A full-stack goal-based investment platform for planning financial goals, evaluating feasibility, analyzing mutual funds, understanding tax-optimization opportunities, and generating data-driven fund recommendations.
The platform is designed around a goal-first investment workflow:
Goal Details
↓
Feasibility Analysis
↓
Strategy Preview
↓
Tax Optimization Insights
↓
Fund Evaluation
↓
Deterministic Recommendation Ranking
↓
User Review & Confirmation
The system separates deterministic financial calculations from explanatory/intelligent layers so that recommendations remain traceable and verifiable.
Users can create goals such as:
- Vacation
- House
- Car
- Education
- Wedding
- Retirement
- Healthcare
- Custom goals
Goal inputs include target amount, target date, SIP/lumpsum contributions, risk level, inflation assumptions, priority, deadline flexibility, and importance.
The feasibility engine evaluates whether the planned investment can reach the target.
Possible outcomes include:
ACHIEVABLESTRETCHEDDIFFICULTUNREALISTIC
Alternative strategies can be generated, such as increasing contributions or extending the timeline.
The platform evaluates funds using:
Metric Purpose
CAGR Annualized historical return
Volatility Historical return variation
Maximum Drawdown Peak-to-trough decline
Sharpe Ratio Return relative to total risk
Sortino Ratio Return relative to downside risk
Alpha Benchmark-relative excess performance
Beta Sensitivity to benchmark movements
R² Benchmark explanatory power
Each fund has a separate data-confidence classification:
HIGHMEDIUMLOWINSUFFICIENT
Confidence considers historical observations, data freshness, available history, and peer reliability.
Funds with insufficient data are not assigned a recommendation score.
Funds are ranked against peers in the same subcategory instead of being returned alphabetically.
The recommendation score is normalized to 0–100.
Higher CAGR, Sharpe, and Sortino generally improve the score, while lower volatility and smaller drawdowns improve the score.
Missing metrics are handled through weight re-normalization, subject to the engine's eligibility rules.
The score is persisted separately as:
asset_metrics.recommendation_score
Benchmark infrastructure supports benchmark-relative analytics through:
- Benchmark mappings
- Historical benchmark observations
- NIFTY data ingestion
- Benchmark repository
- Benchmark ingestion service
This supports Alpha, Beta, R², and rolling consistency calculations.
Tax information is presented separately from fund ranking.
The tax layer can provide educational insights around:
- Section 80C
- ELSS
- Equity mutual-fund taxation
- Debt mutual-fund taxation
- ELSS lock-in considerations
- Potential tax-saving capacity
Tax advantages do not directly override fund-quality ranking. Tax information is educational and should be independently verified where professional advice is required.
Frontend
│
▼
FastAPI API
│
├── Goals
├── Feasibility
├── Recommendations
├── Funds
├── Universe
├── Portfolio
├── Simulation
├── Retirement
├── Emergency Fund
├── Dashboard
├── Chatbot
└── Tax / Strategy
│
▼
Service Layer
│
├── Goal Strategy
├── Feasibility Engine
├── Analytics Engine
├── Metrics Engine
├── Recommendation Engine
└── Tax Opportunity Engine
│
▼
Repository Layer
│
▼
Supabase
│
├── Goals
├── Fund Universe
├── Asset Metrics
└── Benchmark Data
.
├── backend/
│ └── app/
│ ├── core/
│ ├── db/
│ │ └── migrations/
│ ├── middleware/
│ └── modules/
│ ├── auth/
│ ├── chatbot/
│ ├── dashboard/
│ ├── emergency_fund/
│ ├── feasibility_engine/
│ ├── funds/
│ ├── goals/
│ ├── learning/
│ ├── portfolio/
│ ├── profile/
│ ├── recommendation/
│ ├── retirement/
│ ├── simulation/
│ └── universe/
│ ├── providers/
│ └── recommendation/
│
└── frontend/
└── src/
├── components/
├── pages/
├── services/
└── types/
POST /api/v1/goals
GET /api/v1/goals
GET /api/v1/goals/{goal_id}
PUT /api/v1/goals/priorityPOST /api/v1/goals/strategy/preview
POST /api/v1/goals/strategy/finalize
POST /api/v1/goals/strategy/recommendationsPOST /api/v1/goals/feasibility/preview
POST /api/v1/goals/feasibility/applyGET /api/v1/universe/recommendations
POST /api/v1/universe/recommendations/computeThe application uses Supabase/PostgreSQL.
Important data areas include:
goals
asset_universe
asset_metrics
fund_cache
benchmark_mapping
benchmark_historical_observations
Recommendation scores are stored in asset_metrics.recommendation_score
and indexed for efficient ranking.
cd backend
python -m venv venvWindows:
venv\Scripts\activateInstall dependencies:
pip install -r requirements.txtRun the API:
uvicorn app.main:app --reloadHealth check:
GET /health
Expected response:
{
"status": "ok"
}Use the project's environment configuration / .env.example.
Typical backend variables include:
SUPABASE_URL=
SUPABASE_ANON_KEY=
SUPABASE_SERVICE_ROLE_KEY=
SENTRY_DSN=
ENV=
ALLOWED_ORIGINS=Never commit real secrets.
Run the backend test suite:
venv\Scripts\pytestThe test suite covers goal planning, feasibility, recommendation scoring, confidence handling, benchmark providers, fund operations, universe refresh, and strategy/tax logic.
External Fund Providers
↓
Provider Layer
↓
Universe Ingestion
↓
asset_universe
↓
Historical NAV Data
↓
Analytics Engine
↓
asset_metrics
↓
Benchmark Comparison
↓
Recommendation Scoring
↓
recommendation_score
↓
Ranked Fund Recommendations
The same inputs and data should produce reproducible recommendation results.
Fund selection should be based on evaluation metrics rather than alphabetical ordering or arbitrary selection.
The platform distinguishes the quality of the data from the quality of the fund.
Metrics, confidence, recommendation scores, feasibility, tax insights, and user preferences remain separate concepts.
Users should be able to review feasibility, alternatives, strategy, tax considerations, and recommendations before confirming a goal.
The system should avoid producing misleading recommendations when sufficient data is unavailable.
The backend uses authentication, user-scoped Supabase clients, Row Level Security, environment-based secrets, and sanitized error handling.
Never commit:
- JWT tokens
- Supabase keys
- Passwords
- Other credentials
This project is an investment-planning and educational decision-support application. Historical performance does not guarantee future returns. Tax rules and market conditions can change. Tax-related information should be independently verified, and users should consult qualified financial or tax professionals where appropriate.
The platform is actively evolving toward a complete interactive goal-based investment decision-support system.
Current major layers include:
- Fund data ingestion
- Fund analytics
- Benchmark evaluation
- Data confidence
- Recommendation scoring
- Goal feasibility
- Strategy planning
- Tax optimization insights
- Interactive recommendations
- User confirmation