[0.2.3] - 2026-04-14

"Ticker Keyword Intelligence + Correlation Reliability"

Added:

Added generate_keywords(ticker) in news.py to auto-build keyword lists for arbitrary tickers using Yahoo Finance metadata.
Keyword generation now includes ticker symbol, company name variants, and executive surnames (top officers).
Added timeout-bounded metadata fetch behavior so hung network requests cannot block analysis.
Added graceful fallback to [ticker] for timeout, network error, or unknown ticker.
Added Docs/CONTRIBUTORS.md.
Added 5 targeted tests in tests/test_core.py:
known ticker keyword generation
unknown ticker fallback
network failure fallback
timeout fallback
correlate_news substring false-positive regression
Test suite now totals 42 tests.

Fixed:

Fixed false-positive news correlation in signals.py by replacing substring matching with token-safe word-boundary regex matching.
Fixed stale data refresh behavior in main.py to avoid automatic stale-trigger reruns and require explicit user refresh action.
Fixed refresh epoch update consistency for manual refresh and scan-trigger paths.

Changed:

Added sidebar control in main.py to enable or disable auto background scans during active sessions.
Aligned background scan and manual refresh state transitions for more predictable rerun behavior.
Updated CI concurrency policy in ci.yml to preserve all main-branch runs while still canceling in-progress non-main runs.

Documentation:

Expanded code_flow.md with keyword generation flow and updated news-correlation behavior.
Expanded variable_list.md with new symbols/constants:
generate_keywords
_CORP_SUFFIXES
_KW_PATTERN_CACHE
_kw_re
Updated README.md:
fixed disclaimer/docs links
added Docker quick-start TOC entry
standardized scan command usage to python -m app.scan
added contributor documentation references
Updated CONTRIBUTING.md scan examples to python -m app.scan --dry-run.

Technical:

Added _KW_PATTERN_CACHE in signals.py to reuse compiled regex patterns and reduce repeated compilation overhead.
Added _kw_re helper for centralized safe keyword pattern creation.
Intentionally excluded industry/sector fields from generated keywords to reduce broad cross-asset noise.

Changelog:

https://github.com/The-Pulse-Engine/Pulse-Engine_Market_Intelligence_Platform/blob/main/Docs/CHANGELOG.md

[0.2.2] - 2026-04-12

"Dashboard Stability + Security + Test Expansion"

Added

• tests/test_logic_coverage.py — edge case coverage for signal scoring, sentiment, deduplication, and contradiction detection
• tests/test_storage_and_scan.py — storage round-trip, retention policy, dry-run scan, and synthetic backtest tests
• Signal score legend added to the sidebar for quick reference
• Loading spinner shown in the dashboard while live analysis is running

Changed

• Pinned runtime dependencies tightened after pip audit security review; no vulnerable packages remain in requirements.txt
• Dashboard cache invalidation logic reduced to avoid unnecessary reruns on stale data
• Dashboard stale-refresh handling tightened — refresh now triggers only when data is genuinely outdated
• Signal legend copy in sidebar clarified for readability

Technical

• Total test count increased from 14 to 37
• All test files use package-based imports consistent with the v0.2.1 modular restructure

---

Changelog - https://github.com/The-Pulse-Engine/Pulse-Engine_Market_Intelligence_Platform/blob/main/Docs/CHANGELOG.md

v0.2.1.0

[0.2.1] - 2026-04-07

"Modular Package Restructure + Asset Organisation"

Partial progress toward v0.3. Arbitrary ticker support, local installer, and open issue backlog (#10, #11, #12) remain outstanding before v0.3.0 is reached.

Changed

• Reorganized all top-level Python files into proper packages with __init__.py files:
  • app.py → app/analysis.py
  • scan.py → app/scan.py
  • backtest.py → app/backtest.py
  • dashboard.py → dashboard/main.py
  • ui_components.py → dashboard/components.py
  • styles.py → dashboard/styles.py
  • dashboard_data.py → dashboard/data.py
  • storage.py → storage/storage.py
  • config.py → config/settings.py
• All import statements updated to use absolute package-based imports (e.g. from config.settings import X, from storage.storage import X)
• config/settings.py BASE_DIR updated to use .parent.parent to correctly resolve the project root from the new subdirectory location
• Moved image assets out of the project root into dedicated subdirectories:
  • favicon.ico → assets/icons/favicon.ico
  • pulseengine_logo.png → assets/logo/pulseengine_logo.png
• Dashboard entry point changed from streamlit run dashboard.py to streamlit run dashboard/main.py
• Scan CLI entry point changed from python scan.py to python app/scan.py

Fixed

• Added sys.path.insert(0, ...) at the top of dashboard/main.py to ensure the project root is on sys.path when Streamlit is launched, resolving ModuleNotFoundError: No module named 'config' that occurred because Streamlit adds the script directory (dashboard/) to sys.path rather than the project root

Technical

• No logic, function names, arguments, or behaviour changed — pure structural reorganization
• src/ package (engine, price, news, signals, context, explanation, sentiment) remains unchanged in location; only its from config import statements updated to from config.settings import

v0.2.0 - Interface Overhaul & Performance Upgrade

##v0.2.0 - Interface Overhaul & Performance Upgrade

Rebranded to PulseEngine

This release introduces a major upgrade to both the interface and core performance of PulseEngine, focusing on usability, speed, and stability under load.

The dashboard has been fully redesigned to improve clarity and information flow. Signals, confidence levels, contradictions, and supporting news are now structured in a way that makes rapid interpretation easier, especially during active analysis.

On the performance side, several inefficiencies in the data pipeline and UI execution have been addressed. The system now avoids redundant computations, reuses fetched data where possible, and reduces unnecessary reruns—resulting in faster load times and smoother interaction.

Additionally, improvements were made to ensure the system behaves reliably even when external data (such as price feeds or news sources) is incomplete or delayed.

---

Key Improvements

Interface & UX

• Reworked dashboard layout with clearer separation of signal, explanation, and supporting data
• Improved visual hierarchy for faster scanning of key insights
• More consistent styling across components (cards, badges, sections)

Performance & Execution

• Reduced redundant processing during scans and UI updates
• Improved data reuse across components to avoid repeated fetches
• Optimized execution flow to minimize full reruns in the UI layer
• Faster response times during repeated interactions

Scalability

• Better handling of higher usage scenarios through reduced computation overhead
• More efficient data flow across the pipeline, improving responsiveness under load

Data & Stability

• Improved handling of missing or inconsistent price/news data
• More robust pipeline execution without breaking on partial failures

Storage & Efficiency

• Optimized snapshot storage to reduce unnecessary writes
• Improved data persistence approach for consistency and performance

---

This update significantly improves the core experience and lays the groundwork for future expansion.

Full Changelog: https://github.com/Codex-Crusader/le_Market_Intelligence_Platform/blob/main/Docs/CHANGELOG.md

Patch Release (v0.1.1)

v0.1.1 - Test Suite & Stability Update

This release introduces a lightweight test suite and cleans up the development workflow.

What’s new

• Added a minimal pytest suite (~14 tests)

  • Covers core logic invariants (RSI, ROC, sentiment, signal scoring)
  • Includes smoke tests for analyse_asset() and run_full_scan()

• All network calls mocked and storage isolated for safe testing

• Added a test maintenance guide for contributors

Improvements

• Simplified test architecture (removed over-engineered placeholder tests)
• Cleaned up unused and duplicate dependencies
• Updated documentation and developer setup

Why this matters

The project now has a basic safety net:

• Reduces risk of silent regressions
• Makes refactoring safer
• Improves contributor experience

Notes

• Tests are intentionally lightweight and flexible (early-stage project)
• CI/CD not yet configured (planned next step)
• No breaking changes

---

See the changelog for full details.
Full Changelog: https://github.com/Codex-Crusader/le_Market_Intelligence_Platform/blob/main/Docs/CHANGELOG.md

v0.1.0 - Initial Release

Initial Release (v0.1.0)

This is the first working release of the Market Intelligence Platform.

Features

• Multi-asset market analysis (commodities, crypto, equities, indices)
• Signal scoring engine (-10 to +10)
• News ingestion and sentiment analysis
• Market context detection
• Streamlit dashboard visualization
• Snapshot storage and basic backtesting

Running the App

Local

python scan.py --dry-run
streamlit run dashboard.py

Docker

docker build -t market-intel .
docker run -p 8501:8501 market-intel

Then open: http://localhost:8501/

Notes

• This is an early-stage release with ongoing experimentation
• Docker is currently provided as a deployment layer, not for development
• Dependencies are pinned for reproducibility

Known Limitations

• No automated tests yet
• No CI/CD pipeline
• Ongoing improvements to signal accuracy and performance