The AI complex, studied as one organism.
Bubble is a member of the Blaque Baux family. The core repo is the engine and blueprint — a governed, systematic platform (Julia) with a venue-agnostic execution controller and a Layer-3 live-money safety gate. Bubble points that engine in its own direction and inherits the governance wholesale.
Not investment advice. Educational/research software. Nothing here is validated. See LICENSE.
git clone --recursive https://github.com/blaquebaux/bubble.git
julia --project=engine -e 'using Pkg; Pkg.instantiate()' # one-time engine setupTreat every AI-associated name — chips, hyperscalers, model labs, power, thematic ETFs — as a single exposure. Two base threads apply: Boom's finding that such names trade as roughly one factor (diversification is an illusion), and Bleed's interest in Dragon-King / log-periodic bubble diagnostics.
- The AI basket as one factor — effective number of bets (Boom's participation ratio) across the complex.
- Bubble diagnostics — Sornette log-periodic / Dragon-King signals as a RISK flag, honestly tested for tradeability.
- Crowding-aware sizing — bound single-name and single-factor exposure.
Nothing above is implemented or validated. This is the map, not the territory.
Full detail in research/README.md. The scorecard:
| # | Question | Verdict |
|---|---|---|
| 1 | How one-factor is the AI complex? | ✅ crowding — 16 names → 3.8 bets, 49% one factor, 0.86 corr to SPY |
| 1b | Does the tangent supply chain change it? | ✅ confirmed — +8 supply-chain names (semicap/net/power) → 3.6 bets (added zero); neoclouds +~0.5 bets at the frontier; Intel (missed AI, corr 0.55) +0.2 bets — "chip supplier" ≠ "AI complex"; the crowd is the theme, not the industry |
| 2 | Is the bubble fadeable with price? | ❌ no — runup keeps rising (+8.6% fwd 60d); short-extended −0.36 |
The synthesis: the AI complex is one crowded factor (16 names ≈ 4 independent bets; spectrum Bio 36% < AI 49% < Bulk 58% < Basel 81%). Its huge 2016–2026 return is hindsight (AI was the theme, the Boom-Mag7 trap); the prospective lesson is crowding — size it as ~4 bets, not 16. And the bubble is not fadeable with price: after a big runup the complex keeps rising, and shorting the most-extended name loses (−0.36) while riding wins (+0.35). Diagnostics flag but don't time — "you cannot fade the prop" (Brute-Force) and "timing the tail removes the tail" (Bleed). Bubble is a risk/diagnostic sleeve: a factor to size and a tail to insure (via Bleed), never to short outright.
Research: first pass complete — crowding diagnostic; the bubble is not fadeable (research/).
A risk sleeve, not a systematic strategy. No live driver. Nothing validated to the spine's bar.
Blaque Baux is a quantitative research initiative and a subsidiary of Carter Warrens. BlaqueBaux.com is the home for the work; the code lives here on GitHub — open to study, test, and build bespoke strategies on top of.
Anyone can point an AI at a market. The edge is understanding what the data actually says — and turning it into something you can act on. We test relentlessly and put most of it on the record as rejected, with the reason; what survives is built, governed, and validated before it is ever called real. That combination — honest research, reproducible evidence, and execution you can trust — is why Carter Warrens leads on strategy and implementation, not merely uses the tools everyone now has.
This repo is one sleeve of the Blaque Baux family — a single governed engine steered in many directions. The core repo is the base/blueprint and holds the full family roster.
engine/ the Blaque Baux platform (git submodule -> blaquebaux/base)
research/ two Path-A sketches (one-factor crowding, is-it-fadeable) + scorecard
live/ governed live drivers (once a sleeve graduates to paper A/B)
MIT. (c) 2026 Carter Warrens.