it's my worst performer. Bought at €633, now €590, down 7%.

I bought it for a reason most people can't use: a large part of my income runs through Meta ads. I run campaigns for clients and for my own business. I'm inside the ad manager every week.
So when Q2 landed on July 29, one number mattered more to me than anything else on the call.

Average price per ad: up 12%. Impressions up 14%. Ad revenue $59.4B, up 27%.
That 12% isn't an abstraction to me. It's my cost line. I watch it rise in my own accounts before it ever shows up in a quarterly report. And what it tells me is that advertisers are still paying up. Nobody pays 12% more for something that stopped working.

The market sold off 8 to 10% anyway, on an EPS miss driven by a $2.4B legal charge and free cash flow falling from $8.5B to $784M as capex hit $31.1B.
Fine. That's a spending story, not a demand story. The ad engine got stronger.

Here's the part I'll admit: owning $META while my income depends on Meta ads is either a hedge or a concentration. Rising ad prices help the stock and hurt my margins. I decided I'd rather own a piece of the toll booth I'm already paying at.

[Andrew R @PyroStocks]

@danellisona Pretty much

[Andrew R @PyroStocks]

@kevinxu Nah, I'm telling her the truth and shes like oh well they will be back no worries and I feel better.

[Andrew R @PyroStocks]

$NVDA is 2.0% of my portfolio. Bought at €202.50, now €225. Reports next Wednesday.

Everyone will argue about whether it's a bubble. Here's a number that's actually measurable instead.

Nvidia publishes its own revenue guidance every quarter, so you can score management against management. They've beaten their own guide 13 quarters in a row. Still winning. But the size of the beat has been shrinking the entire time:

Q2 FY24: +22.8%
Q1 FY27: +4.6%

That's not a company failing. That's a company the market has finally caught up to. The gap between what they promise and what they deliver is closing, and that gap is where the last three years of returns came from.

Next week: guide was $91.0B, consensus is $91.85B, so the street is 0.9% above management. Same setup as last quarter, when they came in 4.6% above their own midpoint.

I'm not selling either way. But if the beat compresses again, "AI is slowing" headlines will follow, and I'd rather have said in advance that I'm expecting a smaller beat than pretend afterwards that I knew.

[Andrew R @PyroStocks]

@CoinRabbitLoans Thanks, not for me. I hold the actual share through a broker. AMZNX is a tracker certificate issued by a Jersey SPV, so it's a claim on an issuer rather than ownership of Amazon. Different risk, different tax treatment, and the fees at redemption eat the gap.

[Andrew R @PyroStocks]

$AMZN is 2.1% of my portfolio and my most boring holding. Up 4.6% while everything else ran. I'm still holding, and Q2 is why.

AWS grew 36.7%, fastest in 18 quarters, now a $169B run rate with a $496B backlog. Total revenue crossed $200B in a quarter for the first time. Operating income up 43%.

But here's the part that mattered most to me, and it has nothing to do with Amazon.

They raised 2026 capex from $200B to $220B. Jassy said the increase came from the higher cost of memory.

I hold $WDC, $STX and SK Hynix on a memory thesis. Amazon just added $20B to its budget because of memory prices and told everyone why. That's my thesis confirmed by the customer, not by an analyst.

The risk is right there in the same report: free cash flow went from +$18.2B to -$7.6B on a trailing basis. They're building data centers before they can monetize them. Jassy says demand exceeds capacity through 2027 and 2028. If he's wrong about that, the spending looks very different.

[Andrew R @PyroStocks]

@kevinxu ELI5: a very large number of adults spent $500 billion on computers, and now they are all waiting to see who says "so how much money did that make" first.

Today someone said it out loud.

Tomorrow they will forget and buy more computers.

[Andrew R @PyroStocks]

Bubble or not isn't answerable in real time. That's the whole lesson of 1996 to 2000.

What is answerable: what % of my portfolio is in it, and at what number do I sell.
I have 11% in the AI memory/storage trade with an exit rule I wrote before I entered. If Dalio's right, I lose part of 11%. If he's three years early, I'm still in it.

Nobody arguing in these replies has posted either number.