This Week in Stock Picks: What Finance YouTubers Bought (August 24 – August 30)

In the week of August 24 – August 30, 2026, the finance YouTube channels we track published 183 qualified calls across 22 different channels and 82 separate videos — more calls, more channels and more names than the week before. The buy share fell, though: 66% buys, down from 73%, with 33 avoids and 12 sells. Two things drove the week. Nvidia reported, and the board split nine ways on what the print meant — including one creator who bought it on Monday and warned people off it on Thursday. And a wave of 13F season coverage pushed the week’s only unopposed consensus, Visa and Mastercard, onto the board — not from anybody’s own analysis, but from copying somebody else’s filing. Here is what they actually bought, where they split, and which numbers to discount.

TL;DR

• Nvidia was the week’s most-covered and most-contested name: 12 calls from nine channels, nine buys against three avoids — and the bulls and bears are arguing about the same thing.
• Visa and Mastercard were the only widely-covered names with zero opposition — four buys each, from the same four creators. Three of those four trace back to a filing, not a model.
• The 13F trade was the week’s real theme. Three sessions built entirely on other investors’ disclosures produced 25 calls, 14% of the week, and account for most of the “consensus” below.
• Meta’s unopposed streak ended at three weeks — five buys, but also a sell and an avoid.
• One session filed nine avoids on its own — The Compound on 8/28, against Nvidia, Microsoft, Amazon, Alphabet, Meta, Oracle, CoreWeave, OpenAI and Anthropic. That is 27% of every avoid on the board this week from one video.
• Palantir drew three calls from three channels and not one of them was a buy.
• The surprise: a top-3 ranked creator spent his week away from AI entirely — a consumer-brands basket including Celsius, RH, Revolve, e.l.f. Beauty and Honest Company.
• The second surprise: two channels found the same non-obvious AI layer — optical components. Coherent got buys from two creators six days apart, one of them alongside Lumentum.
• Live ranking, updated twice a day: the Terminal.

The week in numbers

The Latest Stock Calls feed carried 183 calls from 22 channels across 82 videos, covering 117 distinct names — up from 171 calls and 20 channels the week before. The buy share fell seven points to 66%, and avoids rose from 20 to 33.

Concentration was extreme, and you should read every count below through it. Parkev Tatevosian, CFA filed 48 calls on his own — 26% of the entire week — across roughly 20 short single-stock valuation videos. Let’s Talk Money! with Joseph Hogue, CFA added 29 more, most of them from two basket videos. Between them, two channels produced 42% of the week’s board. A name with four buyers from four channels is a real signal. A name that appears twice because one creator publishes daily is not.

The week’s calls sorted into three poles: the AI complex, now openly split rather than merely contested; a defensive rotation into toll-booth businesses — payments, waste, home improvement, index providers — almost all of it sourced from 13F filings; and a long tail of single-creator value work that had nothing to do with either.

Nvidia reported, and nine creators disagreed about what it meant

Nvidia drew 12 calls from nine distinct channels — the widest coverage of any name this week, and the most contested. Nine buys, three avoids.

• Everything Money — Buy, 8/24 — a premium business with high returns on capital; even at 45x free cash flow and 33x earnings, his conservative model still implies roughly a 15% return.
• Let’s Talk Money! with Joseph Hogue, CFA — Buy, 8/24 — the “picks and shovels” framing; he expects the AI infrastructure spending boom to hold for at least 8–12 months.
• Parkev Tatevosian, CFA — Buy, 8/24 — calls it the best stock to buy right now, on full-stack AI infrastructure and physical AI as the next catalyst; and again on 8/27, this time arguing that Anthropic’s revenue projections and a potential IPO are a direct read-through to Nvidia.
• A top-3 ranked creator — Buy, 8/25 — undervalued on DCF, and explicitly positioned for the guidance to be sold: he expected a market overreaction to create the entry.
• BWB – Business With Brian — Buy, 8/25 — fair value $290, which puts it in his “buy full” band; cheaper against its own earnings than it has been for over 80% of the last ten years.
• Brian Stoffel — Buy, 8/26 — the cleanest number of the week. On a reverse DCF, after 65% growth for three years, Nvidia needs only 0.5% annual revenue growth in years 4 through 10 to justify today’s price.
• Jerry Romine Stocks — Buy, twice, 8/27 and 8/29 — “fortress monopoly”: ~75% gross margins, revenue doubled year over year, $279 billion in supply-chain commitments, net cash on the balance sheet.

Against that, three avoids that all land on the same mechanism.

A top-3 ranked creator (8/26) acknowledged the strong print and warned anyway: margins and growth are not sustainable at this level, he expects a slowdown in 2027, and he named a “massive amount of circular financing” as the reason. The Compound (8/28) put it harder — Nvidia is increasingly acting as a creditor to its own customers, who are themselves dependent on unprofitable AI startups.

And the one worth sitting with: Everything Money bought Nvidia on 8/24 and filed an avoid on 8/27. Three days. His second session names the “AI ROI chasm” — customer spending running ahead of the revenue it generates — plus major customers designing their own chips, and geopolitical execution risk. Both calls are on the board with dates. We do not smooth those out; a creator changing his mind inside one week is the signal.

The through-line: the bull case and the bear case are now the same fact. Nvidia’s role in financing the buildout is either the moat or the crack, depending on who is holding the model. Background: Nvidia on BullVox.

The only unopposed consensus was a copy of one filing

Visa and Mastercard were the week’s cleanest agreement — four buys each, from four different channels, with nothing filed against either. On raw spread that beats everything except Nvidia. But look at where the calls came from:

• New Money — Buy both, 8/25 — because Bill Ackman’s Pershing Square opened new positions at 5% of portfolio each; wide moat, 50%+ margins, and a third-party DCF showing Mastercard 47% undervalued and Visa about 12%.
• Arte de invertir — Buy both, 8/30 — same Ackman filing, his own framing: an oligopoly with 10–20% earnings growth, inflation-protected because the fee is a percentage of the transaction, trading below its historical average.
• Felix & Friends (Goat Academy) — Buy both, 8/24 — because Donald Trump’s disclosed portfolio bought them. “Toll booths” that take a cut of every transaction regardless of the cycle.
• Parkev Tatevosian, CFA — Buy Visa 8/27 (forward P/E 24.4) and Mastercard 8/28 (forward P/E 24.9, near-60% operating margins) — the only one of the four who got there from his own numbers.

That is the honest read of the week’s top consensus: four creators, one underlying decision. Three of them are reporting a filing; one ran a model. Spread across channels is usually the best proxy we have for independent agreement — this is the case where it isn’t, and it is worth naming.

The same three whale-driven sessions produced 25 of the week’s 183 calls (14%) and most of its defensive tilt: Home Depot (2 buys — Trump’s portfolio via Felix, Bill Gates’ via Arte de invertir, both leaning on a housing-demand deficit and a 21x earnings multiple), Republic Services, Berkshire Hathaway, gold and silver ETFs, and re-entries into Netflix after its 32% drawdown.

Netflix is the cleanest illustration: three buys and one sell, and every one of the four is somebody else’s trade. Ackman re-entered at 20–23x earnings (cited by both New Money on 8/25 and Arte de invertir on 8/30, the latter noting streaming is still only 46% of screen time); a top-3 creator called it an “obvious buy” at 25x forward, 23x on his own DCF; and Felix & Friends filed the sell on 8/24 because Trump’s disclosure showed a sale. Context: Netflix on BullVox.

Meta’s unopposed streak ended

Meta went three consecutive weeks without a single call against it. That ended: seven calls from seven creators — five buys, one sell, one avoid.

The buyers all converged on the same catalyst, and it is a new one. The settlement, not the capex:

• Daniel Pronk — Buy, 8/28 — the $18 billion settlement spread over a decade is non-material: 1.4% of annual operating cash flow. The $1.4 trillion headline number was never the real exposure.
• A top-3 ranked creator — Buy, 8/28 — “obvious buy”, undervalued, growing faster than Netflix, and he reads the settlement terms as favorable because they force TikTok and YouTube into comparable youth-safety requirements.
• Stealth Wealth Investing — Buy, 8/30 — legal headlines run worse than legal outcomes for big tech, and the settlement is small relative to the balance sheet.
• Value Investing with Sven Carlin, Ph.D. — Buy, 8/27 — raised to a 10–11% expected return after the price fell.
• Parkev Tatevosian, CFA — Buy, 8/25 — 38% operating margins, ROIC of 22.8% against a 10.7% WACC.

The two calls against it are unlike each other. Felix & Friends filed a sell on 8/24 — again, a Trump-disclosure sale, not his own thesis. The Compound (8/28) filed an avoid on grounds nobody else on this board uses: not valuation, not legal risk, but product quality — Meta as an example of a “rot economy” where growth metrics are pursued at the expense of the user.

Four weeks of near-unanimity, and the first two objections are a copied filing and a critique of management philosophy. Neither is a valuation bear. That one still hasn’t appeared. Background: Meta Stock Forecast: What Finance YouTubers Say.

One session, nine avoids

The single most consequential video of the week was The Compound’s AI-bubble session on 8/28, which produced nine avoid calls in one sitting: Nvidia, Microsoft, Amazon, Alphabet, Meta, Oracle, CoreWeave, plus OpenAI and Anthropic as private companies.

The argument is a single chain, applied down the stack. OpenAI and Anthropic are unprofitable and dependent on continuous capital raises; their “run rate” revenue is misleading because token spend is not recurring. The hyperscalers’ cloud growth is therefore partly borrowed demand — he claims a significant share of Microsoft’s AI revenue traces to OpenAI. Nvidia sells into hyperscalers who sell into the startups. CoreWeave is the leveraged end of it, “horribly unprofitable” and permanently refinancing GPU purchases at high rates. Oracle he calls a “dog of a company” with flat inflation-adjusted revenue for 15 years, now funding data centers.

Take it or leave it — but count it correctly. Those nine calls are 27% of every avoid filed this week. Strip that one video out and the board’s buy share climbs back to roughly 70%, right where it was the week before. Much of this week’s apparent turn toward caution is one person’s session.

He is not alone on the leveraged names, for what it’s worth: Joseph Hogue independently filed an avoid on CoreWeave on 8/30, citing $10 billion of free cash flow burn over the last year and a rental model exposed to technology obsolescence.

Palantir: three channels, zero buys

Palantir is the week’s clearest negative consensus — three calls from three channels, and not one is a buy.

• A top-3 ranked creator — Avoid, 8/26 — forward P/E over 100, and 57x even two years out; the growth is already in the price.
• Daniel Pronk — Avoid, 8/25 — and this is the fair version of the bear case. He grants the fundamentals outright: revenue nearly doubling, high free cash flow margins, a strong Rule of 40. He passes purely on 90x current-year free cash flow, where even aggressive growth assumptions leave a poor return.
• Felix & Friends — Sell, 8/24 — the Trump-disclosure sale again.

Nobody argued the business is bad. All three arguments are about price. Background: Is Palantir Stock a Buy?

Amazon split three ways, and the split is about Anthropic

Amazon drew seven calls from seven creators — four buys, three avoids — and unusually, both sides are looking at the same asset: the Anthropic stake.

Buying: Parkev (8/27) reads Anthropic’s growth and potential IPO as a direct benefit, since Amazon is both investor and compute supplier. New Money (8/25) notes Seth Klarman’s Baupost raised its stake 20%, to 16% of portfolio, on 37% AWS growth and a $496 billion contracted backlog. Arte de invertir (8/30) cites Druckenmiller and a valuation below its five-year average. Everything Money (8/25) is conditional: a buy at $230, against his $215 fair value at a 15% required return.

Avoiding: Learn to Invest (8/24) makes the sharpest point on the board — the headline 21x P/E becomes 38x once one-off gains from the Anthropic stake are stripped out, which is expensive against the index rather than cheap. Sven Carlin (8/27) has it fairly priced for 8%, below his hurdle, and would rather wait for a crash. The Compound (8/28) counts AWS growth as borrowed from the same startup spending.

One creator’s catalyst is another’s accounting adjustment, on the identical line item.

Two surprises

A top-3 creator skipped AI entirely. In a week where 22 channels mostly argued about hyperscalers, one top-3 ranked creator spent his 8/24 session on consumer brands: Celsius (20% US energy-drink share, international runway, new management), RH (affluent customer base, newly free-cash-flow positive), Revolve into the holiday season, e.l.f. Beauty with a $100–140 year-end range, and Honest Company after a strong earnings report. Two days later he was back on tech — buying AMD as the 2027 share-gain story against Nvidia, plus Salesforce on its Anthropic partnership and ServiceNow — while avoiding Nvidia and Palantir on price. Seasonal, momentum-led, and almost entirely disconnected from the week’s dominant argument.

Two channels found the photonics layer. Ticker Symbol: YOU (8/24) bought Coherent and Lumentum on a specific bottleneck: AI data centers need optical interconnect, and the constraint is lasers. Coherent runs proprietary 6-inch indium phosphide wafer production with capacity expansion underway; Lumentum’s electro-absorption modulated lasers grew revenue 109% year over year at 50.4% gross margins. Six days later, Joseph Hogue (8/30) independently arrived at Coherent, naming Nvidia’s investment and multi-year partnership as the de-risking event. Two channels, six days apart, on a name most of the board never mentions — that is the kind of quiet spread we weight more than a megacap pile-on.

Also on the board

• Micron — three buys, three creators, nothing against: BWB (8/25), Joseph Hogue (8/24, picks-and-shovels), Parkev (8/27, on the Anthropic read-through). Background: Micron on BullVox.
• Broadcom — three buys from two creators, including Joseph Hogue (8/30) explicitly recommending it on post-earnings dips.
• S&P Global — the strangest split of the week: the same Ackman position produced a buy from Arte de invertir (8/30, 23x, oligopoly in ratings and indices, 50% EPS growth over five flat stock years) and an avoid from New Money (8/25, whose DCF read it as ~7% overvalued). Parkev filed a hold and a conditional buy at $339–350 in one video on 8/24.
• Alibaba — two buys, one avoid, all on AI. Parkev upgraded it 8/24 to a $156 fair value; Sven Carlin cut his intrinsic value on 8/30 over negative free cash flow from AI capex and six years of unmet promises.
• Berkshire Hathaway — a rare direct contradiction: bought 8/24 (Trump’s disclosure, via Felix & Friends) and sold 8/27 by Sven Carlin, who says it is fully priced and will deliver 5–6% long-term — real, but not good enough.
• Genuine sells, five of them: Ciena at 70x price-to-free-cash-flow and Delta on $60 billion of debt (both Everything Money, 8/24); MongoDB on decelerating growth into a stretched multiple (Parkev, 8/27); Limbach (Daniel Pronk, 8/25) — the most honest sell of the week, because he states plainly that his thesis broke: he expected a Q2 rebound, guidance said otherwise; and Berkshire above.
• HKCM filed a set of Elliott-wave shorts that stand well outside the rest of the board: Sony down 60%, SK Hynix down over 70% after a 1,500% year, and Bitcoin shorted from around 90,000. In each case he intends to buy back after the correction. Directional, mechanical, and unlike anything else filed this week.
• Roblox — Rational Investing with Cameron Stewart, CFA (8/26) avoids it on $4 billion of stock-based compensation over five years, nearly a full year of revenue.

Who’s earned the mic

A buy call is only worth the track record behind it. We score every qualified call at least 7 days old against the price move that followed, and rank only creators with 20+ scored calls. Among the channels active this week:

• Felix & Friends (Goat Academy) — 69% accuracy over 1,605 scored calls, with a measured copy-portfolio return of +63.4% vs the S&P 500’s +32.8%. The best combination of sample size and margin among this week’s active channels — which is worth remembering, given that his entire session was somebody else’s portfolio.
• Invest with Henry — 65% over 212 scored calls, at +25.2% vs +14.2%.
• Jerry Romine Stocks — 64% over 232 scored calls, at +24.3% vs +13.2%.
• BWB – Business With Brian — 63% over 504 scored calls, at +52.7% vs +28.9% — one of the widest margins over the index on this list.
• Let’s Talk Money! with Joseph Hogue, CFA — 62% over 3,469 scored calls, at +41.6% vs +46.0%. The largest sample here, and still one of the few whose measured portfolio sits behind the index. High hit rate, trailing return.
• ARK Invest — 62% over 293 scored calls, at +61.8% vs +49.2%. One call this week: Moderna (8/26), on Phase 3 melanoma cancer-vaccine results showing a 50% mortality reduction combined with Keytruda.
• Everything Money — 53% over 2,543 scored calls, at +81.8% vs +55.7%. The value-process signature again: low hit rate, high return — and this week, a mind changed in three days.
• Value Investing with Sven Carlin, Ph.D. — 55% over 75 scored calls, at +9.2% vs +2.0% — a thin sample, which matters given how many calls he filed.
• The Compound — 61% over just 33 scored calls. The week’s loudest bear has by far the shortest measured record on this list. Weight the nine avoids accordingly.
• Parkev Tatevosian, CFA — the week’s most prolific channel at 48 calls, at 56% over 326 scored calls and +4.7% vs +2.0% on a short measured window. Frequency is not accuracy, which is why we rank on measured results. See the full accuracy-ranked list.

What to watch next week

Grounded in open calls on the board, not prediction:

• One creator is on both sides of Nvidia inside four days. The 8/24 buy and the 8/27 avoid are both dated and both live. Whichever he reiterates next is a cleaner signal than either call alone.
• The payments consensus has one independent vote. If Visa and Mastercard pick up buyers who are not citing Ackman or Trump, it becomes a real thesis. If not, it is one filing with an echo.
• Meta’s first valuation bear still hasn’t shown up. Four weeks, and the only objections so far are a copied sale and a critique of Zuckerberg’s product philosophy.
• The Anthropic read-through is now load-bearing in three directions. It is the bull case for Amazon, Micron and Nvidia in one creator’s models, and the bear case for the same names in another’s. An actual Anthropic IPO filing would settle a lot of open positions at once.
• Coherent has two independent buyers and no bear. Watch whether a third channel picks up the optical-bottleneck thesis, or whether it stays a two-creator idea.

FAQ

Which stock did finance YouTubers buy most this week? Nvidia had the widest coverage — 12 calls from nine channels between August 24 and August 30 — but it was contested, with nine buys against three avoids. The only widely-covered names with no opposition at all were Visa and Mastercard, four buys each from four channels. Read that consensus carefully: three of the four creators were reporting Bill Ackman’s or Donald Trump’s disclosed positions rather than their own analysis.

What was the most surprising pick of the week? A top-3 ranked creator ignoring AI completely on August 24 for a consumer-brands basket — Celsius, RH, Revolve, e.l.f. Beauty and Honest Company — on seasonal momentum and holiday-quarter strength. The runner-up: two unrelated channels landing on optical components six days apart, buying Coherent (and, in one case, Lumentum) on the argument that lasers, not GPUs, are the current AI data-center bottleneck.

Which stock did finance YouTubers avoid this week? Palantir was the only widely-covered name with no buy at all — two avoids and a sell, and every objection was about price rather than the business, with one creator citing a forward P/E above 100 and another 90x current-year free cash flow. CoreWeave drew two independent avoids on cash burn. Nvidia, Amazon, Microsoft, Alphabet, Meta and Oracle all collected avoids too, but nine of the week’s 33 avoids came from a single session.

Did anyone sell anything this week? Twelve sell calls, and they need sorting. Three — Netflix, Palantir and Meta — came from one creator reporting Donald Trump’s disclosed sales rather than his own view. Four more were Elliott-wave short setups from a single technical channel, including Sony and SK Hynix. The five conviction sells were Ciena on a 70x free-cash-flow multiple, Delta on $60 billion of debt, MongoDB on decelerating growth, Berkshire Hathaway on an insufficient forward return, and Limbach — where the creator says outright that his thesis broke after management cut guidance.

Methodology: we transcribe every new video from the finance channels we track and use AI to extract only qualified calls — a named stock, a clear stance, and real reasoning. See how it works.

Not financial advice. This article aggregates third-party opinions for informational purposes.