2. How to Run Strategy Optimization

2.1. Entry Point and Common kwargs

Run strategy optimization with qt.run(op, mode=2, ...). Common arguments include opti_method, opti_sample_count, opti_output_count, and others.

2.2. Full List of Optimization Algorithms (with Brief Explanations)

The following are common opti_method values and typical use cases; refer to the qteasy 2.0 API as the source of truth.

Algorithm	Meaning	Typical Use Case
grid	Grid search	Few parameters, discrete space; exhaustive or coarse grid search.
montecarlo	Monte Carlo random sampling	Random sampling when there are many parameters.
GA	Genetic algorithm	Medium-to-high dimensions; continuous/discrete mixed spaces.
SA	Simulated annealing	Medium dimensions; helpful when local optima are a concern.
PSO	Particle swarm optimization	Continuous space with multiple peaks.
bayesian	Bayesian optimization	Expensive evaluations; need good results in few steps.

2.3. Full List of Optimization Run Parameters (with Brief Explanations)

Parameter	Type / Values	Meaning
opti_method	str	Optimization algorithm: grid, montecarlo, GA, SA, PSO, bayesian, etc.
opti_sample_count	int	Number of samples/iterations (e.g., Monte Carlo samples or GA generations).
opti_output_count	int	Number of best results to keep (Top-K).
Objective function	—	E.g., Sharpe ratio or drawdown; see configuration or API docs.
Constraints	—	If constraints apply (e.g., max drawdown limit), see configuration.

Backtest date range, asset_pool, and other arguments shared with mode=1 also apply in mode=2.

2.4. Parameter Space

The strategy’s Parameter objects and par_range define the search space; the framework samples or searches within it according to opti_method.

2.5. Minimal Runnable Example

import qteasy as qt

op = qt.Operator(strategies='dma', signal_type='PT', run_freq='d')
# dma 策略带 short_period、long_period 等可调参数
qt.configure(asset_pool='000001.SZ', invest_start='2020-01-01', invest_end='2023-12-31')
result = qt.run(op, mode=2, opti_method='grid', opti_sample_count=100, opti_output_count=10)

2.6. Configuration Highlights

• Backtest range and asset pool: Same as mode=1; they define the backtest environment for each parameter evaluation.

• Objective function: Defines what “better” means (e.g., maximize Sharpe or minimize drawdown) and affects ranking and final selection.