This Week in Stock Picks: What Finance YouTubers Bought (August 17 – August 23)

In the week of August 17 – August 23, 2026, the finance YouTube channels we track published 171 qualified calls across 20 different channels. The headline split was familiar — 73% buys, 12% avoids, 11% holds and eight sells — but the composition underneath it changed shape. The megacaps stopped being a single trade: Nvidia and Meta were bought almost unopposed, while Microsoft and Amazon each collected two avoids, and the objections came from creators who bought the other two in the same session. The week also produced two of the most unusual organizing ideas we have logged — a full video asking which company the US government buys next, and a ranking exercise that generated four of the week’s eight sell calls as a byproduct of filling a tier. Here is what they actually bought, where they split, and which of the sells you should discount. All of it dated and sourced.

TL;DR

• Nvidia had the week’s widest creator spread: eight buys from seven distinct channels, against exactly one bear.
• Meta was unopposed for the third week running — nine buys from seven creators, nothing filed against it. Two of those creators are top-3 ranked.
• The hyperscaler trade split. Microsoft (4 buys, 2 avoids) and Amazon (3 buys, 2 avoids) were both contested — and one creator avoided both while buying Meta.
• The surprise: a session about the US government’s stock portfolio. One creator picked Intel, MP Materials, IonQ and Nvidia as the next candidates for a state equity stake.
• The second surprise: read the sells carefully. Four of the week’s eight sells — AMD, Palantir, CrowdStrike, Robinhood — came from a single tier-ranking video where the creator says outright he needed names to fill the short tier.
• Alphabet drew four buys, one hold and a notable avoid — from ARK Invest, betting against Google returning to the AI frontier.
• The week’s cleanest thematic session bought electricity again: Vistra, Talen, GE Vernova and Constellation Energy, on the argument that only three companies can legally sell power straight to a hyperscaler.
• Live ranking, updated twice a day: the Terminal.

The week in numbers

The Latest Stock Calls feed carried 171 calls from 20 channels — down from 186 and 24 the week before, on a thinner publishing calendar. Concentration was heavier than usual: the three most prolific channels filed 76 calls between them, 44% of the entire week, and a single video on 8/23 produced 19 calls on its own. Read every count below with that in mind. A name with seven buyers from seven channels is a real signal; a name with three buys from one creator is one opinion filed three times.

Sunday 8/23 alone carried 59 calls — a third of the week — because three of the biggest channels published portfolio-wide reviews on the same day. The composition sorted into four poles: the megacap AI complex (now openly contested rather than uniformly bought), the physical layer — power, grid, memory — deep-value turnarounds, and a large block of broad-index ETF buying from one top-ranked channel.

Nvidia: eight buys, seven creators, one bear

Nvidia produced the widest creator spread on the board — eight buy calls from seven distinct channels, with a single avoid against it. On spread, which is the measure we weight most heavily, nothing else came close.

• Parkev Tatevosian, CFA — Buy, 8/18 — a forward P/E of 17.6 and a DCF fair value of $375, with the explicit comparison that Nvidia trades below slower-growth names like Costco and Walmart.
• Ticker Symbol: YOU — Buy, 8/18 — the most structural case of the week: Nvidia is turning GPUs into an investable asset class by building financing platforms with major institutions, shifting AI infrastructure funding off free cash flow and onto borrowed capital.
• Jerry Romine Stocks — Buy, 8/18 — part of a five-name “irreplaceable toll roads” basket; Nvidia controls the bottleneck of AI computing.
• Invest with Henry — Buy, 8/19 — 65% profit margin, PEG under one, forward P/E of 25, expressed by selling puts rather than buying shares outright.
• Let’s Talk Money! with Joseph Hogue, CFA — Buy, twice, in two separate sessions: 8/19 as a government-investment candidate, and 8/23 on valuation — a 25% discount to its own historical average, with H200 shipments to China as the upside catalyst.
• Adriconomics — Buy, 8/23 — top tier, and explicitly preferred over AMD in the same ranking.
• A top-3 ranked creator (8/20) put it on the buy watchlist after the AI-sector pullback.

The dissent came from Everything Money (avoid, 8/22), and it is the most specific bear case filed all week. His objection is not the multiple — it is the circularity: Nvidia helps arrange financing for hyperscalers, who use that money to buy Nvidia chips, and repayment of those loans depends on currently unprofitable AI companies succeeding. He reads it as a fragile chain rather than organic demand.

Set that against the Ticker Symbol: YOU buy from four days earlier and you have the week’s sharpest disagreement: the same debt-financed structure, described by one creator as the bull case and by another as the crack. Both are on the record with dates. Background: Nvidia on BullVox.

Meta: nine buys, seven creators, nothing against

Meta went a third consecutive week without a single call filed against it — nine buys from seven distinct creators, no sell, no avoid, no hold. What changed is the argument. Last week it was capex and monetization; this week almost every buyer led with the legal overhang, and treated it as the reason the price is available.

• Everything Money — Buy, 8/20 — the market is overreacting to AI spending; the app ecosystem plus AI integration into existing products is the catalyst.
• Daniel Pronk — Buy, 8/21 — the $1.4 trillion lawsuit claims are exaggerated, and even tens of billions in fines would be absorbable against Meta’s cash flow and balance sheet.
• Invest with Henry — Buy, 8/19 — $700 within 12 months, expressed through LEAP calls as the capital-efficient version of the trade.
• Value Investing with Sven Carlin, Ph.D. — Buy, 8/19 — the cheapest hyperscaler on his conservative intrinsic-value math even at an 8% growth rate; on Ackman’s 20% projection the stock more than doubles.
• Adriconomics — Buy, 8/23 — also names it the cheapest hyperscaler, but ranks it a half-size position rather than a full one, specifically because of the lawsuits.
• The remaining calls came from two top-3 ranked creators. One bought it three separate times across 8/19, 8/21 and 8/23, citing a 23.3% three-year revenue CAGR, and was explicit about the level: he keeps adding below $600 and expects free cash flow to turn positive again as the investment cycle peaks. The other flagged it on the AI-sector pullback. Those identities sit behind the Premium wall; the direction and the dates do not.

Seven creators, one direction, three weeks running. The dissent to watch for is not a valuation bear — it is the first creator who prices the trial as a real impairment rather than a discount. Background: Meta Stock Forecast: What Finance YouTubers Say.

The hyperscaler trade came apart

The most important development of the week is not a single stock. It is that Microsoft and Amazon each collected two avoids while Nvidia and Meta collected none — and the objections came from creators who were buying the other megacaps in the same breath.

Microsoft: four buys, two avoids, six creators. Parkev Tatevosian, CFA (buy, 8/23) has it at a forward P/E of 20.5, near historical lows for a company with its growth and margins, with an accelerating cloud segment and the best free cash flow generation of its peer group. Invest with Henry (buy, 8/19) sees $500 within six months. Jerry Romine Stocks (buy, 8/18) calls it an enterprise titan converting cloud growth into 30%+ cash flow margins. A top-3 creator added a watchlist buy on 8/20.

Against that, two avoids built on the same single point of failure. Adriconomics (8/18) is considering selling it despite good returns: 70% of Microsoft’s AI revenue comes from OpenAI, and he does not trust the counterparty — unproven business model, senior departures. Sven Carlin (8/19) gets there independently: growth concentrated in two major customers, with current growth rates and P/E leaving no cushion if AI demand slows or the multiple contracts.

Amazon: three buys, two avoids, five creators. Adriconomics (buy, 8/23) rates it the safest of the 20 companies on his risk chart, with robotics as the free option. Parkev (buy, 8/23) reiterates it as a top pick on AWS acceleration now absorbing the bulk of capex. Jerry Romine (buy, 8/18) has it in the toll-road basket. The avoids are both conditional rather than bearish: Invest with Henry (8/19) is long-term bullish but says RSI at 64 and the post-earnings rally make this a bad entry — he would sell puts instead; Sven Carlin (8/19) puts the price close enough to his conservative intrinsic value that the implied return is 8–9%, below his hurdle, and he would rather wait for another $90-type moment.

Two creators looked at four hyperscalers and ranked them in near-opposite order. That is the second week in a row this board has failed to agree on the ordering — and this week the disagreement moved from “which is cheapest” to “which one’s growth is real.”

Alphabet: four buys, one hold, and ARK betting against it

Alphabet drew six calls from six channels — four buys, one hold and the most interesting avoid of the week.

• Felix & Friends (Goat Academy) (buy, 8/19) — the market is overly worried about chatbots eating search; cloud grew 82% last quarter, and the whole thing is mispriced.
• Arte de invertir (buy, 8/23) — framed around Buffett’s position: search plus YouTube plus cloud plus the Waymo and Android optionality, funded by cash flow that pays for the bets.
• Jerry Romine Stocks (buy, 8/18) — tens of billions in cash and consistent buybacks; another toll road.
• A top-3 creator (8/21) calls it the “sleep well at night” AI holding, and was untroubled by the negative free cash flow quarter, reading it as investment-cycle timing.
• Everything Money (hold, 8/17) — great company, rapidly growing cloud, proprietary AI silicon, resilient search; the reservations are rising spend, negative free cash flow and regulatory risk.

And the avoid: ARK Invest (8/19), which is worth reading in full because of who it comes from. The analyst grants Google every input — compute, unique chip architecture, data, distribution — and still wagers against it regaining a place on the “Pareto frontier” of AI models. Not a valuation call, not a regulatory call. A capability call, from a firm whose entire thesis is technology adoption curves. That is the rarest kind of bear case on this board.

Separately, a top-3 creator disclosed on 8/20 that he is holding Alphabet from $92 with a trailing stop — over 100% up, protecting rather than adding.

The surprise: a session about the US government’s stock portfolio

The week’s most unusual video came from Let’s Talk Money! with Joseph Hogue, CFA on 8/19, and the organizing question was not “what is cheap” but “which company does the US government take an equity stake in next?” — building on the precedent of prior state investments that produced large returns.

The four candidates, each with its own logic:

• Intel — the obvious one: another round of funding to nurture domestic chip manufacturing, on top of the backing that has already worked.
• MP Materials — one of the very few US-based rare earth producers, where previous government investment already delivered high returns, making it the natural place to double down.
• IonQ — unprofitable, but the only full-stack quantum platform in an industry the government is keen to support.
• Nvidia — included precisely because it does not need the money: a publicity stake, to have the undisputed chip leader in the portfolio.

Whether or not you find the premise plausible, it is a genuinely different screen from anything else on the board — political economy rather than DCF — and it independently landed on Intel, which Felix & Friends bought the same day (8/19) on fundamentals: the strategic manufacturing role, revenue growth in the foundry business, and a $10 million insider purchase by the CEO. Two creators, two completely different methods, same stock, same day.

The second surprise: half the week’s sells came from filling a tier

On 8/23, Adriconomics published a single video ranking 20 companies into position tiers — full-size long, half-size long, neutral, half-size short — and it generated 19 qualified calls, 11% of the entire week, including four of the week’s eight sells.

Those four sells deserve an asterisk, and he supplies it himself. On Palantir: he acknowledges it “grows like crazy,” says it is not especially risky given AI’s trajectory, and places it in the short tier anyway — hoping merely that it trades sideways. On Robinhood: solid growth, good company, “quite expensive,” and the decision is driven by the need to fill the tier with high-valuation stocks. On CrowdStrike: he likes cybersecurity, but it “can’t compete with Uber or Netflix valuation-wise.” On AMD: he concedes it could balance Nvidia, then sells it because he prefers Nvidia. Even the Tesla avoid in that session is filed “reluctantly,” to populate the tier.

This is a relative-ranking exercise, not four bear cases — and it is exactly the kind of thing that inflates a weekly sell count if you read the tally without reading the reasoning. Strip that one session out and the week produced four genuine sells. We log the calls as stated; we also think you should know how they were generated.

The rest of the same video is more informative than its short tier: full-size longs on Amazon, Nvidia, SoFi, Uber — where he says it is “impossible” for him to sell at this valuation — and Novo Nordisk as the deliberately non-discretionary ballast against SoFi’s credit exposure; half-size longs on Meta, Netflix, Adobe and Duolingo; neutral on Nike, Micron and Constellation Energy; and full shorts on IonQ (250× earnings, plus the whole quantum sector as unproven) and SpaceX (a 350× multiple he says no growth story justifies).

Note the collisions. Asymmetric Investing by Travis Hoium bought Robinhood on 8/18 on its shift to financial infrastructure via the Trade PMR acquisition and RIA integration. Brian Stoffel upgraded CrowdStrike on 8/18 to “anti-fragile,” arguing that AI-driven cyber threats make its tools more essential the messier the environment gets. Both were sold five days later by the tier exercise. And Joseph Hogue bought IonQ on 8/19, four days before it was named a full short.

Buying the electricity, again — and the drones

Felix & Friends (Goat Academy) filed the week’s cleanest thematic session on 8/22, and it is the second week running that a creator has bought the power behind the AI buildout rather than the compute inside it. His framing is sharper than last week’s: only three companies can legally sell power directly to major tech buyers, because they own generation in deregulated markets.

• Vistra — a legal monopoly on direct power sales, well off its highs, with the underlying contracts not yet in the financials.
• Talen Energy — owns Susquehanna, with an $18 billion contract with Microsoft already signed. Weak trailing fundamentals, contracted future.
• Constellation Energy — nuclear in deregulated markets, signing long-term contracts with buyers he describes as desperate; current margins low, future margins the point.
• GE Vernova — the shovel rather than the mine: six years of gas-turbine production backlog and a $42 billion grid backlog including transformers, against a global shortage of exactly those components.

The same creator ran two other non-obvious sessions. On 8/17 he bought the drone supply chain: Unusual Machines as the pure-play NDAA-compliant US component supplier (687% year-on-year revenue growth) and DRNZ as the diversified, lower-fee ETF version for anyone who wants the theme without the single-stock volatility. And on 8/19 he bought ZIM, a shipping company, on pure balance-sheet arithmetic: a $3.4 billion market cap against $2.7 billion of cash, with $1.5 billion of cash generated in the last year — cyclical, and he says so.

Reddit, SoFi and the memory trade

Reddit collected three buys from two creators, and the case is now entirely about licensing. A top-3 creator bought it twice (8/21, 8/23) after the pullback erased the S&P-inclusion gains, and was unusually candid about the bear case: if Reddit fails to extract significant payments from AI companies for its data, that is the thesis breaking — so he is building the position more slowly than planned. Joseph Hogue (8/23) put numbers on the same setup: a potential billion-dollar settlement with Anthropic plus the Google training negotiation, targeting $170–200. Our earlier write-up: Is Reddit Stock a Buy?

SoFi drew three buys and a hold from three creators, and the most useful thing on it is one creator contradicting himself five days apart. Adriconomics marked it a hold on 8/18 — up 22% in two weeks, fair value $29, wait for a clean break of the 200-day moving average — then made it a full-size long on 8/23, still cheap, still growing, with recession credit risk balanced elsewhere in the book. A top-3 creator (8/21) argued the multiple is too low for the growth and that SoFi beats its own rate assumptions comfortably; Joseph Hogue (8/23) sees at least $20. Background: SoFi on BullVox.

Memory quietly became a theme. Parkev Tatevosian, CFA bought Micron twice (8/17 and 8/23) on management’s guidance for tight supply and strong demand beyond 2027, at a mid-single-digit forward P/E — while Adriconomics parked the same stock in neutral, calling the price okay but the expectations “so so high.” Jerry Romine Stocks (8/23) went around them both and bought SK Hynix instead, on a head-to-head comparison: best topline growth, best margins, HBM pole position, cheapest valuation per unit of profit. And a top-3 creator bought the Roundhill Memory ETF twice, on the grounds that the whole sector was on sale.

The value corner: Ackman’s book and Buffett’s book

Two creators spent the week re-underwriting somebody else’s portfolio, and both produced full baskets.

Value Investing with Sven Carlin, Ph.D. filed thirteen calls, most of them on 8/19, working through Pershing Square’s positions: buys on Uber (an intrinsic value of $118 on 20% free cash flow growth, even after stock-based compensation, if it becomes the logistics and autonomous-vehicle platform), Visa (the toll-taker of finance, with a P/E he considers too low for the consistency), S&P Global (the multiple has contracted from 43 to 25 since he last looked), Intercontinental Exchange (mid-20s annualized returns if current momentum holds), Meta and Netflix; a hold on Alcon; and avoids on Microsoft, Amazon and Restaurant Brands. Separately he filed the week’s most considered sell — see below — and a Nike buy on 8/21, arguing it is near a bottom at a high-teens P/E with takeover interest as the margin of safety.

Arte de invertir (8/23) did the same for Berkshire’s disclosures: buys on Alphabet, Berkshire itself — Buffett historically buys his own stock below 1.6× book, and it is around 1.4× — and Delta Air Lines, on sector consolidation improving margins and a structural shift in spending toward travel and experiences. Plus the exit: Constellation Brands, sold entirely, on a structural decline in alcohol demand as younger consumers move to healthier habits.

Nike ended the week with two buys and a neutral from three creators — the only non-megacap to draw that much cross-channel attention. Stealth Wealth Investing (8/23) makes it conditional: it is a buy if management can fix product quality and win back the casual and daily-wear share it lost. Adriconomics stays neutral for the mirror-image reason — “too beaten down and it’s cheap,” but no growth.

The ETF week

One top-3 ranked creator filed 24 calls across three sessions (8/18, 8/20, 8/23), and 17 of them were ETFs — the broadest index-buying block we have logged from a single channel. The list runs from the boringly foundational to the outright thematic: an S&P 500 tracker, a total-market fund, a dividend-equity fund, the Nasdaq-100, a semiconductor ETF he bought three separate times, a quantum computing ETF, a memory ETF and a space-innovation ETF.

The through-line is not selection but timing: every one of the three sessions was framed as buying the AI-sector pullback, with the semiconductor and memory funds justified explicitly as “on sale.” His single-stock activity that week was minimal by comparison — watchlist buys on Nvidia, Meta, Microsoft, AMD, Okta and Wolfspeed, and one disclosed hold.

It is worth stating plainly what this does to the week’s arithmetic: a tenth of all calls logged were broad-index ETF buys from one channel. That is why the consensus sections above are built on creator spread rather than raw counts.

The sells and avoids

Eight sells, four of which came from the tier exercise described above. The other four:

• Berkshire Hathaway — Sven Carlin (8/20): a great business and a safe one, but he models a likely 4–5% annual return from here over ten years, which is not enough for an enterprising investor. Note the direct collision: Timo Baudzus bought it on 8/18, pointing at insider purchases by the CFO and Head of Legal as a vote of confidence in Greg Abel, with $500 or $460 as his entry levels. Same company, same week, opposite sides.
• Marvell Technology — Joseph Hogue (8/23): profit-taking, stated as such. The stock has tripled in eight months to 23× price-to-sales, more than double its valuation a year ago; he wants a dip to re-enter.
• Snowflake — Parkev Tatevosian, CFA (8/23) filed three directions on one stock in one session: a downgrade to hold, an avoid for new money, and a trim, with covered calls at $350–375 into earnings. The number behind all three: $333 market price against a $199 intrinsic value estimate after the stock more than doubled since his buy rating.
• Constellation Brands — the Buffett exit described above.

The bearish signal, as usual, ran through avoids:

• Tesla — two avoids and a hold, no buys. Stealth Wealth Investing (8/18) makes the general argument with Tesla as the case study: investors who bought the 2021 hype are still underwater, and even great companies are bad investments at the wrong price. By 8/23 he had moved to a reluctant hold — fundamentals still struggling, margins eroding, valuation complicated by the private ventures.
• Apple — split. Parkev (avoid, 8/17): fair value $212 against a $306 price, foldable iPhone notwithstanding. Jerry Romine (buy, 8/18): cash pile, quarterly buybacks, toll road.
• The junk drawer. Stealth Wealth Investing filed four avoids in one session on 8/23 — NIO (geopolitical noise, management trust, inconsistent profitability), GameStop (digital distribution has broken the model; he dismisses the squeeze narrative outright), Lucid (overvalued since the CCIV days, can’t scale) and McDonald’s (not a growth stock, limited dividend growth, better opportunities elsewhere).
• Salesforce — Joseph Hogue (8/23): tradeable around earnings if sentiment gets low enough, but a difficult long-term hold while AI fears keep pressing on traditional software.
• AppLovin — a top-3 creator (8/18) is skeptical of the trillion-dollar ambition: decelerating growth, concentration in mobile gaming. Context: Is AppLovin Stock a Buy?
• Booking Holdings — Adriconomics (8/18): excellent company, excellent management, but his $230 fair value no longer implies undervaluation and he is bracing for weak earnings.

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 1,588 scored calls, with a measured copy-portfolio return of +62.6% vs the S&P 500’s +32.6%. The widest margin over the index of any creator who filed a full basket this week, on by far the largest sample among them.
• Invest with Henry — 66% over 208 scored calls, at +24.5% vs +14.0%.
• Timo Baudzus — 64% over 53 scored calls, at +6.9% vs +1.7% — a short measured window, so weight the sample accordingly.
• Jerry Romine Stocks — 63% over 225 scored calls, at +22.8% vs +13.2%.
• Let’s Talk Money! with Joseph Hogue, CFA — 62% over 3,452 scored calls, at +40.3% vs +45.7% — the largest sample on this list, and one of the few active channels whose measured portfolio sits behind the index over its scored window. High hit rate, trailing return.
• Everything Money — 54% over 2,532 scored calls, at +82.6% vs +55.3%. The week’s most specific bear again shows the value-process signature: low hit rate, high return.
• Adriconomics — 58% over 149 scored calls, at +25.9% vs +15.7% — a decent margin on a thin sample, which is worth remembering given how many of the week’s calls came from his one session.
• Parkev Tatevosian, CFA — the week’s most prolific channel at 27 calls, at 58% over 298 scored calls and +4.0% vs +1.8% 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:

• Nvidia’s bull and bear cases now share a mechanism. One creator bought it because it is financing its own customers; another avoided it for exactly that reason four days later. Any credit event in the AI financing chain settles this one, and both positions are dated.
• Meta has gone three weeks unopposed. Nine buys from seven creators this week, zero against. The first creator who prices the trial as impairment rather than discount is the datapoint worth waiting for.
• Microsoft’s OpenAI concentration is now an explicit thesis risk. Two creators independently named customer concentration as the reason to avoid it. That is a testable claim against the next Azure disclosure.
• Reddit has a stated trigger and a stated bear case. A top-3 creator named the licensing deals as the thing that has to land; another named $170–200 as the target if the Anthropic settlement and Google negotiation resolve.
• The tier exercise created four short positions its author half-disowned. Palantir, CrowdStrike, Robinhood and AMD were all bought or upgraded by other channels within the same week. Watch whether he holds those shorts once the ranking is no longer the reason for them.

FAQ

Which stock did finance YouTubers buy most this week? Nvidia had the broadest agreement: eight buy calls from seven different channels between August 17 and August 23, against a single avoid. Meta was close behind and completely unopposed — nine buys from seven creators, with nothing filed against it for the third consecutive week. Because three channels accounted for 44% of the week’s calls, creator spread is the more reliable measure than raw counts.

What was the most surprising pick of the week? A session on August 19 that screened for companies likely to receive a US government equity stake rather than for value — landing on Intel, MP Materials, IonQ and Nvidia. The runner-up: a repeat of the “buy the electricity” trade on August 22, this time argued on the specific point that only three companies own generation in deregulated markets and can therefore sell power directly to a hyperscaler — Vistra, Talen and Constellation Energy, with GE Vernova as the equipment play.

Which stock did finance YouTubers avoid this week? Tesla was the only widely covered name with no buy at all — two avoids and a hold. The most contested megacaps were Microsoft and Amazon, each with four and three buys against two avoids; in both cases the bear case was customer concentration or entry price rather than business quality. The most specific bear case of the week was against Nvidia, on the circularity of its customer financing.

Did anyone sell anything this week? Eight sell calls, but read them carefully: four came from a single video ranking twenty companies into position tiers, where the creator states outright that he needed high-valuation names to fill the short tier — AMD, Palantir, CrowdStrike and Robinhood. The four genuine sells were Berkshire Hathaway on an insufficient forward return, Marvell as profit-taking after a triple in eight months, Snowflake on a $333 price against a $199 intrinsic value estimate, and Constellation Brands as a Buffett exit on structurally declining alcohol demand.

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.