Skip to main content
XGCS Signal Compass

Private applied R&D laboratory

Researching structure in noisy systems.

XG Capital Strategies studies systems whose structure is hidden in noise, and builds the instruments, infrastructure, and decision systems needed to test what it finds. Markets are the deepest of those subjects; they are not the only one.

Enter the research Bring us a research problem

Applied research / Where the work has landed

The laboratory also works on problems brought to it.

The research below is how XGCS decides what to believe, and the same method gets pointed at problems other people bring: a research question, a strategy, a system that needs architecture, a market, a transaction, or something that doesn't fit a category yet. Sometimes the result supports a decision. Sometimes it maps a field, tests a hypothesis, or produces a piece of software, a framework, or evidence that the original idea doesn't hold up. Past work has been strategy, process, and deal intelligence.

See the applied work

A sample of the range, not a services menu

And others across adjacent sectors that don't reduce neatly to an industry label. This is a record of where the work has landed, not a menu of services on offer, and it keeps growing in directions the list below doesn't predict.

  • Real estate professionals

    Deal evaluation and process design

  • Property developers

    Strategy and deal intelligence across project decisions

  • Construction companies

    Process design and operational decision support

  • Civic infrastructure non-profits

    Strategy and process for organisations without a commercial mandate

  • Mergers and acquisitions

    Deal intelligence for buyers and sellers

  • Syndication deals

    Structuring and evaluation support

  • Companies raising capital

    Advisory across several fundraising routes

  • Cryptocurrency mining operations

    Strategy and operational decision support

  • Web3 and digital-asset ventures

    Process design and diligence in an unsettled market

HOW THE LABORATORY WORKS

How does a research system learn from failure?

Most of what this laboratory produces is rejected. What happens to the rejections is the part that compounds.

  1. A candidate appears One way of reading the market, proposed by the search.
  2. It meets evidence The same question every candidate gets: is this real, or is it noise?
  3. REJECTED It was not distinguishable from noise. Its trajectory ends here.
  4. The result is kept The failure is not deleted. It is recorded as a place already tested.
  5. Another appears The next candidate starts against a record that already includes the first.

The five steps above are a schematic, drawn to explain the shape. The figures below are the recorded research record.

That is one trajectory. The search runs a great many of them.

Kinds of strategy the space allows
648
5 axes of behaviour, every combination a distinct niche
Niches actually reached
374
across 3 of 8 programs; the broadest has reached 150 of 648, or 23.1%
Rejections kept on the record
8,855
recorded across 4 programs, all but 101 attributed to a failure mode

The search records where it has already been and what it found there. That record is used to steer toward parts of the space nothing has reached yet, and to tell a region that was properly tested and came back empty apart from one that was never tested hard enough to say. It does not make a candidate exempt from any check.

Recorded XGCS research activity, sanitized for publication. Counts are process activity, not predictive quality, return, or confidence.

Explore the laboratories

XGCS / Living research census

What the laboratory is testing, and why.

RECORDED / VERIFIED AT EXPORT8 PROGRAMS

XGCS is testing whether different ways of reading market structure survive contact with evidence.

The eight lanes are real research programs. Each mark is a sanitized recorded event sampled from that program's bounded public history. The loop shows activity without implying live computation or successful prediction.

01Broad crypto searchstopped
02Broad cross-asset searchrunning
03Crypto searchrunning
04Crypto surf searchstopped
05Field direction searchstopped
06Field surf searchrunning
07Market field world modelrunning
08Volatility risk-premium searchrunning
Bounded recorded activity cycleEvent step 0 / 35

Looping replay / pauses offscreen

generated challenged retained rejected replay position
Who
Eight approved XGCS Heatmap Strategy Lab research programs.
What
Recorded research state changes across candidate-search programs, not trades or recommendations.
When
Website-safe snapshot exported Aug 26, 2026; it is not presented as a continuous live feed.
Where
Inside HSL research systems; proprietary formulas, topology, positions, and identities remain private.
Why
To search broadly while preserving evidence about both surviving structures and rejected ideas.
How
Programs generate, recombine, challenge, archive, reject, revive, and prospectively observe candidate structures.
Open the full census, archive, and failure memory →
This is a compact orientation view of recorded HSL activity, not a performance chart. Event volume and state changes are not quality, return, confidence, or predictive success.

Registry / Current public work

Questions before conclusions.

The public registry contains work with an inspectable public record: a stated question, visible evidence, and a boundary on what it does not show.

Project Research question Program state Lab / provenance Evidence mode
Ebb & Flow Can interacting pressure, momentum, confluence, and regime signals make timing-sensitive decisions more legible? PUBLIC FRAMEWORK XGCS / E&F
Revision date not published		 Conceptual record
Heatmap Strategy Lab How can diverse candidate structures be searched, challenged, retained, and remembered without exposing proprietary research logic? HASH-VERIFIED SNAPSHOT XGCS / HSL
Aug 26, 2026, 10:19 PM UTC		 Recorded research replay

Adjacent research programs

One laboratory. Different problem classes.

Valence, Tikkara, Berean, Portable Human Context, and the XG Content Engine are independent proving grounds for recurring XGCS questions: how signals become structure, how state remains auditable, and how evidence bounds a conclusion.

Research relationship only · Build labels are not evidence states

01

Valence

Intent coordination / exchange graphs

Build state:
OPERATING INTERNALLY
Surface:
Private implementation

02

Tikkara

Deterministic systems / causal provenance

Build state:
PROTOTYPE
Surface:
Implementation repository

03

Berean

Inspectable reasoning / evidence discipline

Build state:
OPERATING INTERNALLY
Surface:
Research prototype

04

Portable Human Context

Open interoperability standard / typed context

Build state:
RESEARCHING
Surface:
Draft specification with conformance examples, no reference server

05

XG Content Engine

Evidence communication / explanatory pipeline

Build state:
PROTOTYPE
Surface:
Internal pipeline, no published output
Read the research boundaries →

How / Research lifecycle

A candidate must become more than an interesting pattern.

It's a map of distinct research approaches that remain under consideration after structured challenge, not a scoreboard.

01

Map the behavior

Place each candidate by analysis family, reach, market regime, directional relationship, and cost sensitivity.

02

Challenge the evidence

Test whether the relationship survives holdout, placebo, persistence, spanning, and cost gates, not merely whether it fit once.

03

Retain the result

Current survivors occupy the archive. Typed failures stay in Failure Memory so null results remain part of the research record.

Inspect the full instruments → Open Signal Lab →

Record / Research library

Observations become durable records.

The weekly log is compiled from the research programs' own event logs, not written by hand. It records what ran, what started, what stopped, and what was discarded. Earlier writing from 2023 and 2024 is kept separately, reproduced as published.

Browse the research library → Read the weekly research log →

Applied Research / A few shapes it has taken

Work brought to the laboratory looks something like this.

Not a menu. XGCS accepts focused engagements when the question is real, the evidence can be inspected, and responsibilities remain explicit, whether or not it arrives already framed as a decision.

01

Market & regime research

Question

What is changing beneath the current consensus, and which assumptions are becoming fragile?

Work / output

Pressure maps, regime briefs, and scenario structure. The client retains decision ownership.

02

Forecasting & risk

Question

Which outcomes are plausible, what would signal a transition, and where is downside asymmetric?

Work / output

Forecast architecture, thresholds, exposure maps, and review cadence. Forecasts remain inputs, never guarantees.

03

Decision-system architecture

Question

How should evidence move from observation to recommendation, escalation, and human action?

Work / output

Inspectable system plans, control layers, and implementation sequences. Accountability remains with the operating organization.

04

Strategic technology research

Question

Which emerging capabilities are material now, later, or not at all?

Work / output

Technology maps, adoption sequencing, and bounded experiments. Novelty is not treated as evidence of fit.

Review applied research →

XGCS / Possibility space

Some research should remain research.

Kairos / Decision state

Some is ready to become infrastructure.

Kairos Dynamics is the separate company being formed to turn selected XGCS research into software other organizations can run. XGCS is intended to license the foundational technology; Kairos builds and sells the product. The licence is not yet executed.

Understand the transition →

Contact / Intake

Bring a decision, research problem, or system question.

The first conversation establishes the question, available evidence, decision horizon, and whether XGCS is the right laboratory for the work.

Open a research inquiry