Goal-Based SIP Investment Predictor using Financial Modeling + Machine Learning

This project helps individuals plan their monthly SIP (Systematic Investment Plan) contributions based on:

• Their financial goal (target amount),
• Time horizon (in years), and
• Risk appetite (Conservative, Balanced, Aggressive)

It uses real historical data from 2007–2021 of asset classes:

• Bitcoin
• Gold
• Nifty50 (Indian equity index)
• Fixed Deposit (FD) (assumed stable return)

Once we calculate how these asset classes behave, we simulate portfolio growth and generate synthetic data to train a Machine Learning model — helping anyone figure out how much they should invest monthly to reach their goal.

---

Project Workflow (Step-by-Step)

---

1. Data Collection & Cleaning

• Merged overlapping historical CSVs for Bitcoin (BTC-INR) from Yahoo Finance.

• Aligned monthly closing prices for:

  • Bitcoin
  • Gold
  • Nifty50

• Fixed Deposit data was assumed to have a constant return of 6% per annum.

Result: A clean dataset of monthly prices (2007–2021)

---

2. Yearly Return Calculation

We resampled monthly data to yearly, using the last closing price of each year, and computed annual returns:

**Annual Return (year)** = (Price_end - Price_start) / Price_start

This gave us a dataframe of yearly % returns for each asset.

---

3. CAGR: Compound Annual Growth Rate

We computed the average annual growth assuming compounding — more realistic than average returns.

$$ \text{CAGR} = \left( \frac{\text{Final Value}}{\text{Initial Value}} \right)^{\frac{1}{\text{Years}}} - 1 $$

Result:

• Bitcoin CAGR: ~170%
• Gold CAGR: ~10%
• Nifty50 CAGR: ~6.3%

---

4. Volatility (Standard Deviation of Returns)

Measures the risk or fluctuation in returns. Higher volatility = more uncertainty.

$$ \sigma = \sqrt{ \frac{1}{N-1} \sum_{i=1}^N (R_i - \bar{R})^2 } $$

---

5. Sharpe Ratio

Measures risk-adjusted return — how much excess return you're getting per unit of risk.

$$ \text{Sharpe Ratio} = \frac{\text{Return} - \text{Risk-Free Rate}}{\text{Volatility}} $$

Assuming risk-free rate = 4%.

---

6. Risk Profiles (Asset Allocations)

We defined 3 sample profiles:

risk_profiles = {
    'Conservative': {'FD': 0.60, 'Gold': 0.30, 'Nifty50': 0.10, 'Bitcoin': 0.00},
    'Balanced':     {'FD': 0.30, 'Gold': 0.30, 'Nifty50': 0.40, 'Bitcoin': 0.00},
    'Aggressive':   {'FD': 0.00, 'Gold': 0.20, 'Nifty50': 0.60, 'Bitcoin': 0.20}
}

---

7. Portfolio Return & Volatility (Realistic Modeling)

For each risk profile, we calculate:

Portfolio Return:

Weighted average of asset CAGRs:

$$ \text{Portfolio Return} = \sum_{i} w_i \cdot \text{CAGR}_i $$

Where:

• $w_i$: weight of asset $i$
• $\text{CAGR}_i$: CAGR of asset $i$

Portfolio Volatility:

We don't just average volatilities — instead, we use the covariance matrix of asset returns and calculate true portfolio volatility:

$$ \sigma_p = \sqrt{ \mathbf{w}^T \cdot \Sigma \cdot \mathbf{w} } $$

Where:

• $\mathbf{w}$: vector of weights
• $\Sigma$: covariance matrix of asset returns

We excluded FD from the covariance matrix since it has 0 volatility — its impact is scaled down appropriately.

---

8. Monte Carlo Simulation (Investment Growth)

To simulate realistic portfolio growth, we model monthly SIP investments over a given time horizon using random monthly returns:

Process:

• For each month, add monthly SIP

• Apply a random return drawn from a normal distribution:

  R_month ~ N(r / 12, σ / √12)
  

  Where:

  • r is the annual expected return (CAGR)
  • σ is the annual volatility
  • R_month is the simulated monthly return

• Repeat for all months and for 1000 simulations

• Average the final portfolio values from all simulations

This gives us the expected goal amount for a given SIP, duration, and risk profile.

---

9. Generating Synthetic Data for ML

We flipped the simulation logic to generate data for machine learning:

Inputs:

• Monthly SIP amount (₹1000 to ₹50000)
• Investment duration (1–20 years)
• Risk profile (Conservative, Balanced, Aggressive)

For each combination:

• Simulate investment growth via Monte Carlo
• Record the goal amount reached

Outputs:

• $X =$ [Goal Amount, Duration, Risk Profile (encoded)]
• $y =$ Monthly SIP

We generated 480 synthetic data points to train a regression model.

---

10. Machine Learning Model

We trained a RandomForestRegressor model to predict the required monthly SIP, given:

• Desired goal amount
• Duration (years)
• Risk profile

Evaluation:

• MSE: ~12.9 lakhs²

• R² Score: ~0.95 → Good fit

• Feature importance:

  • Goal Amount: ~64%
  • Duration: ~30%
  • Risk Profile: ~6%

This lets users answer:

"How much should I invest monthly to reach ₹X in Y years with Z risk tolerance?"

Model Accuracy:

The model achieved an R² Score of 0.948, indicating that ~95% of the variance in SIP prediction is explained by the inputs.

This level of accuracy is excellent for financial estimation models, especially when working with real market volatility.

---

11. Gradio UI

We built a clean, interactive web UI using Gradio where users can:

• Enter their goal, duration, and risk profile
• View predicted SIP amount
• See an investment growth chart
• Visualize their asset allocation pie chart
• Read an investment summary (CAGR, total return, volatility)

---

Final Outputs

• A trained ML model that predicts SIP for any user input
• Financial engine using real market data + Monte Carlo simulation
• Risk-adjusted metrics: CAGR, Volatility, Sharpe Ratio
• Gradio-powered interactive investment planner

---

Glossary of Financial Terms

Term	Meaning
SIP	Systematic Investment Plan – monthly investing strategy
CAGR	Compound Annual Growth Rate – smoothed annual return
Volatility	Standard deviation of returns – measures investment risk
Sharpe Ratio	Return per unit of risk (vs. a risk-free asset like FD)
Covariance	Measures how two assets move together – used for portfolio optimization
Monte Carlo	Simulation technique using random sampling to model uncertain outcomes

---

Why This Project Matters

• Most SIP calculators hardcode returns and ignore real volatility.
• We use historical data + statistical modeling to provide realistic, personalized estimates.
• This builds financial literacy + planning confidence in investors, even if they're new.

---