Private pre-deployment quantitative R&D initiative
Version 4.0.0 candidate · Unreleased · Last verified 2026-09-01
Informational documentation only. Nothing in this repository is investment advice, an offer, a solicitation, or a representation of legal registration, regulatory authorization, trading-system deployment, trading activity, or investment performance.
Disuza Quantitative is a private, pre-deployment quantitative R&D initiative building and validating rule-based systematic trading infrastructure. Its present work concerns research governance, validation discipline, software design, and the controlled development of a systematic research stack.
This repository is the initiative's public documentation surface. It does not contain a production trading platform, proprietary source code, model weights, credentials, private datasets, or a performance record.
Every architecture or capability statement in this repository must use one of these labels:
| Label | Meaning |
|---|---|
| CURRENT | Publicly supportable state at the date shown in the document. |
| IN PROGRESS | Work is being designed, implemented, or evaluated; completion is not claimed. |
| TARGET | Intended future state; not a statement of present capability. |
The machine-readable source for these definitions and the repository's current
positioning is PUBLIC_FACTS.yml.
- CURRENT — a private quantitative R&D initiative, a documentation and governance corpus, and research tooling under controlled development.
- CURRENT — forward-only market-data capture and monitoring are operational for research data collection. They do not place orders, operate accounts, connect capital, or constitute a trading runtime.
- CURRENT — a bounded execution-engine contract is qualified on deterministic synthetic fixtures only. It has not processed market data and does not qualify a strategy, an economic result, paper execution, live execution, or deployment.
- IN PROGRESS — validation controls, data and execution abstractions, risk controls, and their supporting software are being developed and evaluated.
- TARGET — an auditable, deployment-ready systematic research and execution stack, subject to explicit qualification gates.
Descriptions of a target architecture do not imply that its components are integrated, deployed, connected to capital, or operating in markets.
- CURRENT — the programme research scope covers BTC and ETH perpetual markets and NQ and ES index-futures markets.
- CURRENT — optional CFD adaptations are evaluated as a separate research lane.
- CURRENT — the primary design horizon is approximately 30 minutes to 4 hours.
- CURRENT — primary signal generation is rule-based. Machine learning may be evaluated only in auxiliary roles and only after the related hard-rule baseline qualifies out of sample.
- CURRENT — the current synthetic engine route uses purged, embargoed, anchored walk-forward evaluation and Deflated Sharpe Ratio (DSR).
- CURRENT — Combinatorial Purged Cross-Validation (CPCV) and Probability of Backtest Overfitting (PBO) are programme methods only where applicable; they are not claimed for the current synthetic engine route.
Programme research scope is not qualified engine coverage or current trading activity. The narrow engine contract does not establish coverage of the programme research universe or design horizon.
These methods reduce avoidable research error. They do not establish economic value or future performance.
The following explicit non-claims apply:
- No live or paper trading, production trading platform, customer-facing trading service, order placement, account operation, custody, or capital-management activity is represented.
- No historical or expected return, validated alpha, or investment-performance claim is made.
- No investment service, investment advice, financial product, offer, or solicitation is made.
- No company-registration or financial-services-licence claim, regulatory approval, or authorization is represented.
Research methodology can reduce avoidable error; it cannot establish future performance.
| Section | Scope |
|---|---|
| Documentation index | Reading order and status conventions |
| Public status | Current work, controls in progress, and target direction |
| Overview | Initiative scope and public boundaries |
| Research methodology | Rule-based policy and anti-overfit controls |
| Architecture | Current, in-progress, and target system views |
| Components | Component responsibilities and status |
| Data pipeline | Data-quality and lineage design |
| Execution | Execution-control design, not live execution |
| Risk | Research and operational risk controls |
| Operations | Pre-deployment operational readiness |
| Technology | Tools used or evaluated in R&D |
| Regulatory | Public disclaimers and non-claims |
| FAQ | Short answers about scope and status |
| Team | Public project roles |
| Downstream alignment | Post-release checklist for surfaces excluded from v4.0.0 |
The public repository intentionally stays at a documentation-safe level. It
must not include secrets, private identifiers, detailed control thresholds,
private infrastructure coordinates, proprietary algorithms, or unpublished
research results. See CONTRIBUTING.md for the publication
rules.
General inquiries: contact@disuza.com.
Unless a file says otherwise, the current public documentation is licensed
under Creative Commons Attribution 4.0 International. Historical
licence grants remain valid for the exact material and versions to which they
were originally applied; see NOTICE.md.
Disuza Quantitative · Public Technical Reference · v4.0.0 candidate
Unreleased · Last verified: 2026-09-01