Research of Investor Attributes - Behavior Analysis Using Report of Possession and NLP
Research of Investor Attributes - Behavior Analysis Using Large Possession Reports and NLP
This research creates a database of "Reports of Large Possession" (5% rules) disclosed on EDINET using Natural Language Processing (NLP). We analyze activity characteristics for each investor type and investigate detailed investment behavior of activist funds.
Reports of Large Possession contain information such as the holding ratio of investors who own 5% or more of shares, the purpose of holding, and joint holders. By systematically collecting and analyzing these data, it becomes possible to quantitatively understand the behavior patterns and investment styles of market participants.
Key Research Areas
- Analysis of Investor Types: Characterization of investment behavior for each type, such as security companies, asset managers, trust banks, and activists.
- Activist Research: Detailed analysis of investment strategies, characteristics of held stocks, and shareholder proposal tendencies of individual activists.
- Natural Language Processing: Automatic classification of investment intent from text data such as the "purpose of holding."
- Multi-Agent Simulation: Analysis of disclosure documents and generation of questions from different investor perspectives.
Database Configuration
We build and utilize the following databases in this research:
| Database | Overview | Key Content |
|---|---|---|
| Activist_Action_DB | Large Possession Report DB | Structured data from reports on EDINET using NLP |
| Activist_Details_DB | Activist Detail DB | Detailed database of activity information of specific activists |
| tairyou_hoyu.db | Integrated Analysis DB | Integrated management of investor attributes, groups, and time-series data |
Investor Classification System
Organizational Types
| Type Code | Name | Description |
|---|---|---|
| ASSET_MANAGER | Asset Manager | Investment trust, Investment advisory companies |
| ACTIVIST | Activist | Activist funds that make shareholder proposals |
| HEDGE_FUND | Hedge Fund | Hedge funds using various strategies |
| INDIVIDUAL | Individual | Individual investors, founders, etc. |
Activist Analysis System
Analysis Modules
| Module | Analysis Content | Output |
|---|---|---|
| Portfolio Analysis | Investment destination composition, Sector dispersion, Concentration | List of held stocks, Statistics |
| Time-series Analysis | Transition of holding ratio, Purchase/Sale patterns | History of changes, Tendency |
| Purpose Analysis | Intent classification by text mining | Net investment / Important proposal actions, etc. |
Conclusion
Understanding investor behavior in the stock market provides important suggestions for IR strategy formulation, shareholder composition analysis, and corporate governance review. In this research, we aim to contribute to constructive dialogue between companies and investors by quantitatively grasping the behavior patterns of each investor type, maximizing the use of public information called "Reports of Large Possession."