def generate_setting(self):
""""""
self.optimization_setting = OptimizationSetting()
self.target_display = self.target_combo.currentText()
target_name = self.DISPLAY_NAME_MAP[self.target_display]
self.optimization_setting.set_target(target_name)
for name, d in self.edits.items():
type_ = d["type"]
start_value = type_(d["start"].text())
step_value = type_(d["step"].text())
end_value = type_(d["end"].text())
if start_value == end_value:
self.optimization_setting.add_parameter(name, start_value)
else:
self.optimization_setting.add_parameter(
name,
start_value,
end_value,
step_value
)
self.accept()
def optimize_strategy(self, symbol: str, strategy, opt_params: List[Tuple], *, target_metric="sharpe_ratio",
alg='genetic'):
engine = BacktestingEngine()
engine.set_parameters(
vt_symbol=symbol,
interval=self._tf,
start=self._from_date,
end=self._to_date,
rate=self._fee,
slippage=self._slippage,
size=self._size,
pricetick=self._min_tick,
capital=self._capital,
)
engine.add_strategy(strategy, {})
setting = OptimizationSetting()
setting.set_target(target_metric)
for o in opt_params:
setting.add_parameter(*o)
if alg:
results = engine.run_ga_optimization(setting)
else:
results = engine.run_optimization(setting)
return results
"capital": 1_000_000, # 资金分配需要动态调整
}
strategy_cls_list = [
AtrRsiStrategy, BollChannelStrategy,
AtrStrategy]
for strategy_idx, strategy_cls in enumerate(strategy_cls_list):
df = pb.run_backtesting(
strategy_class=strategy_cls,
setting={},
vt_symbol=vt_symbol,
bk_param_dict=bk_param_dict
)
pb.show_portafolio(df)
setting = OptimizationSetting()
setting.set_target("sharpe_ratio")
op_params_list = pb.backtesting_engine.strategy.parameters
setting.set_target("sharpe_ratio")
setting.add_parameter(op_params_list[0], 3, 100, 10)
setting.add_parameter(op_params_list[1], 3, 100, 10)
op_params_result_list = pb.backtesting_engine.run_ga_optimization(setting)
# show op result
op_df = pb.run_backtesting(
strategy_class=strategy_cls,
setting=op_params_result_list[0][0],
vt_symbol=vt_symbol,
bk_param_dict=bk_param_dict
)
pb.show_portafolio(op_df)
from vnpy.app.cta_strategy.backtesting import BacktestingEngine, OptimizationSetting
from vnpy.app.cta_strategy.strategies.atr_rsi_strategy_digu import (
AtrRsiStrategyDigu, )
from datetime import datetime
engine = BacktestingEngine()
engine.set_parameters(
vt_symbol="000001_SH.SSE",
interval="d",
start=datetime(2019, 1, 1),
end=datetime(2019, 12, 31),
rate=0.3 / 10000,
slippage=0.2,
size=300,
pricetick=0.2,
capital=1_000_000,
)
engine.add_strategy(AtrRsiStrategyDigu, {})
engine.load_data()
engine.run_backtesting()
df = engine.calculate_result()
engine.calculate_statistics()
engine.show_chart()
setting = OptimizationSetting()
setting.set_target("sharpe_ratio")
setting.add_parameter("atr_length", 3, 39, 1)
setting.add_parameter("atr_ma_length", 10, 30, 1)
# engine.run_ga_optimization(setting)
result = engine.run_ga_optimization(setting)
return result
if __name__ == "__main__":
from strategies.AAA import AAA
# 回测参数设置
strategy = AAA # 参数优化策略
product = "IF" # 参数优化品种
start = date(2020, 1, 1) # 回测起始时间
end = date(2020, 6, 30) # 回测结束时间
capital = 500000 # 初始资金
# 参数优化设置
setting = OptimizationSetting()
setting.set_target("total_return")
setting.add_parameter("long_period", 3, 10, 1)
setting.add_parameter("rsi_win", 10, 20, 1)
setting.add_parameter("k_line_win", 6, 20, 1)
setting.add_parameter("donchian_win", 6, 20, 1)
setting.add_parameter("trailing_long", 0.5, 1, 0.1)
setting.add_parameter("trailing_short", 0.5, 1, 0.1)
print("开始运行参数优化")
result = run_optimization(strategy=strategy,
product=product,
setting=setting,
ga=False)
run_backtesting(
strategy_class=strategy_class,
setting={},
vt_symbol=vt_symbol,
interval=interval,
start=start,
end=end,
rate=rate,
slippage=slippage,
size=size,
pricetick=pricetick,
capital=capital,
inverse=inverse, # 正反向合约
)
setting = OptimizationSetting()
setting.set_target(f"{class_name}")
setting.add_parameter("open_window", 10, 15, 5) # 开始值，结束值，步长
setting.add_parameter("boll_length", 10, 500, 10)
setting.add_parameter("sl_multiplier", 0.3, 1, 0.02)
setting.add_parameter("dc_length", 10, 200, 5)
setting.add_parameter("prop", 1, 2, 0.2)
# setting.add_parameter("", 1, 3, 1)
# setting.add_parameter("atr_window", 1, 3, 1)
#setting.add_parameter("risk_level", 0.2, 0.2, 0.1)
# setting.add_parameter("trailing_tax", 0.7, 2.5, 0.1)
# 调用算法类型（遗传，穷举）
result = backtests(backtest, setting)
to_csv_result(
class OptimizationSettingEditor(QtWidgets.QDialog):
"""
For setting up parameters for optimization.
"""
DISPLAY_NAME_MAP = {
"总收益率": "total_return",
"夏普比率": "sharpe_ratio",
"收益回撤比": "return_drawdown_ratio",
"日均盈亏": "daily_net_pnl"
}
def __init__(
self, class_name: str, parameters: dict
):
""""""
super().__init__()
self.class_name = class_name
self.parameters = parameters
self.edits = {}
self.optimization_setting = None
self.use_ga = False
self.init_ui()
def init_ui(self):
""""""
QLabel = QtWidgets.QLabel
self.target_combo = QtWidgets.QComboBox()
self.target_combo.addItems(list(self.DISPLAY_NAME_MAP.keys()))
grid = QtWidgets.QGridLayout()
grid.addWidget(QLabel("目标"), 0, 0)
grid.addWidget(self.target_combo, 0, 1, 1, 3)
grid.addWidget(QLabel("参数"), 1, 0)
grid.addWidget(QLabel("开始"), 1, 1)
grid.addWidget(QLabel("步进"), 1, 2)
grid.addWidget(QLabel("结束"), 1, 3)
# Add vt_symbol and name edit if add new strategy
self.setWindowTitle(f"优化参数配置：{self.class_name}")
validator = QtGui.QDoubleValidator()
row = 2
for name, value in self.parameters.items():
type_ = type(value)
if type_ not in [int, float]:
continue
start_edit = QtWidgets.QLineEdit(str(value))
step_edit = QtWidgets.QLineEdit(str(1))
end_edit = QtWidgets.QLineEdit(str(value))
for edit in [start_edit, step_edit, end_edit]:
edit.setValidator(validator)
grid.addWidget(QLabel(name), row, 0)
grid.addWidget(start_edit, row, 1)
grid.addWidget(step_edit, row, 2)
grid.addWidget(end_edit, row, 3)
self.edits[name] = {
"type": type_,
"start": start_edit,
"step": step_edit,
"end": end_edit
}
row += 1
parallel_button = QtWidgets.QPushButton("多进程优化")
parallel_button.clicked.connect(self.generate_parallel_setting)
grid.addWidget(parallel_button, row, 0, 1, 4)
row += 1
ga_button = QtWidgets.QPushButton("遗传算法优化")
ga_button.clicked.connect(self.generate_ga_setting)
grid.addWidget(ga_button, row, 0, 1, 4)
self.setLayout(grid)
def generate_ga_setting(self):
""""""
self.use_ga = True
self.generate_setting()
def generate_parallel_setting(self):
""""""
self.use_ga = False
self.generate_setting()
def generate_setting(self):
""""""
self.optimization_setting = OptimizationSetting()
self.target_display = self.target_combo.currentText()
target_name = self.DISPLAY_NAME_MAP[self.target_display]
self.optimization_setting.set_target(target_name)
for name, d in self.edits.items():
type_ = d["type"]
start_value = type_(d["start"].text())
step_value = type_(d["step"].text())
end_value = type_(d["end"].text())
if start_value == end_value:
self.optimization_setting.add_parameter(name, start_value)
else:
self.optimization_setting.add_parameter(
name,
start_value,
end_value,
step_value
)
self.accept()
def get_setting(self):
""""""
return self.optimization_setting, self.use_ga
inverse=inverse, # 正反向合约
)
"""
open_window = 15
boll_length = 80
boll_dev = 2.0
bias = 2.0
cci_length = 6
cci_exit = 10.0
atr_window = 16
atr_multiple = 2.0
long_trailing = 4.0
short_trailing = 4.0
"""
setting = OptimizationSetting()
setting.set_target(f"{class_name}")
# setting.add_parameter("open_window", 15, 60, 15)
# setting.add_parameter("rsi_window", 10, 150, 10)
# setting.add_parameter("exit_window", 30, 80, 5)
# setting.add_parameter("ema_window", 30, 80, 5)
# setting.add_parameter("pos_trailing", 2, 3, 0.2)
# setting.add_parameter("open_window", 15, 15, 15)
# setting.add_parameter("fast_length", 20, 140, 5)
# setting.add_parameter("slow_length", 20, 300, 20)
# setting.add_parameter("open_window", 30, 30, 30)
# setting.add_parameter("boll_length", 100, 600, 20)
# setting.add_parameter("boll_dev", 1.4, 3, 0.1)
# setting.add_parameter("bias", 0.1, 0.5, 0.03)
cost_multiple = 2.0
interval = 'd'
keep_last_open = True
note_str = 'double_ma_exit_atr_rein_daily'
commodity_list = [
"cu", "al", "zn", "pb", "ni", "sn", "au", "ag", "rb", "hc", "bu", "ru",
"sp", "m", "y", "a", "b", "p", "c", "cs", "jd", "l", "v", "pp", "j",
"jm", "i", "SR", "CF", "ZC", "FG", "TA", "MA", "OI", "RM", "SF", "SM"
]
# for test
# commodity_list = ["cu", "al", "zn"]
# commodity_list = ["sp"]
turtle_gen = OptimizationSetting()
turtle_gen.add_parameter("fast_window", 5)
turtle_gen.add_parameter("slow_window", 50)
turtle_gen.add_parameter("atr_multi", 1.0, 2.5, 0.5)
turtle_gen.add_parameter("exit_atr_multi", 2, 4, 1)
turtle_gen.add_parameter("max_rein", 1, 3, 1)
# turtle_gen.add_parameter("entry_window", 5, 10, 5)
# turtle_gen.add_parameter("exit_window", 20)
# turtle_gen.add_parameter("sl_multiplier", 3.5)
turtle_settings = turtle_gen.generate_setting()
# load custom setting
# df = pd.read_csv('custom_turtle_settings.csv')
# series = dict(df.iterrows()).values()
run_backtesting(
strategy_class=strategy_class,
setting={},
vt_symbol=vt_symbol,
interval=interval,
start=start,
end=end,
rate=rate,
slippage=slippage,
size=size,
pricetick=pricetick,
capital=capital,
inverse=inverse, # 正反向合约
)
# 优化函数
setting = OptimizationSetting()
# 优化指示名称 如果夏普率
setting.set_target(f"{class_name}")
setting.add_parameter("open_window", 15, 15, 5) # 开始值，结束值，步长
setting.add_parameter("boll_length", 60, 100, 2)
setting.add_parameter("sl_multiplier", 0.6, 0.8, 0.01)
setting.add_parameter("dc_length", 70, 100, 5)
setting.add_parameter("prop", 1.8, 1.8, 0.2)
# 多进程穷举优化
result = engine.run_optimization(setting)
to_csv_result(
result=result,
signal_name=str(strategy_class.__qualname__), # 策略名称
run_backtesting(
strategy_class=strategy_class,
setting={},
vt_symbol=vt_symbol,
interval=interval,
start=start,
end=end,
rate=rate,
slippage=slippage,
size=size,
pricetick=pricetick,
capital=capital,
inverse=inverse, # 正反向合约
)
setting = OptimizationSetting()
setting.set_target(f"{class_name}")
setting.add_parameter("open_window", 5, 5, 1)
setting.add_parameter("nloss_singnal", 2.5, 4, 0.1)
setting.add_parameter("atrstop_window", 40, 60, 5)
# setting.add_parameter("", 16, 32, 2)
# setting.add_parameter("", 7, 11, 1)
# setting.add_parameter("", 1, 3, 1)
# setting.add_parameter("atr_window", 1, 3, 1)
#setting.add_parameter("risk_level", 0.2, 0.2, 0.1)
# setting.add_parameter("trailing_tax", 0.7, 2.5, 0.1)
result = engine.run_ga_optimization(setting) #遗传算法
to_csv_result(
result=result,
engine = BacktestingEngine()
engine.set_parameters(
vt_symbol="510050.SSE",
interval="1m",
start=start,
end=datetime(2020, 4, 1),
rate=1/10000,
slippage=0.002,
size=10000,
pricetick=0.001,
capital=1_000_000,
)
engine.add_strategy(test_strategy_cls, {})
setting = OptimizationSetting()
setting.set_target("sharpe_ratio")
# setting.add_parameter("hedge_range_param", 10, 200, 10)
# setting.add_parameter("hedge_multiple_param", 40, 100, 20)
# setting.add_parameter("hedge_size", 10)
setting.add_parameter("hedge_range_percent", 20)
setting.add_parameter("hedge_multiple_percent", 30, 100, 10)
# setting.add_parameter("atr_range", 5, 30, 5)
# setting.add_parameter("hedge_multiple_percent", 40, 100, 20)
results = engine.run_optimization(setting)
start_str = dt_to_str(start)
engine = BacktestingEngine()
engine.set_parameters(
vt_symbol="AG2005.SHFE",
interval="1m",
start=datetime(2010, 1, 1),
end=datetime.now(),
rate=2.25 / 10000,
slippage=1,
size=10,
pricetick=1,
capital=30000,
)
engine.add_strategy(CincoP2005Strategy, {})
engine.load_data()
engine.run_backtesting()
df = engine.calculate_result()
engine.calculate_statistics()
setting = OptimizationSetting()
setting.set_target("sharpe_ratio")
setting.add_parameter("boll_window", 16, 30, 2)
setting.add_parameter("boll_dev", 1.2, 3.2, 0.1)
result_values = engine.run_ga_optimization(setting)
print(result_values)