Optional support for defining hyperopt parameters in a strategy file and reusing common hyperopt/strategy parts.

freqtrade/optimize/hyperopt_auto.py

@@ -0,0 +1,83 @@
+"""
+HyperOptAuto class.
+This module implements a convenience auto-hyperopt class, which can be used together with strategies that implement
+IHyperStrategy interface.
+"""
+from typing import Any, Callable, Dict, List
+from pandas import DataFrame
+from skopt.space import Categorical, Dimension, Integer, Real  # noqa
+
+from freqtrade.optimize.hyperopt_interface import IHyperOpt
+
+
+# noinspection PyUnresolvedReferences
+class HyperOptAuto(IHyperOpt):
+    """
+    This class delegates functionality to Strategy(IHyperStrategy) and Strategy.HyperOpt classes. Most of the time
+    Strategy.HyperOpt class would only implement indicator_space and sell_indicator_space methods, but other hyperopt
+    methods can be overridden as well.
+    """
+    def buy_strategy_generator(self, params: Dict[str, Any]) -> Callable:
+        assert hasattr(self.strategy, 'enumerate_parameters'), 'Strategy must inherit from IHyperStrategy.'
+
+        def populate_buy_trend(dataframe: DataFrame, metadata: dict):
+            for attr_name, attr in self.strategy.enumerate_parameters('buy'):
+                attr.value = params[attr_name]
+            return self.strategy.populate_buy_trend(dataframe, metadata)
+        return populate_buy_trend
+
+    def sell_strategy_generator(self, params: Dict[str, Any]) -> Callable:
+        assert hasattr(self.strategy, 'enumerate_parameters'), 'Strategy must inherit from IHyperStrategy.'
+
+        def populate_buy_trend(dataframe: DataFrame, metadata: dict):
+            for attr_name, attr in self.strategy.enumerate_parameters('sell'):
+                attr.value = params[attr_name]
+            return self.strategy.populate_sell_trend(dataframe, metadata)
+        return populate_buy_trend
+
+    def _get_func(self, name) -> Callable:
+        """
+        Return a function defined in Strategy.HyperOpt class, or one defined in super() class.
+        :param name: function name.
+        :return: a requested function.
+        """
+        hyperopt_cls = getattr(self.strategy, 'HyperOpt')
+        default_func = getattr(super(), name)
+        if hyperopt_cls:
+            return getattr(hyperopt_cls, name, default_func)
+        else:
+            return default_func
+
+    def _generate_indicator_space(self, category):
+        assert hasattr(self.strategy, 'enumerate_parameters'), 'Strategy must inherit from IHyperStrategy.'
+
+        for attr_name, attr in self.strategy.enumerate_parameters(category):
+            yield attr.get_space(attr_name)
+
+    def _get_indicator_space(self, category, fallback_method_name):
+        indicator_space = list(self._generate_indicator_space(category))
+        if len(indicator_space) > 0:
+            return indicator_space
+        else:
+            return self._get_func(fallback_method_name)()
+
+    def indicator_space(self) -> List[Dimension]:
+        return self._get_indicator_space('buy', 'indicator_space')
+
+    def sell_indicator_space(self) -> List[Dimension]:
+        return self._get_indicator_space('sell', 'sell_indicator_space')
+
+    def generate_roi_table(self, params: Dict) -> Dict[int, float]:
+        return self._get_func('generate_roi_table')(params)
+
+    def roi_space(self) -> List[Dimension]:
+        return self._get_func('roi_space')()
+
+    def stoploss_space(self) -> List[Dimension]:
+        return self._get_func('stoploss_space')()
+
+    def generate_trailing_params(self, params: Dict) -> Dict:
+        return self._get_func('generate_trailing_params')(params)
+
+    def trailing_space(self) -> List[Dimension]:
+        return self._get_func('trailing_space')()