Supply Chain Bottleneck Hunter

Problem

Traditional research often starts with large caps and market narratives, then works backward to justify a stock. In super-trends such as AI infrastructure, energy transition, and semiconductor reindustrialization, the harder question is which physical supply chain link will bind first. Bottleneck Hunter flips that process: it validates whether a trend is durable, physical, large, and accelerating, then breaks it down into materials, equipment, components, and test steps to find concentrated, slow-to-expand, hard-to-substitute chokepoints.

How it works

• Trend screening: starts from $ARGUMENTS or the default trend list, then checks industry forecasts, capex, and demand-versus-capacity expansion.
• Physical decomposition: moves beyond concepts to items such as HBM, optical modules, lasers, InP substrates, SOI wafers, test, and substrates.
• Bottleneck scoring: evaluates supplier concentration, ramp time, customer dependence, and revenue purity, with at least two independent sources per key conclusion.
• Valuation gate: requires PS, PE, TAM, and growth checks; downgrades loss-making names, extreme valuations, and post-offering momentum spikes.
• Living map: writes to reports/bottleneck-map/, maintains master-map.md, watchlist.md, and dated scans, and supports an hourly report-only-when-signal mode.

Boundaries

It is best used for industry-chain research, bottleneck tracking, and early name screening, not as a direct buy or sell order. Results depend on search freshness, disclosures, and supply-chain data; missing data must be labeled as uncertain rather than filled with narrative. First-layer bottlenecks that are already fully priced, or names with stretched valuations, should be downgraded or parked on the watchlist.