REST vs MCP: Stock Data Fetching Comparison

This project demonstrates two different approaches to fetching stock data from Alpha Vantage:

1. REST Example: Traditional direct API integration
2. MCP Example: Model Context Protocol with plugin architecture

Purpose

This is a learning project designed to showcase the architectural differences between:

• Direct REST API calls (tightly coupled, stateless)
• MCP protocol integration (abstracted, pluggable, stateful)

Both implementations fetch the same stock quote data, making it easy to compare the approaches.

Architecture Comparison

Visual comparison: Side-by-side architecture diagrams showing REST (blue) vs MCP (green) approaches.

Aspect	REST Approach	MCP Approach
Communication	Direct HTTP GET requests	JSON-RPC over http
Abstraction	Minimal - direct endpoint calls	Protocol layer + plugin interface
State	Stateless - each request independent	Stateful session with MCP server
Coupling	Tightly coupled to Alpha Vantage API	Loosely coupled via plugin interface
Swapping providers	Requires code changes	Just swap plugin implementation
Complexity	Simple, straightforward	More setup, but more flexible
Best for	Single data source, simple needs	Multiple providers, modular architecture

Setup

1. Get an Alpha Vantage API Key

Visit Alpha Vantage and get a free API key.

2. Install Dependencies

Using uv (recommended):

uv pip install -r requirements.txt

Or with pip:

pip install -r requirements.txt

3. Set API Key

export ALPHA_VANTAGE_API_KEY=your_api_key_here

Or create a .env file (copy from .env.example):

cp .env.example .env
# Edit .env and add your API key

Usage

REST Example

uv run rest_example/main.py IBM

Or without uv:

cd rest_example
python main.py IBM

This will:

1. Make a direct HTTP GET request to Alpha Vantage API
2. Parse the JSON response
3. Display the stock quote

MCP Example

uv run mcp_example/main.py IBM

Or without uv:

cd mcp_example
python main.py IBM

This will:

1. Connect to the Alpha Vantage MCP server via http
2. Call the get_quote tool through MCP protocol
3. Display the same stock quote data

Example Output

Both examples produce similar output:

Stock Quote - REST API
==================================================
Symbol:              IBM
Price:               $178.25
Change:              +2.15 (+1.22%)
Volume:              45123456
Latest Trading Day:  2026-06-04
==================================================

Testing Different Symbols

Try various stock symbols:

uv run rest_example/main.py MSFT    # Microsoft
uv run rest_example/main.py GOOGL   # Google
uv run rest_example/main.py TSLA    # Tesla

uv run mcp_example/main.py MSFT     # Same symbols via MCP
uv run mcp_example/main.py GOOGL
uv run mcp_example/main.py TSLA

Key Differences in Code

REST Version (rest_example/rest_stock_client.py)

# Direct HTTP call
response = requests.get(
    "https://www.alphavantage.co/query",
    params={'function': 'GLOBAL_QUOTE', 'symbol': symbol, 'apikey': api_key}
)
data = response.json()

MCP Version (mcp_example/alphavantage_mcp_plugin.py)

# MCP protocol call
result = await self.session.call_tool(
    "get_quote",
    arguments={"symbol": symbol}
)

Plugin Interface (mcp_example/stock_plugin.py)

class StockDataPlugin(ABC):
    @abstractmethod
    def get_quote(self, symbol: str) -> Dict:
        pass

This interface makes it trivial to add other MCP stock servers:

• FinnhubMCPPlugin
• PolygonMCPPlugin
• YahooFinanceMCPPlugin

Just implement the StockDataPlugin interface and swap it in main.py.

Error Handling

Both examples handle common errors:

• Missing API key
• Invalid stock symbol
• Network failures
• API rate limits (free tier: 25 requests/day)

Learning Takeaways

1. REST is simpler for single-source integrations - less boilerplate, direct control
2. MCP provides abstraction that shines when you need to support multiple data sources
3. Plugin pattern (in MCP example) makes swapping providers trivial without changing client code
4. Protocol layer adds overhead but provides standardization and future flexibility
5. Both approaches can achieve the same result - choose based on your needs

Next Steps

To extend this project:

• Add more MCP plugins (Finnhub, Polygon, etc.)
• Implement historical data fetching
• Add caching layer
• Create a unified CLI that switches between REST and MCP
• Build a web UI that uses both backends

Resources

• Alpha Vantage API Documentation
• Model Context Protocol Documentation
• MCP Python SDK

License

This is a learning project - use freely for educational purposes.