the nangoy foundation
A personal record of one small book, run patiently.
august 2026 · book as of 2 september 2026
Going into August, the S&P 500 and the Dow sat at record highs, fueled by the AI complex and the optimism around the revolutionary technology. Then the data started to complicate the picture. The first week brought a disappointing July jobs report — payrolls fell by 23,000 — and the rate-cut case began to build. It was further supported by cooling inflation, with headline CPI at 3.4% and core CPI at 2.5%; by mid-month the market was leaning toward a September cut. That leaning reversed at the end of the month. On August 26, the Fed’s preferred inflation gauge, core PCE, came in a touch hot at 3.3%. The next day, NVIDIA’s blowout earnings briefly carried the tape higher — but on Friday the 28th, Kevin Warsh, in his first Jackson Hole address as Fed Chair, made clear the Fed’s job of getting inflation back to its 2% target is not done, calling price stability its predominant focus. Yields pushed higher and the market turned down, the S&P 500 falling 0.25% and the tech-focused Nasdaq losing 0.52%. In the space of a week, the market went from pricing a September cut to pricing a possible September hike, with those odds jumping to around 57%.
Geopolitically, the earlier ceasefire with Iran broke down in July, and an active flare-up near the Strait of Hormuz followed in late August, keeping uncertainty around the chokepoint. Oil eased over the month even so, though a Hormuz risk premium lingered. Combined with the macro data, this kept the 10-year Treasury choppy — roughly flat-to-firmer on the month, sitting around 4.72% as I write this on August 29th.
Cash & T-bills combines the cash balance with the SGOV position — the true dry powder, just under a third of the book. Emerging Asia is Tencent Music and Taiwan Semi; International developed is the VEA holding.
VEA Vanguard Developed Markets · 13.3%The un-clever part of the book on purpose: cheap, broad exposure to developed markets outside the US, and my bet that the whole world isn’t priced like the S&P.
MCO Moody’s · 11.7%A toll booth on global debt issuance, a ratings duopoly with pricing power I’d love to own cheaper. Holding, not adding, at today’s multiple.
AMZN Amazon · 7.0%Really an AWS bet now, and I’ll be honest, the quality work isn’t finished. The open question — can I actually know AWS’s returns on the capital it’s pouring in, or am I guessing — is one you have to ask of every hyperscaler and most of the AI spenders, and it’s genuinely hard to answer. It probably won’t be, until time answers it for us.
TME Tencent Music · 6.0%Priced for the business to shrink, which it isn’t. It’s cheap enough that the margin of safety survives real pessimism. The risk that matters isn’t the numbers, it’s the China ownership structure behind it.
TSM Taiwan Semiconductor · 5.1%The company the whole AI complex quietly runs on, and an elite compounder on returns. I own it, I’m not chasing it up here, and I’d want more on a broad selloff.
WMT Walmart · 5.0%A long-term compounder I’m comfortable owning, which is why I added on the drop rather than flinching at it. The stock fell last quarter on disappointing comparable-sales numbers, but that was drug-price caps, not weak demand. Still owe myself the full valuation work, but this is a business I’m happy to hold.
GOOGL Alphabet · 4.5%I own a small position, and it’s the honest oddity of the book: my own valuation work says pass at this price. I’m very bullish on Search and ads, especially if Google can integrate AI into them well — but ‘well’ is a big ask, and that’s why the position is small. Same unanswerable question as Amazon, and I’m now reconsidering whether the hyperscaler model wins the AI revolution at all. Work for September.
MCD McDonald’s · 3.5% · newDown ~24% to an 11-year-low multiple on a de-rating, not a broken business. Most of the growth ‘slowdown’ is a smaller buyback by choice, not the P&L. Started small while I finish the work; the crux is whether guest traffic can carry growth now that the pricing and margin levers are spent.
PGR Progressive · 3.1%A best-in-class insurer, and the combined ratio is why: it runs in the mid-80s — 86.8 in July — where most of the industry scraps just to stay under 100. I’m building it in tranches: small starter in, adding only at lower prices, and I stop cold if that combined ratio ever pushes past 96%.
Since inception on 3 July 2025, the book has returned +22.4% on a time-weighted basis — the clean measure of how the portfolio itself performed — against +24.2% for the S&P 500 over the same stretch. A shade behind the benchmark. Against a wider field it sits mid-pack: ahead of Hong Kong and China, behind the Nasdaq, Nikkei and Singapore.
Time-weighted return since inception (3 Jul 2025 – 27 Aug 2026). The month-by-month track record builds from here.
for context — other markets, same period
I’ll be upfront: NVIDIA is stock of the month not because it went up, but because the quarter was genuinely a monster and because it captures the thing I keep circling back to about this market.
The numbers, briefly. Revenue came in at $96.2 billion, up 106% from a year ago, with the data center segment alone at $89 billion. Gross margins actually widened to 75%. Management then guided the next quarter to $108 billion and, more striking, pointed to something like 70% revenue growth for the following year, nearly double what the street had penciled in. On the call they also took the “circular financing” question head-on. Jensen Huang pushed back on the label, saying OpenAI would pay its lease and that its buildout could mean roughly $600 billion of NVIDIA compute over the coming years. But he also said something that stuck with me: that the frontier AI labs are “growing faster than their balance sheets and long-term credit profiles can support.” NVIDIA is now helping arrange over half a trillion dollars of outside capital for its own customers to solve exactly that. Read that twice.
So the business isn’t the problem. The problem is the price, and what the price now demands. Near $228, the stock isn’t paying you for a great quarter; it’s paying for years of near-flawless execution on top of hyperscaler spending that has to keep compounding at a pace nobody has actually guided to. My own work puts fair value well below here, and this quarter, as good as it was, doesn’t close that gap. It mostly raises the bar for what has to keep going right.
And this is where NVIDIA stands in for something bigger. I think the AI overvaluation is real, and not only for NVIDIA — it runs across the whole complex. What worries me isn’t whether I’m comfortable owning one beautiful, performing business. It’s that when sentiment turns, the correlations inside the AI complex go to one. These names don’t fall politely, one at a time. They fall together. Buffett had the image for it: everyone’s at the ball, having too good a time to leave, all planning to slip out just before midnight, except there are no clocks on the wall and the exit is narrow. The music is still playing. I just don’t intend to be the one still dancing when it stops.
Which brings it back to my own book. NVIDIA is about 8.8% of the portfolio, but that understates the real exposure. Once you count Taiwan Semi, Alphabet and Amazon, the AI-linked names are roughly a quarter of everything I own. For risk purposes, that’s one position, not four. So my move here is discipline, not conviction: I’m not adding to NVIDIA or the complex at these levels, I’m treating the cluster as a single bet and keeping it capped, and I’d sooner hold cash and look boring than crowd toward that narrow door with everyone else. If the thesis I care about breaks, it won’t break in NVIDIA’s income statement. It’ll break in the price of the whole group, all at once.
The one real change this month was selling out of QQQ. The problem with it was overlap: it doubled down on the AI names I already own outright — Nvidia, Taiwan Semi, Alphabet, Amazon — so it wasn’t a separate bet, just more of the same correlation. Selling it is the same discipline as this month’s NVIDIA note: treat the AI complex as one position and don’t let it quietly grow. It also raises cash, which is exactly what I want with valuations where they are.
I also started a small McDonald’s position and added to Walmart; the reasoning for both is in the roster above, so I won’t repeat it here.
For now the proceeds simply sit in cash. I’d rather hold it and wait for a price than redeploy into more of what is already expensive.