Actionable insights — Stock Picking in Difficult Times with George Noble

2:39 1. Decompose the stock into macro factors before you pick it

1. For any group or stock, ask three questions: "Are rates going up or down? Is the dollar going up or down? Is the economy slowing or increasing?" Most stocks' behaviour "can be described as a combination of various factor influences."
2. Decide whether the macro is "quiescent" or a "gale force hurricane." In quiet times, stock-pick freely; in storms, the factors dominate.
3. Only fight a factor with a very strong stock-specific case: owning homebuilders or mortgage lenders "and rates are going up and to the right, it's hopeless."

• The direction of the 10-year, the dollar index and growth data; sectors whose stocks all move together on a macro print (factor-driven), versus ones that diverge (stock-driven).

7:18 2. Track which way the market's inflation error has been running

1. Compare what markets expected for inflation with what arrived, over several years, not one print.
2. Note the sign of the error: post-GFC to ~2021 the market "was constantly overestimating" inflation; "since 2022, the market's been constantly underestimating inflation."
3. Explain the flip with the regime: with slack (deflationary bust risk) stimulus doesn't inflate; with "binding constraints" and fiscal dominance it does ("real inflation because you got bottlenecks").
4. Treat forward rates and bond futures as commentary, not forecasts — Noble's senior thesis found they "do a horrible job" predicting rates. Don't anchor on single CPI prints ("3.8 instead of 3.6… who cares?").

• Breakevens versus realised inflation; alternative gauges (Truflation) versus CPI; supply bottlenecks (energy, labour, power) that turn stimulus into inflation.

12:48 3. Judge a policy intervention by its size and whether the fundamentals agree

1. Measure the tool against the market it's aimed at: a few billion of buybacks against $40T of debt "is like trying to kill a whale with a BB gun" (Eisman).
2. Ask whether the fundamentals support the target price. "Intervention… only really works when the fundamentals are with you." If not, it is "open mouth operations."
3. Look for the Norman Lamont pattern: an official defending "an artificial rate," which invites the market to attack — "when there's a wrong price, markets have a way of attacking."
4. Watch credibility: a quick move through the defended level (Eisman: "rates go to five") ends the policy's effect.

• The size of buyback/intervention programs versus issuance; the 10-year's behaviour after each announcement; the 5% level.

13:45 4. When the supplier funds its customer, ask why nobody else will

1. Flag any vendor financing, equity stakes or funding packages from a supplier to its buyers: "The fact that they're having to do that tells me all I need to know."
2. Translate it into Noble's homebuilder test: a builder lending to buyers at 0–1% when mortgages cost 7%, not accruing the gap, and booking it all as income.
3. Check the receivables footnote for concentration (Eisman: "70% of accounts receivable is from five accounts").
4. Ask whether independent lenders are funding the same customers. If "they're not doing it," the market is telling you the credits are weak.
5. Stress-test: circular financing is harmless only if the end customers become "insanely profitable" — so value the chain on their path to profit, not on the vendor's reported earnings.

• New vendor-financing announcements; receivables growing faster than revenue; the top-customer share; third-party lending terms for the AI labs.

16:05 5. Separate "the technology will win" from "the stocks will work"

1. Grant the bull's usage forecast; it may be right. Peter Berezin: internet traffic compounded "43% a year for the next 25 years."
2. Then ask what return the capital earns: that growth "didn't stop Global Crossing and all the rest of them from going bankrupt."
3. Size the capital spent against the likely profit pool: Garran's "24 times" the dot-com malinvestment against a market "maybe… worth 50 or hundred billion dollars."
4. Check pricing: if a cheaper model is "good enough for 95% of us at 95% less cost," usage growth won't earn back the spend.

• AI capex versus AI revenue; price per token; open-weight model share; heavy borrowers in the build-out.

27:17 6. Price a government bond against its peers' debt loads

1. Plot debt/GDP against 10-year yields across countries (Robin Brooks' chart): Switzerland and Sweden low, Europe mid, the US higher, Japan highest in debt.
2. Read each country's fair yield off the line — "the more debt the country is, the higher the rate should be, just as you would with a subprime buyer."
3. Adjust for who owns the debt (Japan's ~240% is "more like 130" net of domestic holders).
4. A big gap between fair and actual yield is pressure that must show up somewhere — in the yield, or in the currency.

• JGB yields versus the line; USD/JPY; the Bank of Japan's reaction; the same test on the US at a 10-year of 4.8%.

30:14 7. Watch the yield gap that keeps foreign money in US assets

1. Identify the big foreign holders of a market (Japan as the "world's largest creditor"; European pensions heavy in US tech).
2. Track the spread that motivates them: US 10-year 4.8% versus JGB ~3% — "only 180 basis points; the gap used to be much bigger."
3. Add their domestic pressure: a falling yen raises rice prices and inflation, so at some point they say "bring the money home."
4. Watch intervention mechanics: defending the currency means selling dollar assets, which means Treasuries — unless a swap line substitutes.

• US-JGB 10-year spread; Japanese Treasury holdings (TIC data); intervention sizes; Fed swap-line usage.

40:20 8. Time a short on an overvalued IPO to its float schedule

1. Split the overvaluation into its causes: (a) "the story, the narrative," (b) "the incredibly small float."
2. Map the lock-up calendar: SpaceX went "from 5% of shares… to 20 or 25%," with "every few weeks 7% more" unlocking.
3. Identify forced or motivated sellers: early holders who bought at a fraction of today's price, and institutions whose position "has gotten so big" they must diversify.
4. If nothing changes fundamentally, a float heading from 5% to 100% points the price down. Hold on a year-end horizon, not daily moves: "others speculate, we invest."

• Lock-up expiry dates; free-float percentage; secondary filings; index-inclusion flows that absorb the new supply.

43:51 9. Add up the parts and set the total against the market cap

1. List the segments and value each on its own: Starlink "200 billion, 300 billion"; the AI arm "a fraction" of the $200–250B paid in stock; rockets "cool" but small.
2. Check where the stated TAM really sits — the S-1's $29T TAM is $22T AI, so it is "really an AI play… with a rocket thing attached."
3. Sum it ("300 billion, 400 billion, 500 billion") and compare with the market cap ($1.7T). Convert to a per-share range and be generous at the top ("I'll give you 60").

• Segment disclosure in filings; private-market marks for comparable assets (satellite broadband, AI labs); changes in the AI arm's revenue.

46:39 10. Value the "option" by scaling its observed adoption

1. Find the business that carries most of the valuation ("the vast majority of the valuation of Tesla depends on the robo taxis").
2. Take the disclosed usage in the live pilot (rides per million of population in Austin, about a million people).
3. Extrapolate that rate to the full market (350 million US people) and put a value on the result.
4. Compare with what the stock implies: "800 million not 800 billion."

• Company-disclosed ride counts; geofence expansion; per-city adoption rates as new markets open.

44:48 11. Read monthly sales and the comparison base before the quarter prints

1. Use public monthly registration/sales data by region (US July −26%, Europe and China down).
2. Check the year-ago quarter for one-offs: last year's Q3 "benefited immensely from… pre-buying… ahead of the removal of credits," so this quarter compares against an inflated base.
3. Layer on cash use: capex rising from ~$6–8B to $30B while the core business shrinks means "seriously cash flow negative."

• Monthly regional sales; quarterly delivery numbers; capex guidance; free cash flow.

48:11 12. In a passive-driven mania, avoid the bloated heavyweights and index the rest

1. Spot stocks held up by "passive buying and also the story and the momentum" rather than fundamentals ("narrative dominance"; "no price discovery").
2. Recall Japan in the '80s: a market at 60× earnings with banks at 100× earnings. "Once the worm turned, all you have to do is avoid that and just index everything else and you kill it."
3. Check the long-run scoreboard: a stock "unchanged over 5 years… despite all the hype" is running on liquidity and narrative.

• Index weight versus earnings share; passive flow data; the equal-weight index versus cap-weight.