← David Hay hub  ·  Research hub  ·  Research library

David Hay — Portfolio Update: Four-stock Edition

Four position reviews written as deliberate "quasi-tweets," with the hit rate stated up front: "three are winners and one is a loser. Coincidentally, that's about the hit vs miss rate our picks have had overall since 2022." EOG has done what the house predicted — "we have been opining that EOG would experience an upside range expansion or breakout," and "this month, it broke above overhead resistance extending back to the powerful energy sector rally during the early stages of Russia's attack on Ukraine" — though "this is not yet decisive" until it clears "the upper $140s where it had previously hit the wall." Since the June 17 "vehement endorsement" it has gone "from $133 at that point to $151 today, over six times greater than the return on the S&P," and "the valuation metrics for EOG remain extremely undemanding… the Price/Sales ratio is particularly modest." FANG, recommended "April 7th, 2025, at $121" and added to the Buy List then, "pretty much caught the three-year low… and it's now up almost 75%" versus "the S&P's roughly 50% rally since the Tariff Tantrum," and still trades "at the low end on a Price/Sales basis" — the metric the house prefers "for a cyclical company like FANG," with 2022's profit burst given as the worked example of why P/E misleads there. The instruction on both breaks the usual rule: "we often suggest trimming positions that have been extremely rewarding. In this case, though, we are inclined to sit tight and, hopefully, let them run," with Super-Major acquisition optionality noted as "always longshots." NEM gets the opposite verdict — "for those who simply want to cut to the chase: it's trim time, in our view" — after a full accounting of the position's history (the 37% dividend cut, the ~25% drawdown, the correct out-of-consensus call that the $5 2025 EPS estimate was low against almost $7 delivered) and the observation that it is "the only gold miner in the S&P 500." What triggers the trim is chart shape and valuation, not the thesis: "what concerns us the most is the straight up nature of the recent recovery run… a non-trivial retracement strikes us as highly probable," plus a Price/Sales ratio that "looks a bit stretched." BABA is the loser, "recommended on May 14, 2026, at $162.11" and down 20%, and the buy is maintained on evidence rather than hope: cloud revenue "grew 45% year-over-year, its fastest pace in 22 quarters," the EPS miss is "the accounting cost of building what this powerful competitive advantage requires," and Qwen has "surpassed 3 billion downloads, accounting for more than 50% of all open-source model downloads worldwide and outpacing Meta's Llama by more than 13 to 1." Verdict: "we think the discount has overshotmaintaining a buy rating," but only "for investors who have a long-term horizon and can stomach a hefty dose of volatility. For those who don't fit this description, we'd suggest trimming or exiting."
2026-AUG-24 · Haymaker (Substack newsletter, paid) · The Haymaker Team / David Hay · Portfolio Update · ↗ Read · article text · actionable insights
One-line take: the clearest single statement in this archive of Haymaker's valuation methodology for cyclicals and of its rule for when a winner is trimmed and when it is left alone — and it is worth reading for those two decision rules more than for the four calls. The methodology first: Price/Sales is the house yardstick for a cyclical, and the reason is spelled out with a worked example rather than asserted. In 2022 "oil prices vaulted to over $100 during the first half of that year, creating an unusual profits burst. Earnings per share more than doubled from 2021 to 2022, deflating the P/E to an unsustainably low point. The Price/Sales ratio was also depressed but to a significantly lesser extent." The point generalises: at a cyclical peak the denominator of P/E is inflated by a margin that will not persist, so the multiple flashes cheap exactly when the earnings are least repeatable — sales move far less, so P/S is "more reliable with companies where profits can fluctuate considerably." Second, the trim rule is stated and then deliberately not applied: "as readers should be well aware, we often suggest trimming positions that have been extremely rewarding. In this case, though, we are inclined to sit tight and, hopefully, let them run." Three conditions are named as the reason for the exception — "the chart patterns, the valuation, and the fundamental stories all remain highly encouraging" — which is the actual test: a winner is trimmed when one of those three has deteriorated, not merely because it has gone up. NEM is the control case in the same post, and it fails two of the three. The thesis is intact and Hay says so (the EPS call was "dead-on," the S&P-membership catalyst "has undoubtedly happened"), but the chart is "straight up" and the multiple is "a bit stretched," so "cashing in some of your gains on this one is a prudent move." Note the corroborating cross-check he uses rather than eyeballing the chart alone: NEM "is back at its early 2025 peak whereas the senior gold miner ETF, GDX, remains about 10% below its high" — a single name that has outrun its own sector index is extended relative to the thing driving it, which is a measurable statement rather than an aesthetic one. BABA is the third pattern: a losing position defended by asking whether the evidence has improved since the recommendation, not whether the price has. It has — 45% cloud growth, Qwen's open-source dominance, a PEG of 0.59 — so "our original thesis… is more supported today than in May" and the miss is reclassified as capex ("the accounting cost of building what this powerful competitive advantage requires"). The honesty is in the suitability caveat, which is the real risk disclosure: this is only for holders who "can stomach a hefty dose of volatility. For those who don't fit this description, we'd suggest trimming or exiting the position." A post-publication addendum reports the $10.2 billion share issuance at an 8.4% discount, "with 100% of proceeds earmarked for AI infrastructure," which "closed within hours with $28 billion in total demand, nearly three times oversubscribed" — dilution answered by demand. Portfolio housekeeping: the Buys table carries a single line, "PALL Cost/Date corrected" — a bookkeeping fix to an existing Buy-List lot, not a new position; the Trims/Holds and Sells tables are empty this issue.

1. Stocks & names mentioned

TickerNameResearchViewWhat he saidAt
EOGEOG ResourcesQT · SA · STK · FAPositiveThe predicted breakout arrived; let it run. "As alert readers realize, we have been opining that EOG would experience an upside range expansion or breakout," and it has — first "relative to shorter-term resistance late last year," then "this month, it broke above overhead resistance extending back to the powerful energy sector rally during the early stages of Russia's attack on Ukraine." Confirmation is explicitly still pending: "it is fair to note, however, that this is not yet decisive; our speculation is that it will soon be confirmed by a continuing surge above the upper $140s where it had previously hit the wall or, more accurately, the ceiling." Performance since the call: "back on June 17th, we gave EOG a vehement endorsement, one that has been rewarded. It has popped from $133 at that point to $151 today, over six times greater than the return on the S&P over that timeframe." Valuation is unchanged by the run: "irrespective of this surge, the valuation metrics for EOG remain extremely undemanding… despite having moved above trough levels hit in June. The Price/Sales ratio is particularly modest, especially considering the exceedingly favorable backdrop for a U.S.-based oil and gas producer." No trim despite the gain — "we are inclined to sit tight and, hopefully, let them run" — with takeout optionality named but discounted: "both have the potential to be acquired by one of the Super Majors, like Exxon; however… those are always longshots."read ↗
FANGDiamondback EnergyQT · SA · STK · FAPositiveUp ~75% from a three-year low, still cheap on the right metric, no trim. "A very similar security that has also rewarded Haymaker readers — at least for those who actually act on our recommendations — is Diamondback Energy (FANG). We recommended FANG back on April 7th, 2025, at $121, and added it to our Buy List at that time." The entry is credited to the panic, not to prescience alone: "we pretty much caught the three-year low on this one and it's now up almost 75%… in less than a year and a half. Of course, that was during the Liberation/Obliteration Day period of extreme broad market weakness," and it beats "the S&P's roughly 50% rally since the Tariff Tantrum." Valuation: "FANG also continues to trade very inexpensively. Relative to its own history, it's definitely at the low end on a Price/Sales basis. As we've so often written, this is our preferred valuation metric for a cyclical company like FANG" — justified with the 2022 example in which a doubling of EPS on $100 oil "deflat[ed] the P/E to an unsustainably low point" while P/S fell "to a significantly lesser extent." Same instruction as EOG: "the chart patterns, the valuation, and the fundamental stories all remain highly encouraging," so sit tight and let it run; Super-Major takeout is possible but "always [a] longshot."read ↗
BABAAlibaba GroupQT · SA · STK · FAPositiveBuy maintained on a 20% loss — the evidence improved even though the price didn't. "We recommended Alibaba on May 14, 2026, at $162.11." The print: "revenue of RMB 268.95 billion grew 9% year-over-year and beat consensus; adjusted EPS of RMB 8.52 missed the RMB 10.72 consensus by a wide margin. The miss was largely due to heavy AI infrastructure capital expenditure compressing near-term profitability. Cloud revenue grew 45% year-over-year, its fastest pace in 22 quarters, and AI-related product revenue grew by triple digits." The thesis-advancing development is the model ecosystem: "Qwen's emergence as the dominant open-source AI model ecosystem globally — surpassed 3 billion downloads, accounting for more than 50% of all open-source model downloads worldwide and outpacing Meta's Llama by more than 13 to 1. Over 300,000 derivative models have been built on Qwen," with Hugging Face's 2026 report calling it "the default workflow for developers deciding what models to fine-tune and deploy," and Qwen3.8-27B taking "3 million downloads in its first weekend while running on consumer hardware." The link to revenue is stated explicitly: "the 45% cloud growth is the commercial evidence of this flywheel: developers building on Qwen deploy on Alibaba Cloud." Pain acknowledged — "the 20% loss is real and it's been a painful few months to be a holder" — but "our original thesis (that Alibaba's cloud and AI infrastructure was being systematically undervalued) is more supported today than in May. The EPS miss is the accounting cost of building what this powerful competitive advantage requires." The numbers behind the discount: "$148 billion in trailing revenue, $15 billion in net income, 45% cloud growth, the world's most downloaded open-source AI model family, and a PEG of 0.59 (46% below its own historical average P/E and 1.78x EV/EBITDA)… we think the discount has overshot." Verdict and suitability caveat: "we are maintaining a buy rating on BABA… this stock is for investors who have a long-term horizon and can stomach a hefty dose of volatility. For those who don't fit this description, we'd suggest trimming or exiting the position." Post-publication addendum: "Alibaba issued 710 million new shares at HK$112.70, an 8.4% discount to Friday's close, raising $10.2 billion with 100% of proceeds earmarked for AI infrastructure. The book closed within hours with $28 billion in total demand, nearly three times oversubscribed, including $6 billion from sovereign wealth funds and long-only investors," with Hong Kong shares down as much as 10% at the open but recovering "to only slightly lower (~1.5%)."read ↗
NEMNewmont CorporationQT · SA · STK · FANeutral"It's trim time, in our view" — take gains, don't exit. The stance is given without preamble: "with this entity, the largest North American gold miner, we're coming to a different conclusion. For those who simply want to cut to the chase: it's trim time, in our view." The position's full history is disclosed rather than airbrushed: "candidly, NEM was an embarrassment for us early on, including announcing a 37% dividend cut in early 2024. That was soon after we positively wrote it up, partially because the pre-cut yield was higher than a 10-year Treasury and it also provided a hefty hedge against monetary debasement. This, along with soft earnings, caused its own debasement… it lost around 25% of its value in short order. Fortunately, we gave it a mild positive plug when it was on the ash heap." The thesis worked: "by August of 2025, it was breaking out… we focused on the series of positive earnings surprises it was (finally!) reporting. We even stuck our necks out and said we thought the EPS estimate of $5 for 2025 was low. We were dead-on with that one, as it ended up reporting almost $7 of EPS for the full year." The structural point stands too: "it was (and actually still is) the only gold miner in the S&P 500… that might give it an added boost once it became clear it was on an earnings roll, encouraging institutional investors to embrace the name. That has undoubtedly happened and been a contributing factor in its moonshot." What changes is price, not thesis: "perhaps reflecting its special status, NEM is back at its early 2025 peak whereas the senior gold miner ETF, GDX, remains about 10% below its high. But what concerns us the most is the straight up nature of the recent recovery run… a non-trivial retracement strikes us as highly probable. We'd also note its Price/Sales ratio looks a bit stretched." Conclusion: "we believe cashing in some of your gains on this one is a prudent move." A trim, not a sale — the sector view is unchanged ("almost all gold miners have all been on fire since 2024, and are again after a nasty correction earlier this year").read ↗
GDXVanEck Gold Miners ETFQT · SA · STKNeutralPassing mention — the benchmark that makes NEM's extension measurable, not a call. Cited once, as the comparator in the trim argument: "perhaps reflecting its special status, NEM is back at its early 2025 peak whereas the senior gold miner ETF, GDX, remains about 10% below its high." The role is diagnostic — a single name trading back at its own peak while the sector index it belongs to is still 10% below its own is extended relative to the sector driving it, which is what converts "the chart looks parabolic" into a measurable statement. The surrounding sector remark is constructive rather than cautious — "almost all gold miners have all been on fire since 2024, and are again after a nasty correction earlier this year" — but no view, target or action is given on GDX itself in this post; it is rowed Neutral on that basis. (Haymaker's standing position in the ETF is unchanged by this issue.)read ↗

"View" is Haymaker's stance in this Portfolio Update. Four names are formally reviewed — EOG, FANG and BABA rowed Positive (BABA a maintained buy on a 20% loss, with an explicit suitability caveat), NEM Neutral as an explicit trim — "it's trim time" — rather than a sale, since the thesis is untouched and only the chart and the multiple have stretched. GDX is rowed Neutral as a passing mention: it appears only as the sector benchmark against which NEM's extension is measured, with no view taken. Not rowed — deliberately: Exxon is named only as the hypothetical Super-Major acquirer in a possibility Hay himself calls "always [a] longshot"; Meta / Llama and Hugging Face appear only as the yardstick and the referee for Qwen's download share; Bloomberg is the chart source. Portfolio tables: Buys carries one housekeeping line — "PALL Cost/Date corrected" (a bookkeeping fix to an existing Buy-List lot, not a new recommendation); Trims/Holds and Sells are empty. Research: QT Qualtrim · SA Seeking Alpha · STK Stock Analysis.

2. Talking points

The frame: four quick notes, and an unprompted hit-rate disclosure

EOG — the breakout that was predicted, and the level that confirms it

EOG — the scoreboard on the June call

FANG — catching a three-year low during a panic

Why Price/Sales, not P/E, for a cyclical — the worked example

The trim rule — and why it is suspended here

NEM — the position's history told against itself

NEM — the call that worked

NEM — what triggers the trim

BABA — the quarter, decomposed

BABA — Qwen as the thesis-advancing evidence

BABA — defending a loser without pretending it isn't one

BABA — the suitability caveat, which is the real risk disclosure

The addendum — dilution answered by demand

Housekeeping

3. In plain English

A jargon-free note on why each name matters here. (Companion to the table above; renders on each name's consolidated page.)

EOG — EOG Resources Positive

EOG Resources is one of the larger American oil and gas producers — it drills wells, mostly on US shale acreage, and sells the barrels. Nothing exotic: its profits rise and fall with the price of oil and with how cheaply it can get a barrel out of the ground.

Hay's argument here is half chart and half price tag. The chart part: for years EOG kept running into the same price level and stopping, going back to the surge in energy stocks after Russia invaded Ukraine in 2022. Traders call that level "resistance" — a ceiling where enough sellers appear to halt every advance. This month the stock finally traded above it. That matters because the sellers who were capping it have been used up; the same level, once broken, often becomes the floor. He is careful not to declare victory: the break is "not yet decisive," and he wants to see the stock hold above the upper $140s — the exact spot where it had previously hit the wall — before calling it confirmed.

The price-tag part is why he isn't taking profits. He first pushed the stock hard on June 17 at $133 and it is $151 now, a move roughly six times the S&P's over the same weeks. Normally a fast gain invites a trim. But the shares are still cheap compared with the company's sales — his preferred yardstick for a business like this — and the industry backdrop for an American oil and gas producer is, in his words, exceedingly favourable. When the chart, the valuation and the underlying business all still look good, his rule is to leave a winner alone: "sit tight and, hopefully, let them run."

One lottery ticket comes attached and he tells you not to pay for it: EOG is big and cheap enough that one of the oil super-majors, an Exxon, could try to buy it. He mentions the possibility and immediately calls takeovers "always longshots" — a bonus if it happens, not part of the case.

FANG — Diamondback Energy Positive

Diamondback Energy is a shale oil producer focused on the Permian Basin in west Texas — the most productive oil field in the United States. Like EOG, it is a straightforward business: pump oil, sell oil.

This is a scoreboard entry as much as a recommendation. Haymaker recommended it in April 2025 at $121, right in the middle of the market panic around the tariff announcements — what Hay calls the "Liberation/Obliteration Day" period. Buying into that fear turned out to catch almost exactly a three-year low, and the stock is up close to 75% since, against roughly 50% for the S&P over the same stretch.

The interesting part is the valuation method, which is the most reusable idea in the whole post. Most people judge a stock by its price-to-earnings ratio — the share price divided by profit per share. For a company whose profits swing wildly with a commodity price, that ratio actively misleads you, and Hay explains exactly how with a real example. In the first half of 2022 oil went above $100, Diamondback's earnings per share more than doubled from the previous year, and the P/E ratio therefore collapsed to a very low number. The stock looked its cheapest at precisely the moment its profits were least repeatable. Sales, by contrast, move much less violently than profits, so the price-to-sales ratio fell too but nowhere near as far — it gave a truer reading. That is why, for cyclical businesses, he compares price to sales instead of to earnings, and on that basis Diamondback is still near the low end of its own historical range.

Same conclusion as EOG, for the same three reasons — the chart, the valuation and the business story all still look encouraging — so no trimming despite the gain. The possibility of a super-major buying it is mentioned and dismissed as a longshot.

BABA — Alibaba Group Positive

Alibaba is China's largest e-commerce company, and — the part this note is really about — the operator of China's largest cloud computing business, renting out servers and AI computing power the way Amazon Web Services does in the West. Haymaker has described it as China's Amazon and Google rolled into one.

The position is down about 20% since the May recommendation at $162.11, and Hay does not soften that: "the 20% loss is real and it's been a painful few months to be a holder." What he does instead is ask a better question than "is it down?" — namely, has the evidence behind the original reason for owning it got better or worse since?

The last quarter looked bad on the surface. Revenue grew 9% and beat expectations, but profit per share came in far below what analysts wanted. The reason for the shortfall is the crux: Alibaba is spending enormous sums building AI data centres, and that spending runs straight through the profit line today while the benefit arrives later. Hay's phrase is that the miss is "the accounting cost" of building the advantage — a company that is investing heavily looks less profitable than one that isn't, even when the investment is the whole point.

Meanwhile the part of the business the thesis rests on accelerated: cloud revenue grew 45% from a year earlier, its fastest in 22 quarters, and AI-specific product revenue tripled or better.

The new development since May is Qwen, Alibaba's family of open-source AI models. "Open-source" here means the model itself is published for anyone to download, run and modify for free, rather than being locked behind a company's paid service. Qwen has passed 3 billion downloads and now accounts for more than half of all open-source AI model downloads worldwide — outpacing Meta's Llama, the previous default, by more than 13 to 1 — with over 300,000 modified versions built on top of it and Hugging Face, the main hub where developers share models, calling it the default starting point. A version released on August 14 took 3 million downloads in a weekend and runs on ordinary consumer hardware rather than a data centre.

Giving software away only helps if it sells something. It does: developers who build on Qwen tend to rent Alibaba's servers to run their applications, which is why Hay treats the 45% cloud growth as the commercial proof that the free model is working as a funnel rather than as charity.

On price: a company with $148 billion of trailing revenue and $15 billion of net income trades 46% below its own historical average earnings multiple, with a PEG of 0.59. PEG divides the P/E by the growth rate — below 1 conventionally means you are paying less per unit of growth than the growth alone would justify. His conclusion is not that the discount is unjustified but that it has "overshot."

The rating stays a buy, with an unusually blunt condition attached: this is only for people with a long horizon who "can stomach a hefty dose of volatility," and anyone who doesn't fit that description should trim or exit. After the piece was written, Alibaba sold 710 million new shares at an 8.4% discount to raise $10.2 billion, all of it earmarked for AI infrastructure. Issuing new shares dilutes existing holders, which is why the Hong Kong listing dropped as much as 10% at the open — but the order book closed within hours with $28 billion of demand, nearly three times what was on offer, including $6 billion from sovereign wealth funds, and the shares recovered to roughly 1.5% down. Heavy demand from long-term institutional buyers at a discount is the sort of evidence that argues the capital is wanted, not forced.

NEM — Newmont Corporation Neutral

Newmont is the largest gold miner in North America and, unusually, the only gold miner in the S&P 500 index. That last detail is not trivia — it means index funds and big institutions that only buy S&P names can own gold-mining exposure through Newmont and essentially nowhere else.

Hay tells the whole history of this position, including the embarrassing part. Haymaker wrote it up positively, and shortly afterwards, in early 2024, Newmont cut its dividend by 37% — which broke one of the two reasons for owning it, since the pre-cut yield had been higher than a 10-year Treasury bond. Combined with weak earnings, the stock lost about a quarter of its value quickly. He credits himself only with a "mild positive plug when it was on the ash heap."

Then it worked. By August 2025 the stock was breaking out on a run of better-than-expected earnings, and Haymaker took the unusual step of saying publicly that Wall Street's $5 earnings-per-share estimate for 2025 was too low. Newmont delivered almost $7. The S&P-membership argument paid off too — once the earnings improved, institutions piled in, which he thinks contributed to what he calls the moonshot.

So why trim now? Nothing in the business has broken. Two things about the price have. First, a clean relative check: Newmont is back at its early-2025 peak while GDX, the fund that owns the big gold miners as a group, is still about 10% below its own high. One stock outrunning the sector it belongs to that far is a measurable sign of stretch, not just a feeling about the chart. Second, the recovery has been, in his word, "straight up" — a vertical move with no pauses, the shape that typically gives some ground back. He adds that the price-to-sales ratio, the same yardstick that says EOG and Diamondback are cheap, now "looks a bit stretched" for Newmont.

The instruction is therefore to bank some gains, not to leave: "cashing in some of your gains on this one is a prudent move." He remains constructive on gold miners generally — almost all of them have been on fire since 2024 and are running again after this year's correction. This is a call about how much of one stock you should be holding after a vertical run, not a change of mind about gold.


Summary derived from the paid Haymaker Portfolio Update (text in transcript.txt) for personal study. Charts referenced in the original (Bloomberg) are omitted. Not investment advice. © Haymaker / David Hay for source material.