feat(portfolio): Read data and calculate mean-variance for returns

Cargo.toml

@@ -34,9 +34,22 @@ rand = "0.9.2"
 rand_distr = "0.5.1"
 rayon = "1.10.0"
 statrs = "0.18.0"
+ndarray = "0.17.1"
+ndarray-stats = "0.7.0"
+csv = "=1.3.1"
 reqwest = { version = "0.12.23", features = ["json"] }
 serde = "1.0.219"
 serde-aux = "4.7.0"
+<<<<<<< HEAD
+ndarray = "0.16.1"
+ndarray-stats = "0.6.0"
+<<<<<<< HEAD
+csv = "=1.3.1"
+=======
+csv = "=1.3.0"
+>>>>>>> 39c208f (chore(deps): bump tokio from 1.47.1 to 1.48.0 (#90))
+=======
+>>>>>>> 8ae270c (fix: code review changes + allocated arrays)
 tokio = "1.48.0"
 
 [dev-dependencies]

bindings/python/README.md

@@ -7,7 +7,10 @@
 [![codecov-quantrs][codecov-badge]][codecov-url]
 ![Crates.io MSRV][crates-msrv-badge]
 ![Crates.io downloads][crates-download-badge]
+<<<<<<< HEAD
 [![PyPI version][pypi-version]][pypi-url]
+=======
+>>>>>>> 459f544 (feat(bindings): add python bindings for fixed_income (#79))
 
 [actions-test-badge]: https://github.com/carlobortolan/quantrs/actions/workflows/ci.yml/badge.svg
 [crates-badge]: https://img.shields.io/crates/v/quantrs.svg
@@ -19,10 +22,13 @@
 [codecov-url]: https://codecov.io/gh/carlobortolan/quantrs
 [crates-msrv-badge]: https://img.shields.io/crates/msrv/quantrs
 [crates-download-badge]: https://img.shields.io/crates/d/quantrs
+<<<<<<< HEAD
 [pypi-version]: https://img.shields.io/pypi/v/quantrs.svg
 [pypi-url]: https://pypi.org/project/quantrs
 <!-- [![Python versions](https://img.shields.io/pypi/pyversions/quantrs.svg)](https://pypi.org/project/quantrs/) -->
 <!-- [![PyPI downloads](https://img.shields.io/pypi/dm/quantrs.svg)](https://pypi.org/project/quantrs/) -->
+=======
+>>>>>>> 459f544 (feat(bindings): add python bindings for fixed_income (#79))
 
 Quantrs is a tiny quantitative finance library for Rust. These are the Python bindings for it.
 It is designed to be as intuitive and easy to use as possible so that you can work with derivatives without the need to write complex code or have a PhD in reading QuantLib documentation.

examples/portfolio_optimization.rs

@@ -0,0 +1,12 @@
+use quantrs::portfolio::{Portfolio, ReturnsCalculation};
+
+#[warn(unused_variables)]
+fn main() {
+    let data_path = "examples/data/asset_prices.csv";
+    let risk_free_rate = 0.01; // 1%
+    let expected_return = 0.1; // 10%
+    let returns_calc = ReturnsCalculation::Log;
+    let portfolio = Portfolio::new(data_path, risk_free_rate, expected_return, returns_calc);
+
+    println!("{}", portfolio);
+}

src/lib.rs

@@ -41,9 +41,19 @@ mod macros {
 pub mod data;
 pub mod fixed_income;
 pub mod options;
+<<<<<<< HEAD
+pub mod portfolio;
+<<<<<<< HEAD
+=======
+=======
+>>>>>>> 459f544 (feat(bindings): add python bindings for fixed_income (#79))
 
 pub use fixed_income::*;
 pub use options::*;
 
 #[cfg(feature = "python")]
 pub mod python;
+<<<<<<< HEAD
+=======
+>>>>>>> af72ef0 (feat(bindings): add python bindings for fixed_income (#79))
+>>>>>>> 459f544 (feat(bindings): add python bindings for fixed_income (#79))

src/portfolio.rs

@@ -0,0 +1,5 @@
+//! Module for Portfolio Optimization techniques
+
+pub use mean_variance::{Portfolio, ReturnsCalculation};
+
+mod mean_variance;

src/portfolio/mean_variance.rs

@@ -0,0 +1,206 @@
+//! Module implementing Mean-Variance Portfolio Optimization
+//!
+//! Current Support:
+//! - Portfolio struct to hold assets, mean returns, covariance matrix
+//! - Calculation of mean returns and covariance matrix from CSV data
+//! - Support for Simple and Log returns calculation methods
+//!
+//! Next Steps:
+//! - Implement optimization algorithms (e.g., Markowitz optimization)
+
+use csv::ReaderBuilder;
+use ndarray::{s, Array2, Axis, Zip};
+use ndarray_stats::CorrelationExt;
+use std::fmt;
+
+#[derive(Debug)]
+pub enum ReturnsCalculation {
+    Simple,
+    Log,
+}
+
+#[derive(Debug)]
+/// Struct representing a portfolio of assets.
+pub struct Portfolio {
+    /// List of asset tickers in the portfolio.
+    tickers: Vec<String>,
+    /// Mean returns of the assets.
+    mean_returns: Vec<f64>,
+    /// Covariance matrix of the asset returns.
+    covariance_matrix: Array2<f64>,
+    /// Risk-free rate for the market
+    risk_free_rate: f64,
+    /// Expected return for the portfolio
+    expected_return: f64,
+    /// Method used to calculate returns (Simple or Log)
+    returns_calculation: ReturnsCalculation,
+    /// Weights of the assets in the portfolio (if calculated)
+    /// None if not yet calculated
+    weights: Option<Vec<f64>>,
+    /// Internal storage of returns data
+    _returns: Array2<f64>,
+}
+
+impl Portfolio {
+    pub fn new(
+        data_path: &str,
+        risk_free_rate: f64,
+        expected_return: f64,
+        returns_calc: ReturnsCalculation,
+    ) -> Self {
+        // Read data from CSV file given file path
+        let mut rdr = ReaderBuilder::new()
+            .has_headers(true)
+            .from_path(data_path)
+            .expect("Failed to read CSV file");
+
+        // Extract tickers from headers
+        let tickers: Vec<String> = rdr
+            .headers()
+            .expect("Failed to read headers")
+            .iter()
+            .map(|s| s.to_string())
+            .collect();
+
+        // Read records and parse to f64, defaulting to 0.0 on parse failure
+        let records: Vec<Vec<f64>> = rdr
+            .records()
+            .map(|result| {
+                result
+                    .expect("Failed to read record")
+                    .iter()
+                    .map(|s| s.parse::<f64>().unwrap_or(0.0))
+                    .collect()
+            })
+            .collect();
+
+        let n = tickers.len();
+        let m = records.len();
+
+        // Create ndarray from records to store prices
+        let prices = Array2::from_shape_vec((m, n), records.into_iter().flatten().collect())
+            .expect("Failed to create prices array");
+
+        // Calculate returns based on specified method
+        let returns = match returns_calc {
+            ReturnsCalculation::Simple => Self::calculate_simple_returns(&prices),
+            ReturnsCalculation::Log => Self::calculate_log_returns(&prices),
+        };
+
+        // Calculate mean returns
+        let mean_returns = returns.mean_axis(Axis(0)).unwrap().to_vec();
+
+        // Calculate covariance matrix
+        // Using unbiased estimator (N-1 in denominator) for sample covariance
+        // ddof = 1.
+        // If you want population covariance, use ddof = 0
+        // Transpose of returns is used as ndarray-stats expects variables in rows
+        let covariance_matrix = returns
+            .t()
+            .cov(1.0)
+            .expect("Failed to compute covariance matrix");
+
+        Self {
+            tickers,
+            mean_returns,
+            covariance_matrix,
+            risk_free_rate,
+            expected_return,
+            returns_calculation: returns_calc,
+            weights: None,
+            _returns: returns,
+        }
+    }
+
+    /// Function to calculate log returns
+    fn calculate_log_returns(prices: &Array2<f64>) -> Array2<f64> {
+        let (n, m) = prices.dim();
+        assert!(n >= 2, "Need at least 2 rows to compute returns");
+
+        let prev = prices.slice(s![..-1, ..]);
+        let next = prices.slice(s![1.., ..]);
+
+        let mut out = Array2::<f64>::zeros((n - 1, m));
+
+        Zip::from(&mut out)
+            .and(&next)
+            .and(&prev)
+            .for_each(|o, &nxt, &prv| {
+                *o = nxt.ln() - prv.ln();
+            });
+
+        out
+    }
+
+    /// Function to calculate simple returns
+    fn calculate_simple_returns(prices: &Array2<f64>) -> Array2<f64> {
+        let (n, m) = prices.dim();
+        assert!(n >= 2, "Need at least 2 rows to compute returns");
+
+        let prev = prices.slice(s![..-1, ..]); // P_t
+        let next = prices.slice(s![1.., ..]); // P_{t+1}
+
+        let mut out = Array2::<f64>::zeros((n - 1, m));
+
+        Zip::from(&mut out)
+            .and(&next)
+            .and(&prev)
+            .for_each(|o, &nxt, &prv| {
+                *o = (nxt - prv) / prv;
+            });
+
+        out
+    }
+}
+
+/// Implement Display trait for pretty-printing the Portfolio
+impl fmt::Display for Portfolio {
+    fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result {
+        writeln!(f, "Portfolio")?;
+        writeln!(f, "Tickers: {:?}", self.tickers)?;
+        writeln!(f, "Risk-Free Rate: {:.4}", self.risk_free_rate)?;
+        writeln!(f, "Expected Return: {:.4}", self.expected_return)?;
+        writeln!(f, "Returns Calculation: {}", self.returns_calculation)?;
+        writeln!(f, "Weights: {:?}", self.weights)?;
+
+        // Print mean returns as percentages
+        writeln!(f, "\nMean Returns (%):")?;
+        for (i, &mean_return) in self.mean_returns.iter().enumerate() {
+            writeln!(f, "{:>8}: {:>8.4}%", self.tickers[i], mean_return * 100.0)?;
+        }
+
+        // Print covariance matrix in a readable format
+        writeln!(
+            f,
+            "\nCovariance Matrix ({} x {}):",
+            self.covariance_matrix.nrows(),
+            self.covariance_matrix.ncols()
+        )?;
+        // Print header with ticker names
+        write!(f, "         ")?; // spacing for row labels
+        for ticker in &self.tickers {
+            write!(f, "{:>10}", ticker)?;
+        }
+        writeln!(f)?;
+
+        // Print each row with ticker name as label
+        for (i, row) in self.covariance_matrix.outer_iter().enumerate() {
+            write!(f, "{:>8} ", self.tickers[i])?; // row label
+            for value in row {
+                write!(f, "{:>10.6}", value)?;
+            }
+            writeln!(f)?;
+        }
+        Ok(())
+    }
+}
+
+/// Implement Display trait for ReturnsCalculation enum
+impl fmt::Display for ReturnsCalculation {
+    fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result {
+        match self {
+            ReturnsCalculation::Simple => write!(f, "Simple"),
+            ReturnsCalculation::Log => write!(f, "Log"),
+        }
+    }
+}