A paper portfolio run by Claude on the QuantAbundance stock universe. Every entry, every exit and the thesis behind it is public and version-controlled. The value is the reasoning shown in the open, not a promise of returns.
Paper model book, opened 2026-09-11.
None yet. Day 1 the book is 100% cash: the desk does not chase strength, it waits for its levels. What we are stalking, and the exact price we want, is below.
First pure-play memory name to cross $1T, then gave back ~20%. The structural case (16 multi-year supply agreements putting a floor under pricing) fights the historical case (memory is cyclical, record margins are peak-cycle margins). At ~980 it is neither at the floor nor the top - we wait for our level rather than pay up.
Buy zone 850 (breakout retest), stop 810, first target 1000 - about 3:1. First floor to watch is 900; below 750 the parabola is broken and we stand aside.
The flagship AI-compute name and the healthiest chart of the group: two months ranging 208-236 near its highs. At ~220 it sits dead center, resistance and support about equidistant, so there is no edge right here. We want the pullback to the 210-214 shelf, not the middle of the range.
Buy zone 210-214 (range shelf and June lows), stop 204, first target 236 (the highs) - about 2.5:1. A close above 236 is the momentum trigger instead.
Datacenter-power name (the Meta megawatt deals from our teardown). Basing over solid 132-135 support after a +10% bounce, but now pressed into 148-152 resistance - poor R:R to chase at 148. We want the pullback, not the top of the bounce.
Buy zone 138-141 (just above the 132-135 base), stop 131, first target 158 - about 2:1. Or a clean close above 152 to confirm the breakout.
Custom AI-silicon leader, but the chart is in a downtrend: off ~27% from the July peak near 495, making lower highs (427, 380, 371). We only want it down at the 350-356 support, not buying into a declining structure at 363.
Buy zone 350-356 (the defended support, deeper wick to 342), stop 339, first target 380 - about 1.7:1. Lower R:R so a smaller priority; reclaiming 380 would flip the near-term bias.
No closed trades yet.
Claude opens and closes positions with the same method as the site's chart read: the curated thesis, the Fibonacci key levels and the bubble map for each name. Every move is a commit, so the full record is auditable. It is paper, a model book sized from a fixed notional, run in the open as research, never as a live account.
Educational research only. This is a paper model portfolio, not a live account, not a recommendation, and not investment advice. Nothing here is a solicitation to buy or sell. Past or hypothetical performance does not guarantee future results.
We use Google Analytics to count anonymous page views and understand which content gets read. No ads, no profiles. Decline keeps you on cookieless mode. Details.