Investment Research & Portfolio Intelligence Platform #49

Issue type: Epic
Working title: Build Institutional Investment Research & Portfolio Intelligence Platform
Priority: P0
Target: Web application, responsive desktop-first
Objective: Turn the 10 research workflows in the reference designs into a unified, data-backed investment research application.

1. Product goal

Build a platform that lets an investor move through one continuous workflow:

Discover opportunities → research a company → value it → analyze catalysts/technicals → understand macro exposure → construct a portfolio → measure risk → monitor changes.

The application must distinguish calculated data, third-party data, AI interpretation, and user assumptions. AI must not fabricate financial figures when required market data is unavailable.

---

2. Core application shell

APP-001 — Dashboard & navigation

Create the application shell with:

Dashboard

Discover

Company Research

Portfolio

Macro

Watchlists

Reports

Settings

Dashboard widgets:

Portfolio value and performance

Watchlist

Market overview

Upcoming earnings

Macro calendar

Portfolio risk alerts

Recent research

Saved screens

Analysis shortcuts

Acceptance criteria

Users can navigate between every module without losing portfolio/research context. Global ticker search is accessible from every screen. Recent analyses can be reopened from the dashboard.

---

APP-002 — Global security search

Search by:

Company name

Ticker

Exchange

Company result should resolve to a canonical security record containing ticker, company, exchange, currency, sector and industry.

Selecting a company opens its unified research page.

---

APP-003 — Company research workspace

Every security gets a common company page with tabs:

Overview | Financials | Valuation | Earnings | Technicals | Quant | Competition | Ownership | News

Header displays current market information and data freshness.

Research generated in one module should be reusable by other modules.

---

3. Institutional Stock Screener

FEAT-001 — Stock Screener

Allow users to screen an investable universe using:

Market cap

Sector/industry

Geography

Revenue growth

EPS growth

P/E

Forward P/E

PEG

EV/EBITDA

Price/free-cash-flow

ROIC/ROE

Gross/operating/net margins

Debt/equity

FCF

Dividend yield

Payout ratio

Momentum

Volatility

Results table should contain relevant fundamentals and valuation metrics.

Support:

Column sorting/filtering

Saved screens

Comparison mode

Watchlist addition

CSV/XLSX export

AI-generated research may additionally explain valuation, growth, balance-sheet quality, dividend sustainability, competitive positioning, bull/bear scenarios and key risks.

Acceptance criteria: Screening values come from structured financial data rather than AI-generated estimates.

---

4. DCF valuation

FEAT-002 — DCF Valuation Engine

Inputs:

Revenue

Revenue growth

Operating margin

Tax rate

D&A

CapEx

Working capital

WACC

Terminal growth

Exit multiple

Net debt

Diluted shares

Generate a five-year forecast containing:

Revenue → EBIT → taxes → NOPAT → D&A → CapEx → working capital → unlevered FCF.

Calculate terminal value using both:

Perpetuity growth

TV = FCF(n+1) / (WACC − g)

and exit multiple methodologies.

Output:

Enterprise value

Equity value

Implied value/share

Current price comparison

Implied upside/downside

Base/bull/bear scenarios

Include WACC × terminal-growth sensitivity matrix and WACC × exit-multiple matrix.

Every assumption must be editable.

Acceptance criteria: DCF results recalculate deterministically whenever an assumption changes.

---

5. Portfolio risk

FEAT-003 — Portfolio Risk Engine

Portfolio input supports ticker + quantity, value or percentage.

Calculate:

Position concentration

Sector concentration

Geographic exposure

Currency exposure

Correlation matrix

Historical volatility

Portfolio beta

Maximum drawdown

VaR

Expected shortfall

Liquidity indicators

Interest-rate sensitivity where relevant

Provide stress scenarios such as:

Equity market decline

Recession

Rate increase/decrease

USD appreciation/depreciation

Technology selloff

Inflation shock

Create a risk heat map showing:

Risk | Exposure | Severity | Contributors | Potential mitigation

Do not present scenario estimates as guaranteed outcomes.

---

6. Earnings intelligence

FEAT-004 — Earnings Analyzer

For each company display:

Next earnings date

Last four quarters

Reported EPS

Consensus EPS

EPS surprise

Reported revenue

Consensus revenue

Revenue surprise

Historical price reaction

Show relevant company KPIs where available.

Examples:

Cloud company → ARR/cloud growth
Bank → NIM/deposits/credit losses
Retail → comparable sales
SaaS → NRR/RPO
Semiconductor → inventory/gross margin

Capture management guidance and compare it with consensus.

Where reliable options data is available, calculate market-implied earnings move.

Generate explicit bull/base/bear scenarios rather than pretending a single outcome is certain.

---

7. Portfolio construction

FEAT-005 — Portfolio Builder

User profile:

Investment amount

Investment horizon

Risk tolerance

Income requirements

Liquidity requirements

Account/tax context

Existing holdings

Contribution amount

Asset classes:

Domestic equities

International equities

Government bonds

Corporate bonds

Cash

REITs

Commodities

Optional alternatives

Support:

Core / satellite allocation

Output:

Target allocation

Current vs target allocation

Suggested rebalance amounts

Historical risk/return statistics

Benchmark

Contribution allocation

Rebalancing rules

Allow both ETF-level and asset-class-level construction.

Tax guidance must be contextual rather than presented as personalized professional tax advice.

---

8. Technical analysis

FEAT-006 — Technical Analysis Engine

Timeframes:

Daily

Weekly

Monthly

Calculate:

SMA 20/50/100/200

EMA

RSI

MACD

Bollinger Bands

ATR

Volume averages

Relative strength

52-week range

Detect potential:

Support/resistance

Breakouts

Trend changes

Moving-average crosses

Volume confirmation

Common chart structures

Support Fibonacci retracement as an optional visualization.

Interactive chart must allow indicators to be enabled/disabled.

Trade-planning mode can calculate hypothetical entry, invalidation/stop level, target and risk/reward from user-defined assumptions.

The UI should describe technical signals probabilistically rather than treating them as guaranteed forecasts.

---

9. Dividend research

FEAT-007 — Dividend Portfolio Analyzer

For each security display:

Dividend yield

Annual dividend

Payment frequency

Ex-dividend date

Payout ratio

FCF payout ratio

Consecutive growth history

3/5/10-year dividend CAGR where available

Dividend cuts

Balance-sheet context

Portfolio analytics:

Weighted yield

Estimated annual income

Estimated monthly/quarterly income

Sector diversification

Dividend concentration

Add DRIP simulator with:

Initial investment

Contributions

Dividend growth assumption

Price-growth assumption

Reinvestment toggle

Projection horizon

Generate 5/10/20-year scenarios.

Any internal dividend-safety metric must expose the factors behind the result rather than present an unexplained score.

---

10. Competitive intelligence

FEAT-008 — Competitive Landscape

User selects company or industry.

Identify an appropriate peer set and compare:

Metric Company Peer A Peer B Peer C

Market cap
Revenue
Revenue growth
Gross margin
Operating margin
FCF margin
ROIC
R&D
Valuation

Analyze competitive dimensions such as:

Brand

Cost position

Switching costs

Network effects

Scale

Distribution

IP

Regulation

Ecosystem

Where reliable data exists, show three-year market-share trends.

Provide sourced discussion of:

Innovation pipeline

Capital allocation

Industry threats

Company-specific risks

Potential catalysts

Avoid invented market-share figures when reliable data isn't available.

---

11. Quantitative pattern research

FEAT-009 — Quant Pattern Finder

User chooses security and analysis period.

Calculate historical:

Monthly returns

Day-of-week returns

Volatility

Drawdowns

Win rate

Return distribution

Autocorrelation

Relative performance

Seasonality report should show:

Period | Observations | Mean | Median | Win rate | Statistical significance

Event studies:

Earnings

CPI

Central-bank decisions

Other supported market events

Analyze returns around configurable windows such as:

T−5 → T+5

Additional datasets, where available:

Insider transactions

Institutional ownership

Short interest

Options activity

Important requirement: display sample size and statistical significance so random historical noise isn't described as a trading edge.

---

12. Macro intelligence

FEAT-010 — Macro Impact Dashboard

Track:

Policy rates

Yield curve

Inflation

GDP

Unemployment

Payrolls

Consumer spending

PMIs

Credit spreads

Dollar index

Oil

Major market indexes

Provide historical charts and changes over relevant periods.

Create a macro-to-portfolio exposure map:

Macro factor → affected holdings → mechanism → sensitivity → scenario impact

Support scenarios such as:

Higher-for-longer rates

Rate-cut cycle

Inflation resurgence

Recession

Soft landing

Dollar strengthening

Commodity shock

Separate observed economic data from AI-generated scenario analysis.

---

13. Cross-cutting features

FEAT-011 — Watchlists

Users can create multiple watchlists, add/remove securities and display price movement, valuation, earnings date and selected alerts.

FEAT-012 — Portfolio management

Support multiple portfolios, transaction/position entry, cost basis, cash and benchmarks.

FEAT-013 — Research reports

Any analysis can be saved as a report.

Report records should preserve the data snapshot and assumptions used at generation time, so historical reports don't silently change when live data changes.

Exports:

PDF

XLSX/CSV where appropriate

FEAT-014 — Sources & provenance

Every important externally sourced metric should retain:

provider + source + as-of timestamp

AI-generated conclusions should have access to the source material behind them.

FEAT-015 — Alerts

Support alerts for:

Price

Earnings

Valuation

Dividend changes

Technical events

Portfolio concentration

Macro releases

---

14. AI research layer

AI-001 — Research orchestration

Use AI for interpretation and synthesis, while dedicated services perform calculations.

Architecture:

User → UI → Research Orchestrator → Data Services → Analytics Engines → AI Interpretation → Report

The model should receive structured calculated results rather than being asked to calculate complex financial models from prose.

AI-002 — Grounding

Every generated analysis receives:

Security metadata

Market data

Financial statements

Calculated metrics

Relevant filings

Relevant macro data

User assumptions

AI must explicitly represent missing information rather than manufacture it.

---

15. Data architecture

Core entities:

User
InvestorProfile
Security
Company
MarketPrice
FinancialStatement
FinancialMetric
Estimate
EarningsEvent
Dividend
OwnershipRecord
InsiderTransaction
OptionSnapshot
MacroSeries
Portfolio
Position
Transaction
Watchlist
Alert
ResearchReport
AnalysisRun
Scenario
DataSource

Every time-sensitive financial record should contain:

source
as_of
retrieved_at
currency

---

16. Technical architecture

Recommended stack:

Frontend
Next.js + TypeScript
React
Tailwind
Trading/chart library

API
Python + FastAPI

Analytics
Pandas
NumPy
SciPy
statsmodels

Database
PostgreSQL

Cache
Redis

Jobs
Background worker / queue

AI
LLM research orchestration

Storage
Object storage for reports

Authentication
OAuth/OIDC

Financial data providers should sit behind a provider abstraction so the application isn't locked to one vendor.

---

17. Non-functional requirements

Performance: cached company pages <2s target; long research jobs asynchronous with visible status.

Auditability: store model version, source snapshot, assumptions and calculation version for generated research.

Security: encrypted transport/storage, no secrets in client code, server-side API credentials, user-level portfolio isolation.

Accessibility: keyboard navigation, accessible tables/charts and WCAG-oriented UI.

Responsive design: desktop optimized, tablet supported, core monitoring functionality usable on mobile.

---

18. MVP boundary

The first shippable release should implement the platform shell plus Stock Screener, Company Overview, DCF, Earnings, Technical Analysis, Portfolio Builder, Portfolio Risk, Watchlists, Reports and data provenance.

Dividend, Quant Pattern, Competitive Landscape and advanced Macro functionality can then build on the same security, market-data and analytics infrastructure.

The key architectural constraint is that we should build the financial data/analytics foundation once. Otherwise these ten features turn into ten independent AI prompts, which would be much less reliable and considerably harder to maintain.