Overbought Isn't A Timing Signal — The Semi Blow-Off Is 32 Months Old

Warren Pies of 3Fourteen Research notes that S&P 500 semiconductors first registered a "blow-off" condition — a five-year return of at least 200%, a six-month return above 20%, price within 1% of an all-time high and at least 12% above the 200-day average, all on the same day — thirty-two months ago, and have since gone through two corrections greater than 20% (a 26% drop in one month and a 35% drop in two months) while continuing to make new highs. The group is currently down about 13% from its recent peak. His historical anchor: the 1990s semiconductor bull ran for 63 months after first registering a blow-off, suffering roughly one 20% drawdown per year the entire way up.

This is the disconfirming discipline the framework's positioning work needs, and it deserves the most weight precisely because it argues against where the crowding signals have been pointing. The FMS "most crowded trade on earth," the record single-stock call skew, the de-grossing — all of it registers as "the top is near," but Pies quantifies why that instinct has been wrong for 32 months: overbought and crowded is a condition that persists for years, and the correct response to it has been to buy the roughly annual 20% dip, not to fade the highs. The distinction that keeps this from dissolving the thesis is that Pies is making a price-history and analog argument, while the Modal thesis is a structural-cause argument. His base rate says blow-off tops resolve slowly; it is silent on the specific thing this thesis rests on — hyperscaler free cash flow turning negative, the monetization gap, and the vendor-financing concentration now visibly cracking (bond spreads, the Oracle downgrade, SpaceX below its IPO price). The 1990s semis bull did not run on a buyer base spending negative-FCF capital funded by an investment-grade bond market that is already repricing. So the honest synthesis: overbought alone probably does not top this cycle, and crowding should be retired as a timing trigger — but the analog cannot price the financing layer, which is where this cycle can differ from the base rate.