Lumen
Quantum-enhanced portfolio optimization that shows its work.
Deterministic AI reasoning for traders who demand proof, not promises.
Explainable Optimization
Lumen uses constraint-based reasoning—not statistical approximation—to optimize portfolios. Every recommendation comes with a mathematical proof of optimality. No black boxes. No "trust the algorithm."
Edge-First Architecture
Runs on your device. Sub-second optimization for typical portfolios. Your data never leaves your machine unless you opt for quantum enhancement. Complete privacy.
Quantum When It Matters
For complex portfolios with tax-lot optimization across 50+ positions, Lumen leverages quantum computing to solve problems that would take classical systems hours. Automatic classical-quantum routing.
Provably Optimal
Every solution includes constraint satisfaction proof. You see exactly which constraints drove each recommendation. Understand the trade-offs between tax efficiency, transaction costs, and allocation drift.
Built for Precision Trading
Whether you're rebalancing a buy-and-hold portfolio or sizing positions across correlated assets, Lumen adapts to your strategy with mathematical rigor.
Portfolio Rebalancing
- Define target allocations by percentage or dollar amount
- Minimize transaction costs while reducing allocation drift
- Integer share constraints (no fractional shares where unsupported)
- Rebalancing bands with tolerance thresholds
- Cash reserve requirements and liquidity constraints
Tax-Loss Harvesting
- Identify optimal loss-harvesting opportunities across all lots
- Automatic wash sale rule compliance (30-day tracking)
- FIFO, LIFO, HIFO, or specific lot identification methods
- Short-term vs. long-term capital gains optimization
- Multi-objective: balance tax efficiency, costs, and drift
Position Sizing & Risk Management
- Correlation-aware position sizing across asset classes
- Volatility-adjusted risk allocation
- Portfolio VaR (Value at Risk) calculation
- Risk parity allocation strategies
- Forex pair correlation analysis for multi-currency portfolios
Open Source. Extensible. Apache 2.0.
Lumen is built for developers who want to extend, modify, or integrate deterministic optimization into their own systems. Full source code, comprehensive documentation, and a permissive license.
// Simple Lumen API example
#include <lumen/optimizer.hpp>
#include <lumen/constraints.hpp>
auto portfolio = lumen::Portfolio::from_csv("holdings.csv");
auto target = lumen::Allocation::percentage({
{"VTI", 0.60}, {"BND", 0.40}
});
auto optimizer = lumen::Optimizer()
.with_constraints(
lumen::MinimizeTransactionCosts(0.01),
lumen::IntegerShares(),
lumen::MinimumCashReserve(1000.00)
);
auto result = optimizer.solve(portfolio, target);
result.print_recommendations();
result.export_proof("optimization_proof.json");
Comprehensive Documentation
Quick Start Guide
Get Lumen running in 5 minutes
Architecture
Understand the core design
API Reference
Complete function documentation
CustomConstraints
Build your own optimization rules
Quantum
Connect to D-Wave, IBM, IonQ
Example Gallery
Real-world use cases & code
Extensibility First
Lumen's modular architecture makes it trivial to add new constraint types, optimization objectives, or data sources. Want to optimize for ESG scores? Add a custom constraint. Need to integrate with your broker's API? Implement the portfolio data interface. Everything is designed for extension.
Plugin Architecture
Add solvers, constraints, data sources, or output formats through a clean plugin system. No core modifications needed.
Cross-Language Bindings
C++ core with Python, JavaScript, and Rust bindings. Use Lumen from your language of choice.
Observability Built-In
Integrated PyFlare instrumentation. Monitor performance, track solver decisions, and debug optimization issues in production.
Technology Stack
Lumen leverages best-in-class open-source tools for optimization, symbolic mathematics, and quantum integration.
HiGHS Solver
High-performance open-source linear and mixed-integer programming solver. Powers classical optimization with proven reliability.
SymEngine
Fast symbolic mathematics library for constraint formulation and manipulation. Enables clean expression of complex constraints.
D-Wave Ocean SDK
Quantum annealing integration for QUBO formulations. Solves combinatorial optimization problems classical systems struggle with.
PyFlare
AI/ML observability platform for monitoring optimization performance, tracking solver decisions, and identifying bottlenecks.
IBM Qiskit / IonQ
Gate-based quantum computing integration for QAOA (Quantum Approximate Optimization Algorithm) on large-scale problems.
Cross-Platform C++
C++17/20 core runs on Windows, Linux, macOS (coming), Mobile (coming) support via shared core for iOS and Android applications.