InvestmentResearchAI: probabilistic modelling and evaluation protocol

Public research documentation describes probabilistic asset-relative modelling, time-separated evaluation and explicit limitations, with a scheduled artifact-generation definition.

L2Research workflow Source Repository-documented Outcome Not empirically evaluated here

Evidence & verification

Source level

Repository-documented
Grounded in a maintainer repository/README. Availability, stars and forks do not establish research benefit or correctness.

Outcome verification

Not empirically evaluated here
A documented resource or intended workflow, not a demonstrated academic outcome. No empirical benefit is established by its inclusion.

Sources checked

2026-10-19 (UTC)

Source scope: README, model and news-labelling guides, and a workflow definition as documentation only. Documentation/reference inclusion does not establish an empirical research benefit.

Source-checked does not mean independently reproduced. See the Method page for how these labels are assigned.

Sources (4)

Research applicability

Documented financial/economic model research and evaluation; no investment recommendation.

Getting started

Read the linked primary sources and the stated adoption requirements. Reuse suggestions are editorial and untested; no setup recipe or trial was executed.

Maintenance

Sources last checked on 2026-10-19. Future availability, maintenance, provider terms and compatibility were not verified.

License

Repository licence, data rights and provider terms were not audited.

Workflow & human oversight

Intended L2 for researcher-configured economic modelling: time-ordered data and splits, a small MLP ensemble, calibration, shadow evaluation, challenger criteria and prospective records. A separate news-labelling protocol prescribes blinded labels and held-out comparison. Scheduled automation is not evidence of successful runs, agency or predictive advantage.

Workflow level is not an evidence score; higher is not better evidence.

Approach

Documentation describes a NumPy MLP ensemble for asset-relative predictions over several horizons, with train/validation/shadow splits, embargoes, calibration and challenger criteria. News annotation has a separate protocol excluding future-return information. A workflow invokes tests and pipeline scripts; their execution was not verified.

Limitations

README, model method, news-labelling guide and one workflow definition were read as text. Scripts, weights, generated artefacts, news labels, accounts and the deployed UI were not accessed or executed. CI schedules and permissions are not successful-run evidence. Demo records may remain until valid results exist. The source discloses missing point-in-time universe data, economic cost/performance evidence, formal PBO/DSR and adaptive online conformal evaluation. Cited studies were not audited. No profitability, financial advice, trading action or validated research result is established. Moving HEAD documents are unpinned and non-atomic.

Reproducibility

Source text was checked only within this scope: README, model and news-labelling guides, and a workflow definition as documentation only. No software, dataset, experiment or result was independently reproduced.

Adoption requirements

Untested suggestion: establish legitimate data, point-in-time lineage, independent evaluation and access boundaries before a research trial.

Comments

No comments yet.

Add a comment

Comments are held for moderation before appearing. Critical or corrective comments are welcome — especially evidence that an entry overstates what has been verified.