Skip to content

Latest commit

 

History

History
93 lines (65 loc) · 4.26 KB

File metadata and controls

93 lines (65 loc) · 4.26 KB

Roadmap — Capital Markets Intelligence Platform

Current status: v1.0 complete — 6 proprietary models, 5 dashboards, 4 Excel workbooks, 4 PDFs, 4 research memos. Zero API keys required.

This roadmap outlines what we're building next. Contributions on any milestone are welcome — see CONTRIBUTING.md to get started.


v1.1 — Engineering Hardening (in progress)

Goal: make the platform production-grade and contributor-friendly.

  • Turn pipeline into a Python package with CLI entrypoint (cmi run data|analysis|outputs|all)
  • Add YAML/JSON configuration for tickers, countries, event windows, and scenarios
  • Refactor duplicated data-loading logic into shared utility module (cmi/data_utils.py)
  • Add structured logging and graceful error handling across the pipeline
  • Add pytest test suite with golden-data tests for core models
  • Add GitHub Actions CI (tests + lint on every PR)
  • Add type hints and mypy checks for core modules
  • Add Dockerfile and optional VS Code devcontainer
  • Add a cmi run happy-path command for fast end-to-end validation

v1.2 — Analytics Upgrades (planned)

Goal: deepen the quantitative models across all four verticals.

IPO Event Study

  • Add configurable estimation windows and additional statistical tests (rank tests, parametric vs non-parametric)
  • Add market-adjusted vs market-model comparison
  • Cross-check abnormal returns against an external event-study library
  • Add multi-event aggregation: CARs by sector, country, underwriter (with boxplots)

M&A Deal Screening

  • Move from simple correlations to logistic regression / tree-based classification model with held-out validation
  • Add sector-level M&A summaries (median premium, completion rate, typical deal size)
  • Add time-to-completion analysis as a function of deal characteristics

Sovereign Risk & Macro

  • Decompose sovereign risk index contributions by indicator (tornado charts per country)
  • Add rating-style mapping (IG/BB/B buckets) and watch-list flag for downgrade risk
  • Add commodity-linked stress scenarios (oil exporter vs importer differential)

Cross-Asset Regime Detection

  • Add Markov transition probabilities and regime persistence stats (half-life, expected duration)
  • Add a toy strategy backtest using regime calls (with appropriate disclaimers)

v1.3 — UX, Outputs & Documentation (planned)

Goal: make the platform feel like a real institutional research product.

Dashboards

  • Add a global navigation "home page" HTML linking all dashboards, reports, and workbooks
  • Add interactive filters (country, sector, date range) to Plotly dashboards
  • Standardize chart style, color palette, and fonts across all dashboards

Excel & PDF Reports

  • Add TOC, numbered sections, and figure captions to PDF reports
  • Add hyperlinks in Excel workbooks (tickers → Yahoo Finance, countries → World Bank)
  • Generate a "Data Inventory" appendix (source, coverage period, last updated) in both Excel and PDF

Documentation

  • Add a "Scenario Cookbook" in docs/ (how to add a new IPO, country, or scenario)
  • Add a full worked-example Jupyter notebook for each vertical (IPO, M&A, sovereign, macro)

v2.0 — Advanced Research Platform (future)

Longer-horizon ideas — contributions and proposals welcome:

  • Dynamic term structure models (Nelson-Siegel-Svensson)
  • Multi-factor risk models for cross-asset portfolios
  • Integration with open-source trading agents (FinRL, FinRobot) for research simulation
  • Optional LLM-powered memo generation (narrative summaries from structured model output)
  • Real-time data refresh mode (scheduled pipeline runs)
  • Web-based UI for non-technical users to configure and run analyses

Contributing

We are actively looking for help with:

  • Quant finance: regime-switching models, factor models, advanced event-study statistics
  • Python engineering: packaging, CI, type safety, testing
  • Visualization: Plotly dashboard polish, cross-asset chart design
  • Documentation: notebooks, cookbooks, written explanations of models

See CONTRIBUTING.md and browse open issues — especially those labeled good first issue or help wanted.