Most important take away

Jeff deGraaf’s bubble rule (an index or sector that doubles in 2 years or less) has flashed for Korea’s KOSPI and the U.S. semis (SOX) — historically a 6–12 month yellow flag, not a sell signal. The actionable response is sizing, not shorting: trim position size, recalibrate for higher volatility, and plan to “de-risk on the break” rather than try to call the top. Underneath the parabola, memory chip fundamentals (Micron, SanDisk) really did quadruple, so the move is rational — but Josh and Michael still expect roughly a 75–80% chance of a 40%+ drawdown at some point in the next year.

Summary

Actionable insights and investment advice

• deGraaf’s bubble rule: when a diversified index or sector doubles within 24 months, you are in bubble territory. It is a yellow flag, not a cell ticket. Has flagged Hong Kong/Hang Seng (90s), China, NASDAQ (1998, early by ~2 years), and now KOSPI and the SOX.
• The signal is for sizing, not shorting. Shorting bubbles is “an expensive mistake.” Reduce gross as conditions deteriorate; the win is in the de-risking sequence, not calling the top. Exit on break — don’t fade early.
• Recalibrate for volatility. If you size by vol, semi/memory names now carry far more risk than they did a quarter ago — your nominal weight is “lying” to you.
• Terminal wealth is maximized by selling on the way down, not on the way up. Trying to call the top usually leaves you with half the gains. Accept that you’ll give some back.
• Time is the bear’s best friend in a bubble — odds of a major drawdown rise meaningfully 6–12 months after the doubling signal. Set a watch and start looking for a break.
• Today’s tape: KOSPI printed an outside reversal day (open above prior high, close below prior low). Micron was down ~9% intraday but closed down only ~4%; buyers stepped in. Not necessarily THE top, but a “grain of sand.”

Stocks and instruments mentioned

• KOSPI / Korea: Samsung + SK Hynix = ~43% of the index (vs. ~18–20% in 2000). Bubble signal triggered. MSCI Korea forward EPS quadrupled in a month — earnings genuinely support the rally.
• SOX / semiconductors: Doubled inside 2 years two weeks ago; bubble signal triggered. SMH had a similar bearish-then-recovery candle today.
• Micron (MU): From ~$420 (April 12) to ~$800 in a month; closed down ~4% today after being down 9%. Trades at ~9x 2026 / ~7x 2027 EPS but operating margins swing from +40% to -60% — investors are right to discount cyclicals. Now the 10th largest holding in the S&P 500 and the largest stock in the Russell 1000 Value index — a curiosity.
• SanDisk: $250 (January) to $1,500; was $1,000 three weeks ago. ~11x / 8x earnings. Sales growth +283%, EPS growth +7,028%.
• Seagate, Western Digital: Same memory complex; similar setups.
• Goldman Sachs (GS): Josh’s “capital markets mood ring.” Held the rising 200-day twice in March; consolidating in mid-$900s. A break above $1,000 = signal of another six months of bull market (more IPOs, M&A, underwriting). Failure to break = warning. 70/30 odds of $1,000 before $800. GS has compounded ~23% annualized over 5 years, beating Morgan Stanley and JPM.
• Netflix vs. Spotify (mystery chart): Trading nearly in lockstep at $432 / $87. Michael owns Spotify; views it as defensive — money rotating out of momentum could land here. Closed today’s gap.
• Housing stocks (depression conditions): Whirlpool, Pool Corp (-55% over 5 years), Owens Corning, Mohawk, Home Depot (multi-year lows), Lennar ($185 → $85). Problem is lack of existing-home turnover, not falling prices. Suggests durable goods/housing names remain a “barren wasteland” — avoid until activity returns.
• AI capex / hyperscaler debate: Wall Street Journal piece argues asset-heavy tech is historically a worse stock setup than asset-light tech. Hyperscalers’ balance-sheet assets have doubled in 5 years. Compare to Verizon/AT&T: massive 25-year capex into wireless infrastructure produced poor shareholder returns. Open question: do AI profits accrue to the platforms (hyperscalers), the models (OpenAI/Anthropic), or to a yet-unimagined software/services layer? Position accordingly with humility. A Chicago prop firm is reportedly building markets to trade compute like oil or electricity — could ultimately ease the capex risk.
• Momentum factor: S&P 500 Momentum just printed its largest 6-week return ever (30.5%) across live and back-tested history. Winners are being loved too much; misses (Shake Shack, Toast, Netflix anecdote) are being punished much more than average.
• Concentration / market structure: 115 stocks in the S&P 500 are now $100B+ (vs. ~20 in 2005). 31 are tech (38 with comm services), 19 are financials (zero post-GFC in 2009). Memory chip names entered the $100B club.

Themes

• Bubbles don’t ring a bell at the top; they emerge in rational waves of fundamentals plus momentum. The current AI/memory move is fundamentally justified but still mechanically a bubble pattern.
• Even Isaac Newton blew up in the South Sea Bubble — discipline (sizing, de-risking on the break) matters more than IQ.
• An “AI industrial bubble” (Bezos) need not mean no winners; it means the returns on all the capex will probably be uneven, and shareholders historically don’t love asset-heavy bets.
• The economy is unusually resilient given a depression-level housing sector — illustrates the dominance of AI capex and large-cap tech.

Chapter Summaries

1. Open and sponsors. Betterment Advisor Solutions, Janus Henderson securitized.
2. Jeff deGraaf surprise appearance — the bubble rule. Doubling of an index or sector within 24 months is the bubble signal; analogy to hypoxia at altitude. KOSPI and SOX both qualify.
3. The “Twelve Rules for Riding a Bubble” framework. Yellow flag not sell ticket; shorting is expensive; base rates favor trend; signal is for sizing; time is the bear’s best friend (6–12 months out); reduce gross as conditions deteriorate; de-risk on the break, don’t fade early. Isaac Newton’s South Sea story illustrates the mental anguish of selling on the way down.
4. Today’s tape action. KOSPI’s outside-reversal candle; Micron’s -9% intraday recovering to -4%; SMH similar. Big-picture: not necessarily the peak but a meaningful warning.
5. Fundamentals justify the run. MSCI Korea forward EPS quadrupled in a month; Micron and SanDisk sales/EPS up triple to quadruple digits. The buying is rational; cyclical PEs (8–11x) look cheap but reflect deep cyclicality. Micron is now the largest stock in the Russell 1000 Value — a sign of how distorted the screens are.
6. Earnings season punishment asymmetry. Beats rewarded in line with average (+1%); misses punished much more than average (~ -5% vs. -2.9%). Examples: Shake Shack -38% over four days, Toast -22%, Netflix’s -8% on strong results.
7. AI capex and the “industrial bubble” debate. WSJ piece argues asset-heavy tech historically underperforms asset-light tech. Hyperscaler balance-sheet assets have doubled in 5 years. Telecom/wireless capex analog as a cautionary tale. Possibility that compute becomes a tradable commodity. Open question on where AI profits accrue.
8. Housing stock apocalypse. Whirlpool, Pool Corp, Home Depot, Lennar, Owens Corning — all down sharply on lack of housing turnover. Remarkable that the broader economy is resilient through this.
9. The $100B club. 115 names now vs. 20 in 2005. Tech (31) and financials (19) dominate. Memory chip names entering the club is striking.
10. Make-the-Case: Goldman Sachs as the capital-markets mood ring. Watch the $1,000 level — a break = green light for another 6 months of bull market and capital formation; failure has implications. GS has outperformed MS and JPM over 5 years (~23% annualized).
11. Mystery chart: Spotify vs. Netflix. Trading near-identically. Michael owns Spotify, sees it as defensive, willing to add if it pukes into the 70s.
12. Outro. Animal Spirits Wednesday; Ben Carlson visiting NYC for a live Compound and Friends taping.