AI Market Lab · Paper Portfolio

Frontier Shock Watch

A lightweight market game for tracking AI infrastructure, model-release shocks, and paper-money strategies before any real capital enters the arena. Research surface first; dragon-hoard later.

Not financial adviceManual update buttonLocal browser paper portfolioDeepSeek / frontier model events

The thesis is not “short Nvidia because China made a model.” The thesis is: frontier-model shocks move expectations around compute demand, inference margins, software leverage, and geopolitical chip ecosystems.

This page is a sandbox for testing that thesis with play money. It stores snapshots and paper trades in your browser only. Quotes and durable quote history are loaded from Foundry-hosted JSON files that Ash can update and push on request; if needed, paste CSV rows manually as a fallback. Either way, the point is discipline: thesis → watchlist → simulated allocation → postmortem.

Watchlist

Core names: AI chips, hyperscalers, infrastructure, and ETFs exposed to the DeepSeek-style shock cycle. Ask Ash to update the market lab, then press the button below to load the freshly pushed JSON and redraw movement over time.

No update yet.

Ticker	Name	Last	Change	Foundry history

Contrarian lens

Investor persona · Burry-inspired

Cassandra Value Bear

A deliberately skeptical lens modeled on public reporting around Michael Burry’s recent posture: wary of AI-infrastructure euphoria, interested in mispriced durable cash-flow software, and allergic to stories priced for perfection.

Bubble-awareValuation-firstOptions-risk literate

Current suspicion: chip and AI-infrastructure names may be discounting too much flawless growth.
Counterweight: established software/payment companies can benefit from cheaper AI while being less dependent on capex mania.
DeepSeek read: model efficiency is a threat to excess inference margins, not proof that AI demand disappears.
Game move: test hedges and underweights in paper mode before touching real capital.

Evidence note: based on public reporting/search results, not direct real-time filings. Treat as an interpretive persona, not Michael Burry’s actual live portfolio.

Strategy presets

Balanced AI Basket

Equal-weight broad exposure. Boring on purpose.

NVDA, AMD, TSM, AVGO
MSFT, GOOGL, AMZN, META
SMH, QQQ

DeepSeek Shock Hedge

Tests the idea that model efficiency pressures premium chip multiples while helping AI users.

Underweight: NVDA, SMCI
Overweight: GOOGL, META, MSFT
ETF ballast: QQQ

Compute Still Wins

Tests the counter-thesis: cheaper intelligence drives more usage, not less compute.

Overweight: NVDA, AVGO, TSM
Support: AMD, SMH, MSFT

Paper portfolio

Bankroll

Ticker	Weight	Entry	Shares	Value	P/L
Load a strategy, then start the game.

Operating rules before real money

No autonomous real trades. Ash can research, simulate, and propose; Christopher approves anything real.
Every thesis needs a time horizon, invalidation condition, and postmortem.
Shorting and leveraged products are advanced weapons. Paper-test them first.
DeepSeek-style releases are not automatically bearish. Efficiency can compress margins and expand demand at the same time.