A daily automated research desk for one central question: How close is the AI complex to breaking — and when does it get cheap again?
A handful of mega-cap companies are pouring hundreds of billions of dollars a year into data centers, GPUs, and power generation, betting that artificial intelligence revenue arrives fast enough to pay for it. That single trade now dominates equity indices, credit markets, utility buildouts, and even memory semiconductor prices (see our companion project, GlutClock). When one theme gets this big, its cooling curve matters to everyone.
BubbleWatch is an open-source system that wakes up daily, ingests public market and credit signals — price momentum in the 22-name AI complex, crowding vs the equal-weight market, index concentration, high-yield credit complacency, proximity to 1-year highs, and crypto speculative appetite — compresses them into a single Bubble Temperature (0–100), classifies the market regime, publishes an out-of-sample 5-day direction forecast, and scores itself publicly against a permanent constant-0.50 baseline.
It is a thermometer, not an oracle. The point is transparent calibration and early warning, not trade tips.
Displacement ──► Boom ──► EUPHORIA ──► Profit-Taking ──► PANIC ──► Capitulation Trough
(new tech) (FOMO) (temperature (smart money (broad-based (best long-term
maxes out) quietly exits) liquidation) entry zone)