The Silk - Core Forecast

ALB (Albemarle): Short ALB or reduce long exposure. Mean reversion at >3σ is the highest-confidence signal in the calibration dataset. Even with strong fundamental tailwinds, the velocity of the move (+16.3% in 1 day) is unsustainable.

AMD: Reduce AMD long exposure or hedge with puts. The 7d probability slope was flat (+0.0pp/day) suggesting the signal has been persistent — but persistence at >3σ increases, not decreases, reversion probability per calibration data.

NVDA: Neutral — insufficient edge at current sigma level with conflicting timeframe signals. Prior was 58%, updating to 50% based on 5d weakness and declining 7d slope (-1.2pp/day).

USD/CNY: Mild short USD/CNY but position size should reflect low edge (55% vs 50% baseline). The deceleration in velocity suggests the easy part of the move is done.

Crude Oil: Mild short crude oil bias. The trend alignment across timeframes and stable geopolitical backdrop support continuation. Position small given lack of calibrated signal.

Mean reversion in CRITICAL sigma assets (ALB, AMD): ALB (+3.79σ) and AMD (+3.68σ) at statistical extremes above 30d means. Base rate: 77% mean reversion within 6 days [n=1686]. Both have experienced parabolic 30d moves (ALB via lithium +29.6%, AMD +71.5%). Expect 5-15% pullbacks within 4-day trade windows. Bear case: these are momentum-driven by structural demand shifts (EV battery procurement, AI capex) that could sustain elevated levels longer than historical base rates suggest.

A novel class of electrochemical 'swing reactors' built on entirely new infrastructure: purpose-built modular facilities that synthesize and decompose lithium compounds in real-time, synchronized to commodity price oscillation cycles. Like a chemical oscillator (Belousov-Zhabotinsky reaction) that naturally cycles between states, these reactors exploit the kinetic energy of lithium's mean-reversion pattern — converting lithium carbonate into higher-value lithium hydroxide (or vice versa) precisely at the harmonic resolution point when prices snap back, profiting not from direction but from the TIMING of the cadence. The facilities require a new type of continuous-flow electrochemical architecture — novel membrane stacks, AI-driven reaction-timing controllers, and a dedicated sensor network for real-time commodity-chemistry coupling — none of which exist today.

A new class of internationally governed, purpose-built 'Compute Sovereignty Zones' — physical territories with novel governance frameworks, dedicated power generation (small modular reactors built on-site), and entirely new fabrication-to-deployment supply chains where nations collectively fund and operate advanced AI inference campuses outside existing cloud/hyperscaler paradigms. Value is captured through treaty-based compute credit issuance — a new financial instrument representing guaranteed future inference capacity — tradeable on dedicated clearinghouses, monetizing the post-phase-transition reality where advanced GPU compute becomes a civilizational resource akin to energy reserves.

Build a new hardware-financial infrastructure layer that records and prices interest-rate state transitions using purpose-built cryogenic quantum memory arrays, rather than conventional databases or existing market plumbing. As 2-year Treasury yields decline sharply, demand rises for ultra-fast repricing of refinancing, callable debt, structured credit, and mortgage convexity risk; this system captures value by selling premium access to a newly invented rate-state settlement network and licensing its pricing substrate to issuers, servicers, and dealers.