Smart Money Tracker — MF Intelligence Dashboard

SEBI mandates every mutual fund disclose their full portfolio holdings monthly — publicly, for free. This project scrapes those disclosures, cross-references with live stock prices, and builds a dashboard that answers: what are India's top fund managers actually doing with money right now?

Features

1. Smart Money Flow Detector

Across all large/mid/small cap equity funds, which stocks are being collectively accumulated or exited by top managers this month vs. last? Ranked by net change in aggregate holding %. Conviction scores show the highest-weight stocks across all tracked managers.

2. Fund Manager Style Fingerprint

For each manager (Prashant Jain, Rajeev Thakkar, Saurabh Mukherjea, Nilesh Shah, S Naren, etc.):

• Large vs mid vs small cap tilt
• Growth vs value score (P/E, P/B of holdings vs index)
• Sector concentration (Herfindahl index)
• Turnover ratio (portfolio churn)
• K-Means clustering to find "managers who think alike"

3. Style Drift Detector

Tracks each fund's actual market-cap breakdown over time. Flags funds drifting outside their SEBI mandate — a genuine red flag for investors. Shows mandate compliance gauges and severity ratings (critical/warning/minor).

4. SIP XIRR Calculator with Peer Benchmarking

Computes true XIRR (annualized return accounting for monthly cash flows — not fake point-to-point returns). Benchmarks against Nifty 50 and the best fund in the same category. SIP growth visualization shows how INR 1,000/month would have grown.

5. Portfolio Overlap Heatmap

Pick any 2–5 funds, see what % of portfolios overlap. Supports Jaccard (stock count) and weighted (by AUM %) methods. Shows common holdings detail and unique holdings per fund. Sunburst visualization of fund composition.

6. Alpha Attribution (Brinson-Hood-Beebower)

Decomposes fund outperformance into:

• Allocation effect — did they overweight the right sectors?
• Selection effect — did they pick the right stocks?
• Interaction effect — combined

Fund Managers Tracked (10+)

Manager	AMC	Funds
Prashant Jain	HDFC MF	Flexi Cap, Balanced Advantage, Top 100
Rajeev Thakkar	PPFAS MF	Flexi Cap, Conservative Hybrid
Saurabh Mukherjea	Marcellus	Consistent Compounders
Nilesh Shah	Kotak MF	Flexi Cap, Emerging Equity, Small Cap, Bluechip
S Naren	ICICI Pru MF	Value Discovery, Bluechip, Midcap, Multicap
Neelesh Surana	Mirae Asset MF	Large Cap, Midcap, Flexi Cap
Shreyash Devalkar	Axis MF	Bluechip, Midcap, Small Cap, Flexi Cap
R Srinivasan	SBI MF	Bluechip, Midcap, Small Cap, Flexi Cap
Anish Tawakley	ICICI Pru MF	Flexi Cap
Jinesh Gopani	Axis MF	Growth Opportunities

Tech Stack

Python | Pandas | Plotly Dash | scikit-learn | pyxirr | yfinance | BeautifulSoup

Data Sources (all free)

Source	What you get
amfiindia.com	Daily NAV for every mutual fund scheme
SEBI portfolio disclosures	Monthly stock-level holdings of every fund
yfinance	Historical stock prices for alpha calculation
Value Research scrape	AUM, category, fund manager name mapping
NSE/BSE	Market cap classification (large/mid/small) per stock

Setup

git clone https://github.com/eeshsaxena/smart-money-tracker.git
cd smart-money-tracker
pip install -r requirements.txt
python app.py

Open http://localhost:8050.

Deployment

Every data source is public, so this deploys with no secrets and no database. Holdings and NAV data are fetched on demand and cached under data/cache/.

Render — render.yaml is a ready blueprint; point Render at this repo and deploy. The free plan is sufficient.

Any Docker host:

docker build -t smart-money-tracker .
docker run -p 8050:8050 smart-money-tracker

Anything Procfile-based (Railway, Heroku) works as-is.

The server binds $PORT and runs under gunicorn as app:server. Requests that miss the cache scrape SEBI and AMFI, so the first load after a cold start is slow; the gunicorn timeout is raised to 180s to accommodate it.

Running Tests

pytest tests/ -v

Project Structure

smart-money-tracker/
├── app.py                              # Dash entry point
├── config.py                           # Fund managers, thresholds, peer groups
├── requirements.txt
├── src/
│   ├── scraper/
│   │   ├── sebi_scraper.py             # AMFI/SEBI portfolio disclosure scraper
│   │   ├── nse_classifier.py           # NSE/BSE market-cap classification
│   │   └── value_research.py           # Value Research metadata scraper
│   ├── analysis/
│   │   ├── smart_money_flow.py         # Cross-fund accumulation/exit signals
│   │   ├── style_fingerprint.py        # Manager fingerprint + K-Means clustering
│   │   ├── style_drift.py              # SEBI mandate compliance detector
│   │   ├── xirr_calc.py               # SIP XIRR + growth visualization
│   │   ├── overlap.py                  # Portfolio overlap (Jaccard + weighted)
│   │   └── bhb_attribution.py          # Brinson-Hood-Beebower attribution
│   └── dashboard/
│       ├── layout.py                   # 7-tab Dash layout
│       ├── callbacks.py                # All interactive callbacks
│       └── components.py               # Reusable UI components
└── tests/
    ├── test_analysis.py                # 30+ analysis tests
    └── test_scraper.py                 # Scraper + config tests

What Makes It Different

Most tools (Value Research, Morningstar India) show trailing returns. This shows why — portfolio construction, manager behavior, and forward-looking signals. Strong for fintech DS roles at Zerodha, Groww, ET Money, INDmoney.