Genetic evolution on GPU: thousands of strategy variants compete and mutate to find optimal parameters.
Load this exact public contract, mutate it locally and settle the duel on the same unseen holdout.
CUDA Evolved is the natural evolution of CUDA GPU, integrating a multi-objective evolutionary algorithm that simultaneously optimizes return, Sharpe ratio, maximum drawdown and win rate. Where CUDA GPU optimizes a single objective, CUDA Evolved finds the optimal Pareto front — the set of solutions where improving one criterion necessarily degrades another. The result is a more robust and balanced strategy, less susceptible to over-optimization.
Uses NSGA-II (Non-dominated Sorting Genetic Algorithm II) on NVIDIA GPU. Evaluates each solution on 4 simultaneous objectives: PnL, Sharpe, Max DrawDown, Win Rate. Selects non-dominated solutions (Pareto front). The trader chooses the optimal profile based on risk tolerance. Full recalculation every 2 hours.
Multi-objective evolutionary signals. Self-calibrated technical indicators (RSI, MACD, Bollinger, Stochastic). Dynamic signal allocation based on detected market regime. Adaptive ATR for money management.
High
Multi-objective optimization more robust than single-objective. Pareto front avoids unbalanced solutions. Better overall return/risk ratio. Continuous evolution without human intervention.
Even higher algorithmic complexity than CUDA GPU. Risk of premature convergence of evolutionary algorithm. Significant GPU computation time (NSGA-II is expensive). Difficult to understand why a solution was selected.
Explore all 74 trading strategies across 4 arenas
🏟️ View all strategies