Methodology · v1.0

How we score.

One number per name. Same recipe every week. Weights are public, inputs are auditable, methodology changes only when they're published in a dated post.

Formula

Score = 0.35·V + 0.25·R + 0.15·B + 0.15·S + 0.10·F

Valuation leads — the single best predictor of multi-year drawdowns in themed trades. Runway comes second: cash-rich companies absorb bad quarters; insolvent ones can't.

Inputs

V — Valuation (35%). Forward price-to-sales ÷ trailing growth. A stock at 20× sales growing 100% YoY looks very different from one at 20× growing 10%. Capped at 10.

R — Runway (25%). Years of operating burn covered by net cash. Scored inversely: <1 year = 10, >10 years = 0. Profitable names = 0.

B — Burn (15%). Trajectory of free cash flow. Rewards names turning the corner; penalizes names where the gap is widening.

S — Substance (15%). Does revenue exist? 0 = pre-rev slideware. 10 = profitable, cash-generative. Captures the gap between what's pitched and what's sold.

F — Froth (10%). Listing route (SPAC/reverse merger penalty), listing recency, dilution rate, retail speculation signals. Lowest weight; catches names other screens miss.

Tiers

Tier

Cutoff

Reading

Critical

≥ 6.5

Multiple inputs red. Thesis depends on continued hype.

High

5.5 – 6.5

Fragile on ≥2 inputs. Vulnerable to a bad quarter.

Elevated

4.0 – 5.5

Priced like a winner. Not immediately fragile.

Moderate

< 4.0

Real business, sensible balance sheet.

Data

Market data (price, cap, 52-week range) from end-of-day US exchange feeds. Fundamentals from the most recent 10-Q/10-K. Forward P/S from consensus analyst estimates; falls back to trailing when no estimate exists. Every number sourced and timestamped.

Universe

US-listed common stocks in one of the four themes (AI, Nuclear, Quantum, Space) with ≥1 quarter of reported financials. Excludes ETFs, non-US-reporting ADRs and names under $5M in cap. Reviewed quarterly.

What this is not

Not a price target. Not a "will fall by X% in Y weeks" prediction. A cross-sectional ranking: given the universe of hyped names in these four themes, who is standing on the thinnest ice relative to the others. Low score ≠ good stock. High score ≠ guaranteed drawdown. Hype can persist longer than any model expects.

Versioning

Methodology v1.0, published October 30, 2026. Weight changes or new inputs ship as a dated post with full before/after. Historical scores tag their version and are never retroactively rewritten.