A high-performance MCP (Message Control Protocol) server for evaluating financial formulas and indicators.

• Extensible Architecture: Easily add new financial indicators by implementing the Formula interface.
• High Performance: Built with Go for optimal performance and concurrency.
• RESTful API: Simple HTTP/JSON interface for easy integration.
• Modular Design: Clear separation of concerns with well-defined components.

• Go 1.16 or higher
• Git

1. Clone the repository:

   git clone https://github.com/yourusername/quantmcp-server.git
   cd quantmcp-server

2. Build the server:

   go build -o quantmcp ./cmd/server.go

3. Run the server:

   ./quantmcp

The server will start on http://localhost:8080.

The server includes several performance optimizations:

• L1 Cache: Fast in-memory cache for hot items with lock-free reads
• L2 Cache: Sharded cache for larger storage to reduce contention
• Automatic cleanup of expired entries
• Metrics collection for cache performance monitoring

• Efficient reuse of database connections
• Configurable pool size based on workload

• Batch processing of multiple formula evaluations
• Reduced overhead for high-throughput scenarios

• Object pooling for formula instances
• Efficient memory allocation patterns
• Automatic garbage collection tuning

The server exposes Prometheus metrics at /metrics endpoint:

• Cache Metrics: Hits, misses, size, and entry count
• HTTP Metrics: Request counts, latencies, and response sizes
• System Metrics: Memory usage, goroutines, and GC stats

Example metrics:

# HELP http_requests_total Total number of HTTP requests
# TYPE http_requests_total counter
http_requests_total{method="POST",path="/evaluate",status="200"} 42

# HELP cache_hits_total Total number of cache hits
# TYPE cache_hits_total counter
cache_hits_total 1234

# HELP cache_misses_total Total number of cache misses
# TYPE cache_misses_total counter
cache_misses_total 56

• POST /evaluate - Evaluate a single formula (legacy)
• POST /v2/evaluate - Optimized formula evaluation with caching
• POST /v2/evaluate/batch - Batch process multiple formulas
• GET /metrics - Prometheus metrics endpoint
• GET /health - Health check endpoint
• GET /formulas - List available formulas
• GET /version - Server version information

POST /evaluate
Content-Type: application/json

{
  "command": "compute",
  "instrument": "MACD",
  "params": {
    "fast": 12,
    "slow": 26,
    "signal": 9
  },
  "inputs": {
    "close": [1.23, 1.25, 1.28, 1.30, ...]
  }
}

{
  "status": "success",
  "instrument": "MACD",
  "result": {
    "macd": [0.0023, 0.0031, 0.0027, ...],
    "signal": [0.0025, 0.0029, 0.0028, ...],
    "histogram": [-0.0002, 0.0002, -0.0001, ...]
  }
}

1. Create a new file in the internal/formulas directory.
2. Implement the Formula interface.
3. Register the formula in an init() function.

Example:

package formulas

import "github.com/quantmcp/server/internal/formulas"

func init() {
    formulas.RegisterFormula(NewYourFormula())
}

Run the test suite:

go test ./...

Benchmark tests are available to measure performance characteristics:

# Run all benchmarks
go test -bench=. -benchmem ./...

# Run benchmarks with memory profiling
go test -bench=. -benchmem -memprofile=mem.out ./...

Example using wrk:

wrk -t4 -c100 -d30s -s scripts/loadtest.lua http://localhost:8080/evaluate

When running the server in test mode, metrics are available at http://localhost:8080/metrics for monitoring performance during testing.

This project is licensed under the MIT License - see the LICENSE file for details.