PodcastThe Monetary Matters NetworkJacob Pozharny

Why Fundamentals Don't Work For "New Economy" Stocks | Jacob Pozharny

Traditional valuation metrics break down for intangible‑heavy ‘new economy’ firms because accounting rules understate ROE and book value, causing price‑to‑book relationships to flatten or reverse across sectors. The guests describe a supervised, topic‑level textual pipeline and borrow‑fee signals that, combined with PB–ROE mapping, reveal idiosyncratic opportunities across shipping, non‑US energy and mid/small‑cap international stocks, while warning that ultralow multiples can reflect policy risk. Their approach relies on scale (~10,000 securities), tailored horizons (sentiment ≈3 months; fundamentals ≈12 months) and realistic batting‑average targets (~53–55%) to generate repeatable alpha.

Market ValuationsArtificial IntelligenceEquities

Preview · 14 more points in the app

Read the Full Summary

Key Takeaways

Classical price‑to‑book (PB) vs ROE valuation relationships hold for low‑intangible, old‑economy industries but flatten or invert in intangible‑heavy sectors (pharma, semis, software), meaning standard PBRE models mislead unless adjusted for intangible intensity.

IFRS/US GAAP accounting that does not capitalize much R&D and customer‑relationship investment mechanically depresses reported ROE and book value, making PB multiples look artificially high for R&D‑intensive firms and requiring adjusted metrics or supplemental analysis.

The team buckets intangibles into Brand, Customer Relationships and IP to map stocks (e.g., Amazon/Netflix/Costco as customer‑relationship driven; MRNA/Pfizer/Meta as IP‑driven), which helps explain why PB–ROE signals break down and directs bottom‑up work.

A supervised, topic‑level textual layer (Computex) ingests ~2 million sentences from ~10,000 companies to generate topic‑specific sentiment (sales, margins, cash), deliberately avoiding black‑box LLMs to preserve point‑in‑time validity and use the textual signals as an externality check on quantitative screens.

Because sector‑level sentiment is often priced quickly, the team matches horizons to signal speed (sentiment ≈3 months, aggregate/fundamental ≈12 months) and relies on scale — many bets across ~10,000 securities — to monetize modest per‑bet edges.

Real‑time microstructure signals such as borrow availability and borrow‑fee spikes provide a near‑term read on hedge‑fund short initiation or covering, and when combined with textual tone can flag actionable short opportunities (e.g., names with spiking borrow demand plus deteriorating call‑tone).

TECHNOLOGY

Valuation Breakdown — The historical positive link between price‑to‑book and ROE holds in low‑intangible, old‑economy industries but flattens or reverses in high‑intangible sectors (pharma, semiconductors, software); sector sentiment and returns diverge—software shows weak sentiment/returns, semiconductors improving, and energy can be an outlier where sentiment outpaces returns—so valuation models require contextual adjustment for intangibles.

“What we see is in fact the market is recognizing what it is that I'm I'm I'm describing…”—

Intangible Accounting — IFRS/US GAAP non‑capitalization of many R&D and customer‑relationship investments mechanically depresses reported ROE and book value, making P/B multiples for R&D‑intensive firms (e.g., pharma) appear artificially high and reducing the predictive power of classical DCF/fundamental metrics unless adjusted.

“What he finds and what we confirm and extend to international markets is that the at the cross-section fundamentals have become much less predictive of price action for companies of…”—

MARKETS

PB–ROE Mapping — Updated PBRE/PB–ROE analysis shows the valuation–profitability link holds in low‑intangible industries but breaks in intangible‑heavy sectors; they classify intangibles into Brand, Customer Relationships and IP and map stocks accordingly (e.g., Amazon/Netflix/Costco = customer‑relationship; MRNA/Pfizer/Meta = IP).

“There's this model called the PBRE model…”—

Textual Pipeline — Their on‑market pipeline ingests ~2 million sentences from ~10,000 companies (earnings calls, filings) to generate topic‑level textual signals that act as an externality check on quantitative screens, catching meme dynamics, corporate actions and regulatory breaks (e.g., conflicting Petrobras signals).

“So when we look at this, we take in about two million sentences…”—

Borrow Signals — Prime‑broker feeds on borrow availability and fees give a near‑term read on hedge‑fund short activity because short orders typically require securing a borrow first; large day‑to‑day moves in supply/fees can telegraph significant short initiation or covering.

“Before any sell order is placed, sell short order is placed. There needs to be a borrow that is associated with that…”—

FORECASTS

Analyst Predictability — Poorly performing sell‑side analysts tend to follow stronger, more predictive analysts; by identifying which analysts' FY1 estimates actually predict reported EPS, the team can anticipate consensus FY1 EPS direction ahead of the market.

“The reality is that some sellside analysts are much better than others and the poorly performing cellside analysts know who they are and they tend to follow the analysts that…”—

Many Bets — With a modest per‑bet edge (~51–53%), they run very large coverage (~10,000 securities) and tilt toward mid/small‑cap ex‑U.S.; signal horizons vary (sentiment ≈3 months, fundamentals ≈12 months) and holding periods are matched to signal speed.

“One thing that stands out to me is just how many individual sets of data there are and it's kind of in the quantitative world it's kind of like blackjack…”—

Related Episodes

Important Disclaimers

Not Investment Advice: For education only — not financial, tax or legal advice.

AI-Generated: Made by AI from source transcripts and may contain errors.

No Guarantees: Past performance does not predict future results; investing carries risk.

Our Commentary · Request Removal