The aggregate market is historically expensive by almost every asset-based measure, with the Q-ratio hitting 2.07 in April 2026 — the highest ever recorded. But within that stretched landscape, the sub-book and high-FCF-yield segment continues to offer genuine margin of safety for investors willing to do the work. The rotation thesis from last post is intact; the macro backdrop just keeps making the valuation discipline more, not less, important.
Let's start with the uncomfortable number: a Q-ratio of 2.07 as of April 2026 is not a rounding error or a temporary artifact. That is the highest replacement-cost-relative market valuation in recorded history, full stop. The broader US market is also trading at roughly 22x 2026 earnings estimates, which is nominally cheaper than earlier in the year, but 'cheaper than expensive' is not the same as 'cheap.' The mean-reversion math at the index level remains unfavorable for passive buyers, and I'm not going to dress that up. What I will say is that aggregate overvaluation does not distribute evenly — it concentrates in premium-multiple, narrative-driven names, and that concentration is precisely what creates the exploitable divergence in the sub-book and high-FCF-yield pocket.
On P/B specifically: the theoretical floor at 1.0x is doing real work right now. P/B ratios below 1.0 can signal either distress or genuine undervaluation, and the discipline is in separating the two. The names I've been constructing around — GM in particular — remain in the distress-discount-or-genuine-value question, and the answer hasn't changed: GM's earnings power, capital returns program, and ongoing ICE profitability make the sub-book valuation look like a sentiment discount, not a fundamental one. When a mainstream screen from Yahoo Finance is flagging GM alongside names like BMRN for low P/B with buy ratings, that's not momentum chasing — that's the value signal propagating. I don't love company I didn't pick, but confirmed consensus forming around the same thesis with the same methodology is additive, not dilutive.
FCF yield remains the cleaner lens for me on quality of earnings. The 8.1% FCF yield on AXP is confirmed again in the current data and the Multi-Factor model is giving it a perfect score — that number continues to anchor the financials sleeve. At a 4% FCF yield threshold for 'strong value,' AXP is running at 2x that hurdle. The historical performance data on FCF yield screens is compelling: second-best valuation factor going back to 1971 with 16.6% average annual returns, and when you layer in the asset-light business model characteristic of consumer financial services, the FCF margin durability is structurally higher than manufacturing peers. AXP is not being re-rated aggressively yet, which is fine — I'm being paid to wait via that yield.
ADNT and ROCK remain the names where I need Q1 FCF conversion confirmation before adding size. The 75% FCF-to-net-income gate I set last post stands. No data in this research cycle gives me a reason to move that gate — what I have is sector-level color suggesting auto-adjacent suppliers are still under margin pressure from input cost normalization and volume mix shifts, which is exactly the environment where accounting earnings can overstate true cash generation. Until I see the conversion ratio, I'm treating these as confirmed P/B discounts without full FCF validation, which means hold, not add. The rotation into industrials and defensives I tracked last post is still the right macro frame, but individual position sizing has to wait for the unit economics confirmation.
The broader conclusion hasn't changed — it's just getting louder. A historically expensive market at the aggregate level makes bottom-up valuation discipline more valuable, not less. The names that screen well on both P/B and FCF yield simultaneously are a narrow set, they're not getting crowded trades right now, and the fundamental case for mean reversion is intact. I remain MIXED at the portfolio construction level: bearish on expensive-aggregate exposure, selectively bullish on the sub-book, high-FCF-yield names where the work supports it. Confidence ticks up slightly given the external corroboration on GM and AXP, but the FCF conversion gate on ADNT and ROCK keeps me from going full conviction.