Smarter than manual tuning
Stop guessing lookback windows and multipliers. Bayesian optimization systematically explores your parameter space and converges on the best configuration.
Optimize Nautilus strategieswithout managing infrastructure
Build a Docker image of your NautilusTrader backtest, configure parameter ranges in the dashboard, and let Bayesian optimization find the best settings.
Why HyperOptimizer for NautilusTrader?
High-performance optimization for high-performance trading
Parallel backtests
Run multiple Nautilus backtests simultaneously. Each trial gets its own container; no contention, no shared state. Results come in faster.
Your strategy stays private
Your Nautilus strategy runs inside an isolated container. We never see your source code, trading logic, or historical data. Only the metric lines you print.
Clear results
Leaderboard ranked by Sharpe, PnL, or any custom metric. Convergence plots and Pareto frontiers for multi-objective experiments like return vs. drawdown.
How it works
From backtest to optimized in three steps
1
Dockerize your backtest
Package your Nautilus engine, strategy, data, and dependencies in a Docker image. Set the default command to your backtest script.
2
Parse args & emit metrics
Read --hpo-* flags, run engine.run(), and print metrics from engine.get_result().
3
Read the leaderboard
We run parallel trials, use Bayesian optimization to explore the parameter space, and surface the best strategy configurations in the dashboard.
Example
Emit backtest results as metrics
After engine.run()
result = engine.get_result()
metrics = {
"total_pnl": float(result.stats_pnls.get("PnL", 0)),
"total_orders": result.total_orders,
"elapsed_time": result.elapsed_time,
}
for key, value in metrics.items():
print(f"hpo.metrics.{key}={json.dumps(value, default=str)}")More integrations
Other ways to use HyperOptimizer
Optimize your Nautilus strategy
Follow the NautilusTrader integration guide for a step-by-step walkthrough with full code examples.