QuantDSL

QuantDSL is a domain-specific language (DSL) written in OCaml and designed to enable users to perform financial modeling and simulate financial scenarios with ease.

Features

• Interactive REPL: Tests and debug input in real-time according to the following steps:

  • Lexing: Converts the input DSL scripts into tokens using lexical analysis.
  • Parsing: Converts tokens into an abstract syntax tree (AST), which represents the program structure.
  • Semantic Analysis: Validates variable declarations, types, and function arguments.
  • Type Checking: Ensures type consistency across operations.
  • Optimizing: Performs constant folding and other optimizations to improve execution efficiency.
  • Code Generation: Converts the optimized AST into bytecode instructions for execution.

• Built-in Financial Functions:

  • Historical returns simulation (historical_returns)
  • Covariance matrix calculation (covariance_matrix)
  • Mean and mean matrix computation (mean, mean_matrix)
  • Value at Risk (VaR) calculation (calculate_VaR)
  • Portfolio optimization (optimize_portfolio)
  • Monte Carlo simulation (monte_carlo_simulation)

• Customizable Scripts: Create and run scripts using a simple, intuitive DSL syntax.

Prerequisites

OCaml: Install OCaml and OPAM (OCaml Package Manager).

sudo apt update
sudo apt install ocaml opam

Dune: Ensure you have Dune installed for building the project.

opam install dune

Running the DSL

Run the following command to build the project:

dune build

This will generate the executable at:

_build/default/repl.exe

Interactive Mode

Launch the REPL to run commands interactively:

dune exec ./_build/default/repl.exe

Example REPL session:

QuantEx> let assets = ["AAPL", "GOOG", "MSFT"];
QuantEx> let returns = historical_returns(assets, {"period": "1y"});
QuantEx> let cov_matrix = covariance_matrix(returns);
QuantEx> let expected_returns = mean_matrix(returns);
QuantEx> let optimized_weights = optimize_portfolio(expected_returns, cov_matrix, {"target_return": 0.1});
QuantEx> optimized_weights;
Result: [0.4, 0.35, 0.25]

Script Mode

Run a .dsl script file directly:

dune exec ./_build/default/repl.exe -- path/to/script.dsl

Example Script: optimize_weights.dsl

let assets = ["AAPL", "GOOG", "MSFT"];
let returns = historical_returns(assets, {"period":"1y"});
let cov_matrix = covariance_matrix(returns);
let expected_returns = mean_matrix(returns);
let optimized_weights = optimize_portfolio(
    expected_returns,
    cov_matrix,
    {"target_return":0.1}
);
optimized_weights;

Run the script:

dune exec ./_build/default/repl.exe -- optimize_weights.dsl

Expected output:

Result: [0.4, 0.35, 0.25]

License

License

Copyright © 2024 [Ryan Sun] (https://github.com/Ryii).
This project is MIT licensed.