185 lines
7.4 KiB
Python
185 lines
7.4 KiB
Python
|
"""
|
||
|
|
EMA Crossover Strategy Implementation
|
||
|
|
|
||
|
|
This module implements an Exponential Moving Average (EMA) Crossover trading strategy.
|
||
|
|
It extends the BaseStrategy and generates buy/sell signals based on the crossover
|
||
|
|
of a fast EMA and a slow EMA.
|
||
|
|
"""
|
||
|
|
|
||
|
|
import pandas as pd
|
||
|
|
from typing import List, Dict, Any
|
||
|
|
|
||
|
|
from ..base import BaseStrategy
|
||
|
|
from ..data_types import StrategyResult, StrategySignal, SignalType
|
||
|
|
from ..utils import create_indicator_key, detect_crossover_signals_vectorized
|
||
|
|
|
||
|
|
|
||
|
|
class EMAStrategy(BaseStrategy):
|
||
|
|
"""
|
||
|
|
EMA Crossover Strategy.
|
||
|
|
|
||
|
|
Generates buy/sell signals when a fast EMA crosses above or below a slow EMA.
|
||
|
|
"""
|
||
|
|
|
||
|
|
def __init__(self, logger=None):
|
||
|
|
super().__init__(logger)
|
||
|
|
self.strategy_name = "ema_crossover"
|
||
|
|
|
||
|
|
def get_required_indicators(self) -> List[Dict[str, Any]]:
|
||
|
|
"""
|
||
|
|
Defines the indicators required by the EMA Crossover strategy.
|
||
|
|
It needs two EMA indicators: a fast one and a slow one.
|
||
|
|
"""
|
||
|
|
# Default periods for EMA crossover, can be overridden by strategy config
|
||
|
|
return [
|
||
|
|
{'type': 'ema', 'period': 12, 'price_column': 'close'},
|
||
|
|
{'type': 'ema', 'period': 26, 'price_column': 'close'}
|
||
|
|
]
|
||
|
|
|
||
|
|
def calculate(self, df: pd.DataFrame, indicators_data: Dict[str, pd.DataFrame], **kwargs) -> List[StrategyResult]:
|
||
|
|
"""
|
||
|
|
Calculate EMA Crossover strategy signals.
|
||
|
|
|
||
|
|
Args:
|
||
|
|
df: DataFrame with OHLCV data.
|
||
|
|
indicators_data: Dictionary of pre-calculated indicator DataFrames.
|
||
|
|
Expected keys: 'ema_period_12', 'ema_period_26'.
|
||
|
|
**kwargs: Additional strategy parameters (e.g., fast_period, slow_period, price_column).
|
||
|
|
|
||
|
|
Returns:
|
||
|
|
List of StrategyResult objects, each containing generated signals.
|
||
|
|
"""
|
||
|
|
# Extract EMA periods from kwargs or use defaults
|
||
|
|
fast_period = kwargs.get('fast_period', 12)
|
||
|
|
slow_period = kwargs.get('slow_period', 26)
|
||
|
|
price_column = kwargs.get('price_column', 'close')
|
||
|
|
|
||
|
|
# Generate indicator keys using shared utility function
|
||
|
|
fast_ema_key = create_indicator_key({'type': 'ema', 'period': fast_period})
|
||
|
|
slow_ema_key = create_indicator_key({'type': 'ema', 'period': slow_period})
|
||
|
|
|
||
|
|
# Validate that the main DataFrame has enough data for strategy calculation (not just indicators)
|
||
|
|
if not self.validate_dataframe(df, max(fast_period, slow_period)):
|
||
|
|
if self.logger:
|
||
|
|
self.logger.warning(f"{self.strategy_name}: Insufficient main DataFrame for calculation.")
|
||
|
|
return []
|
||
|
|
|
||
|
|
# Validate that the required indicators are present and have sufficient data
|
||
|
|
required_indicators = [
|
||
|
|
{'type': 'ema', 'period': fast_period},
|
||
|
|
{'type': 'ema', 'period': slow_period}
|
||
|
|
]
|
||
|
|
if not self.validate_indicators_data(indicators_data, required_indicators):
|
||
|
|
if self.logger:
|
||
|
|
self.logger.warning(f"{self.strategy_name}: Missing or insufficient indicator data.")
|
||
|
|
return []
|
||
|
|
|
||
|
|
fast_ema_df = indicators_data.get(fast_ema_key)
|
||
|
|
slow_ema_df = indicators_data.get(slow_ema_key)
|
||
|
|
|
||
|
|
if fast_ema_df is None or slow_ema_df is None or fast_ema_df.empty or slow_ema_df.empty:
|
||
|
|
if self.logger:
|
||
|
|
self.logger.warning(f"{self.strategy_name}: EMA indicator DataFrames are not found or empty.")
|
||
|
|
return []
|
||
|
|
|
||
|
|
# Merge all necessary data into a single DataFrame for easier processing
|
||
|
|
# Ensure alignment by index (timestamp)
|
||
|
|
merged_df = pd.merge(df[[price_column, 'symbol', 'timeframe']],
|
||
|
|
fast_ema_df[['ema']],
|
||
|
|
left_index=True, right_index=True, how='inner',
|
||
|
|
suffixes= ('', '_fast'))
|
||
|
|
merged_df = pd.merge(merged_df,
|
||
|
|
slow_ema_df[['ema']],
|
||
|
|
left_index=True, right_index=True, how='inner',
|
||
|
|
suffixes= ('', '_slow'))
|
||
|
|
|
||
|
|
# Rename columns to their logical names after merge
|
||
|
|
merged_df.rename(columns={'ema': 'ema_fast', 'ema_slow': 'ema_slow'}, inplace=True)
|
||
|
|
|
||
|
|
if merged_df.empty:
|
||
|
|
if self.logger:
|
||
|
|
self.logger.warning(f"{self.strategy_name}: Merged DataFrame is empty after indicator alignment. Check data ranges.")
|
||
|
|
return []
|
||
|
|
|
||
|
|
# Use vectorized signal detection for better performance
|
||
|
|
bullish_crossover, bearish_crossover = detect_crossover_signals_vectorized(
|
||
|
|
merged_df, 'ema_fast', 'ema_slow'
|
||
|
|
)
|
||
|
|
|
||
|
|
results: List[StrategyResult] = []
|
||
|
|
strategy_metadata = {
|
||
|
|
'fast_period': fast_period,
|
||
|
|
'slow_period': slow_period
|
||
|
|
}
|
||
|
|
|
||
|
|
# Process bullish crossover signals
|
||
|
|
bullish_indices = merged_df[bullish_crossover].index
|
||
|
|
for timestamp in bullish_indices:
|
||
|
|
row = merged_df.loc[timestamp]
|
||
|
|
|
||
|
|
# Skip if any EMA values are NaN
|
||
|
|
if pd.isna(row['ema_fast']) or pd.isna(row['ema_slow']):
|
||
|
|
continue
|
||
|
|
|
||
|
|
signal = StrategySignal(
|
||
|
|
timestamp=timestamp,
|
||
|
|
symbol=row['symbol'],
|
||
|
|
timeframe=row['timeframe'],
|
||
|
|
signal_type=SignalType.BUY,
|
||
|
|
price=float(row[price_column]),
|
||
|
|
confidence=0.8,
|
||
|
|
metadata={'crossover_type': 'bullish', **strategy_metadata}
|
||
|
|
)
|
||
|
|
|
||
|
|
results.append(StrategyResult(
|
||
|
|
timestamp=timestamp,
|
||
|
|
symbol=row['symbol'],
|
||
|
|
timeframe=row['timeframe'],
|
||
|
|
strategy_name=self.strategy_name,
|
||
|
|
signals=[signal],
|
||
|
|
indicators_used={
|
||
|
|
'ema_fast': float(row['ema_fast']),
|
||
|
|
'ema_slow': float(row['ema_slow'])
|
||
|
|
},
|
||
|
|
metadata=strategy_metadata
|
||
|
|
))
|
||
|
|
|
||
|
|
if self.logger:
|
||
|
|
self.logger.info(f"{self.strategy_name}: BUY signal at {timestamp} for {row['symbol']}")
|
||
|
|
|
||
|
|
# Process bearish crossover signals
|
||
|
|
bearish_indices = merged_df[bearish_crossover].index
|
||
|
|
for timestamp in bearish_indices:
|
||
|
|
row = merged_df.loc[timestamp]
|
||
|
|
|
||
|
|
# Skip if any EMA values are NaN
|
||
|
|
if pd.isna(row['ema_fast']) or pd.isna(row['ema_slow']):
|
||
|
|
continue
|
||
|
|
|
||
|
|
signal = StrategySignal(
|
||
|
|
timestamp=timestamp,
|
||
|
|
symbol=row['symbol'],
|
||
|
|
timeframe=row['timeframe'],
|
||
|
|
signal_type=SignalType.SELL,
|
||
|
|
price=float(row[price_column]),
|
||
|
|
confidence=0.8,
|
||
|
|
metadata={'crossover_type': 'bearish', **strategy_metadata}
|
||
|
|
)
|
||
|
|
|
||
|
|
results.append(StrategyResult(
|
||
|
|
timestamp=timestamp,
|
||
|
|
symbol=row['symbol'],
|
||
|
|
timeframe=row['timeframe'],
|
||
|
|
strategy_name=self.strategy_name,
|
||
|
|
signals=[signal],
|
||
|
|
indicators_used={
|
||
|
|
'ema_fast': float(row['ema_fast']),
|
||
|
|
'ema_slow': float(row['ema_slow'])
|
||
|
|
},
|
||
|
|
metadata=strategy_metadata
|
||
|
|
))
|
||
|
|
|
||
|
|
if self.logger:
|
||
|
|
self.logger.info(f"{self.strategy_name}: SELL signal at {timestamp} for {row['symbol']}")
|
||
|
|
|
||
|
|
return results
|