Mission

Build Power Fund: an investment intelligence system that helps find, evaluate, size, and manage opportunities before they become consensus trades — with capital preservation as a hard constraint and high growth as the objective when the opportunity set justifies it.

What we are building

Not a black-box trading bot. A research + decision + risk platform that:

  1. Ingests a wide spectrum of financial and company data (prices, filings, earnings, thematic signals — not charts alone).
  2. Surfaces candidates early relative to mainstream narrative and parabolic price moves.
  3. Supports quantitative analysis and repeatable workflows, with a human in the loop for live capital decisions (especially early on).
  4. Records theses, actions, and outcomes so the process improves over time.

Two roadmaps

Same phase numbers, different objects. Do not treat them as one ladder.

PlanLives inWhat it sequences
Softwareplan.mdCodify the PM → augment → expand perception → constrain risk → optional scale of the product
Capitalmandate.md$0–$75k seed → earn the next invested cap → earn most of the $250k book → only then consider outside capital

PowerFund does not allocate capital because capital is available. Capital is released as the investment process earns trust. While invested cost is well under the $75k capital Phase-1 cap, the job is evidence collection and process validation, not racing toward full deployment.

Engineering should not block on fund legal structure. Product and data design should not paint us into a single-user corner either. Software Phase N is not a prerequisite for capital Phase N.

Primary opportunity set

Thematic concentration where capital and policy are reshaping markets. The four core themes are not four unrelated sectors; they are one industrial transformation — a massive increase in the economic value of computation, electricity, and autonomous machines:

  • AI infrastructure
  • Energy (especially as a bottleneck to AI and electrification)
  • Robotics and AI
  • Defence

Other sectors are in scope when they feed the same stack or clear a high bar for asymmetry. See themes.md.

A retrospective “this watchlist would have doubled from 2024” is contaminated by survivorship: names are here partly because they already worked. The job is to find 2028 winners in 2026.

Success criteria

Near term maps to software Phase 1 plus capital Phase 1. Medium term maps to software Phases 2–3 and capital Phase 2. Long term is optional software Phase 4 / capital Phase 4.

Near term (personal operating system)

  • Clear written mandate and risk rules that are actually followed.
  • A usable research workflow: watchlists, company dossiers, signals, portfolio book, decision journal.
  • At least one automated scoring path beyond price technicals (e.g. fundamentals / CapEx / filings-driven).
  • Measurable process quality: fewer late chase entries; documented invalidations; post-mortems.
  • Honest trade labeling: "early" claims backed by evidence that price has not discounted them; otherwise labeled momentum-with-fundamentals with matching exit discipline (see mandate).
  • Decision-grade dossiers that state not only “would we own this?” but at what price we become unusually eager (normal / attractive / dislocation / panic vs thesis impairment).
  • Performance judged against the S&P 500 (success) and QQQ (style), on both NAV and the deployed sleeve. See mandate.md.

Medium term (intelligence edge)

  • Pipelines covering market data, filings/earnings, and thematic/alternative inputs relevant to our themes.
  • Explainable signals with “why now,” confidence, and crowding/valuation filters.
  • Portfolio construction aware of theme concentration, correlation, and drawdown controls.

Long term (optional scale)

  • Evidence that the process can generalize beyond one operator.
  • Path to research distribution or multi-investor capital without rewriting the core platform.

Non-goals (for now)

  • High-frequency trading or market-making.
  • Fully autonomous live trading without human review.
  • Maximizing coverage of every asset class on day one.
  • Raising outside capital before the process and track record exist.

Guiding principles

  1. Early ≠ obscure — early means evidence accumulating before consensus pricing, not microcap lottery tickets by default. Search for companies that control a bottleneck whose economics are changing faster than perception; do not search for “small” as a strategy.
  2. Preservation enables growth — survive wrong theses; size for asymmetry. Stretch return targets are scenarios, never objectives that override risk rules.
  3. Explainability — every actionable signal should be inspectable.
  4. Breadth of data, focus of action — ingest widely; concentrate decisions on the mandate.
  5. Process over prediction — journal, review, and risk rules compound; vibes do not.
  6. Dislocations are a return source — prepare before the scare (ranked watchlist, scenario values, pre-agreed size). Buy when price fell more than intrinsic value, not because something is down 25%. Cash and fully valued names that can be recycled are the option on that volatility.
  7. Capital is earned, not assumed — release more of the $250k only when the process has evidence. Availability of cash is not a reason to deploy it.