|
|
@@ -0,0 +1,821 @@
|
|
|
+"""
|
|
|
+期货合约5日移动平均线偏离分析
|
|
|
+研究期货合约在大幅偏离其5天移动平均线（K5）后是否出现重大反转走势
|
|
|
+
|
|
|
+本程序实现完整的分析流程：
|
|
|
+1. 数据结构设置 - 获取主力期货合约数据
|
|
|
+2. 数据收集 - 扩展时间范围获取OHLC数据
|
|
|
+3. 距离计算 - 计算价格相对K5的上下偏离距离
|
|
|
+4. 阈值确定 - 使用百分位数作为极端偏差阈值
|
|
|
+5. 信号检测与分析 - 识别极端偏差并分析后续走势
|
|
|
+6. 结果输出 - 生成Range_1偏差统计和信号分析结果CSV
|
|
|
+
|
|
|
+注：程序使用动态获取的主力合约进行分析，确保分析结果基于真实的交易活动
|
|
|
+
|
|
|
+作者: jukuan研究团队
|
|
|
+日期: 2025-09
|
|
|
+适用平台: 聚宽在线研究平台
|
|
|
+"""
|
|
|
+
|
|
|
+import pandas as pd
|
|
|
+import numpy as np
|
|
|
+from jqdata import *
|
|
|
+import datetime
|
|
|
+import warnings
|
|
|
+warnings.filterwarnings('ignore')
|
|
|
+
|
|
|
+# =====================================================================================
|
|
|
+# 分析配置参数 - 集中配置部分
|
|
|
+# =====================================================================================
|
|
|
+
|
|
|
+class AnalysisConfig:
|
|
|
+ """期货5日移动平均线偏离分析配置参数"""
|
|
|
+
|
|
|
+ # ==================== 时间范围设置 ====================
|
|
|
+ ANALYSIS_START_DATE = datetime.datetime(2025, 8, 1) # 分析开始日期
|
|
|
+ ANALYSIS_END_DATE = datetime.datetime(2025, 8, 20) # 分析结束日期
|
|
|
+
|
|
|
+ # ==================== 时间扩展参数 ====================
|
|
|
+ HISTORICAL_DAYS = 365 # 历史分析期：优势期前天数
|
|
|
+ FORWARD_DAYS = 20 # 正向分析期：优势期后天数
|
|
|
+ FORWARD_MONITOR_DAYS = 10 # 前向监控窗口：交易日数
|
|
|
+
|
|
|
+ # ==================== 偏差阈值设置 ====================
|
|
|
+ DEVIATION_PERCENTILE = 90 # 偏差阈值百分位数（范围：0-100）
|
|
|
+
|
|
|
+ # ==================== 分析规模控制 ====================
|
|
|
+ MAX_ANALYSIS_CONTRACTS = 10 # 最大分析合约数量限制
|
|
|
+
|
|
|
+
|
|
|
+ # ==================== 移动平均线参数 ====================
|
|
|
+ MA_PERIOD = 5 # K线移动平均线周期（天数）
|
|
|
+
|
|
|
+ # ==================== 输出设置 ====================
|
|
|
+ OUTPUT_ENCODING = 'utf-8-sig' # 输出文件编码格式
|
|
|
+ VERBOSE_LOGGING = True # 是否打印详细日志
|
|
|
+
|
|
|
+ @classmethod
|
|
|
+ def print_config(cls):
|
|
|
+ """打印当前配置信息"""
|
|
|
+ print("=== 期货5日移动平均线偏离分析配置 ===")
|
|
|
+ print(f"分析时间范围: {cls.ANALYSIS_START_DATE.strftime('%Y-%m-%d')} 至 {cls.ANALYSIS_END_DATE.strftime('%Y-%m-%d')}")
|
|
|
+ print(f"历史分析期: {cls.HISTORICAL_DAYS}天")
|
|
|
+ print(f"前向分析期: {cls.FORWARD_DAYS}天")
|
|
|
+ print(f"前向监控窗口: {cls.FORWARD_MONITOR_DAYS}个交易日")
|
|
|
+ print(f"偏差阈值百分位数: {cls.DEVIATION_PERCENTILE}%")
|
|
|
+ print(f"最大分析合约数: {cls.MAX_ANALYSIS_CONTRACTS}")
|
|
|
+ print(f"合约类型: 主力合约")
|
|
|
+ print(f"移动平均线周期: {cls.MA_PERIOD}日")
|
|
|
+ print(f"详细日志: {'开启' if cls.VERBOSE_LOGGING else '关闭'}")
|
|
|
+ print("=" * 50)
|
|
|
+
|
|
|
+class FutureMA5DeviationAnalyzer:
|
|
|
+ """期货5日移动平均线偏离分析器"""
|
|
|
+
|
|
|
+ def __init__(self, config=None):
|
|
|
+ """初始化分析器"""
|
|
|
+ # 使用配置类参数
|
|
|
+ if config is None:
|
|
|
+ config = AnalysisConfig
|
|
|
+
|
|
|
+ self.config = config
|
|
|
+ self.historical_days = config.HISTORICAL_DAYS
|
|
|
+ self.forward_days = config.FORWARD_DAYS
|
|
|
+ self.deviation_percentile = config.DEVIATION_PERCENTILE
|
|
|
+ self.forward_monitor_days = config.FORWARD_MONITOR_DAYS
|
|
|
+ self.analysis_start_date = config.ANALYSIS_START_DATE
|
|
|
+ self.analysis_end_date = config.ANALYSIS_END_DATE
|
|
|
+ self.max_analysis_contracts = config.MAX_ANALYSIS_CONTRACTS
|
|
|
+ self.ma_period = config.MA_PERIOD
|
|
|
+ self.output_encoding = config.OUTPUT_ENCODING
|
|
|
+ self.verbose_logging = config.VERBOSE_LOGGING
|
|
|
+
|
|
|
+ # 期货品种与交易所的映射关系
|
|
|
+ self.exchange_map = {
|
|
|
+ # 大连商品交易所 (XDCE)
|
|
|
+ 'A': 'XDCE', 'B': 'XDCE', 'C': 'XDCE', 'CS': 'XDCE', 'FB': 'XDCE',
|
|
|
+ 'I': 'XDCE', 'J': 'XDCE', 'JD': 'XDCE', 'JM': 'XDCE', 'L': 'XDCE',
|
|
|
+ 'M': 'XDCE', 'P': 'XDCE', 'PP': 'XDCE', 'PG': 'XDCE', 'RR': 'XDCE',
|
|
|
+ 'V': 'XDCE', 'Y': 'XDCE', 'EB': 'XDCE', 'EG': 'XDCE', 'LH': 'XDCE',
|
|
|
+
|
|
|
+ # 郑州商品交易所 (XZCE)
|
|
|
+ 'AP': 'XZCE', 'CF': 'XZCE', 'CY': 'XZCE', 'FG': 'XZCE', 'JR': 'XZCE',
|
|
|
+ 'LR': 'XZCE', 'MA': 'XZCE', 'OI': 'XZCE', 'PM': 'XZCE', 'RI': 'XZCE',
|
|
|
+ 'RM': 'XZCE', 'RS': 'XZCE', 'SF': 'XZCE', 'SM': 'XZCE', 'SR': 'XZCE',
|
|
|
+ 'TA': 'XZCE', 'WH': 'XZCE', 'ZC': 'XZCE', 'CJ': 'XZCE',
|
|
|
+ 'ME': 'XZCE', 'PF': 'XZCE', 'PK': 'XZCE', 'RO': 'XZCE', 'SA': 'XZCE',
|
|
|
+ 'TC': 'XZCE', 'UR': 'XZCE', 'WS': 'XZCE', 'WT': 'XZCE',
|
|
|
+
|
|
|
+ # 上海期货交易所 (XSGE)
|
|
|
+ 'AG': 'XSGE', 'AL': 'XSGE', 'AU': 'XSGE', 'BU': 'XSGE', 'CU': 'XSGE',
|
|
|
+ 'FU': 'XSGE', 'HC': 'XSGE', 'NI': 'XSGE', 'PB': 'XSGE', 'RB': 'XSGE',
|
|
|
+ 'RU': 'XSGE', 'SN': 'XSGE', 'SP': 'XSGE', 'SS': 'XSGE', 'WR': 'XSGE',
|
|
|
+ 'ZN': 'XSGE',
|
|
|
+
|
|
|
+ # 上海国际能源交易中心 (XINE)
|
|
|
+ 'BC': 'XINE', 'LU': 'XINE', 'NR': 'XINE', 'SC': 'XINE',
|
|
|
+
|
|
|
+ # 中金所 (CCFX)
|
|
|
+ 'IC': 'CCFX', 'IF': 'CCFX', 'IH': 'CCFX', 'T': 'CCFX', 'TF': 'CCFX', 'TS': 'CCFX',
|
|
|
+
|
|
|
+ # 广期所 (GFEX)
|
|
|
+ 'SI': 'GFEX', 'LC': 'GFEX', 'PS': 'GFEX'
|
|
|
+ }
|
|
|
+
|
|
|
+ # 存储结果的字典
|
|
|
+ self.contract_data = {} # 主合约数据结构
|
|
|
+ self.signal_results = {} # 信号检测结果
|
|
|
+
|
|
|
+ if self.verbose_logging:
|
|
|
+ print("初始化期货5日移动平均线偏离分析器")
|
|
|
+ print(f"配置参数 - 历史分析期: {self.historical_days}天, 前向分析期: {self.forward_days}天")
|
|
|
+ print(f"偏差阈值: {self.deviation_percentile}分位数, 监控窗口: {self.forward_monitor_days}个交易日")
|
|
|
+ print(f"最大分析合约数: {self.max_analysis_contracts}, 移动平均线周期: {self.ma_period}日")
|
|
|
+ print(f"合约类型: 主力合约")
|
|
|
+
|
|
|
+ def _is_trading_day(self, check_date):
|
|
|
+ """
|
|
|
+ 检查是否为交易日
|
|
|
+
|
|
|
+ 参数:
|
|
|
+ check_date (date): 要检查的日期
|
|
|
+
|
|
|
+ 返回:
|
|
|
+ bool: 是否为交易日
|
|
|
+ """
|
|
|
+ try:
|
|
|
+ # 使用聚宽API检查交易日
|
|
|
+ trade_days = get_trade_days(end_date=check_date, count=1)
|
|
|
+ if len(trade_days) > 0:
|
|
|
+ return trade_days[0].date() == check_date
|
|
|
+ return False
|
|
|
+ except:
|
|
|
+ # 如果API调用失败，简单判断是否为工作日
|
|
|
+ return check_date.weekday() < 5
|
|
|
+
|
|
|
+ def _get_symbol_dominant_history(self, symbol, start_date, end_date):
|
|
|
+ """
|
|
|
+ 获取单个标的的主力合约历史变化
|
|
|
+
|
|
|
+ 参数:
|
|
|
+ symbol (str): 标的编码
|
|
|
+ start_date (date): 开始日期
|
|
|
+ end_date (date): 结束日期
|
|
|
+
|
|
|
+ 返回:
|
|
|
+ list: 主力合约记录列表
|
|
|
+ """
|
|
|
+ results = []
|
|
|
+ current_date = start_date
|
|
|
+ current_dominant = None
|
|
|
+ period_start = None
|
|
|
+
|
|
|
+ if self.verbose_logging:
|
|
|
+ print(f" 获取 {symbol} 主力合约历史...")
|
|
|
+
|
|
|
+ # 按日遍历时间范围
|
|
|
+ while current_date <= end_date:
|
|
|
+
|
|
|
+ # 跳过非交易日
|
|
|
+ if not self._is_trading_day(current_date):
|
|
|
+ current_date += datetime.timedelta(days=1)
|
|
|
+ continue
|
|
|
+
|
|
|
+ try:
|
|
|
+ # 获取当日主力合约
|
|
|
+ dominant_contract = get_dominant_future(symbol, current_date)
|
|
|
+
|
|
|
+ if dominant_contract is None:
|
|
|
+ current_date += datetime.timedelta(days=1)
|
|
|
+ continue
|
|
|
+
|
|
|
+ # 如果是新的主力合约
|
|
|
+ if dominant_contract != current_dominant:
|
|
|
+
|
|
|
+ # 如果不是第一个合约，结束上一个合约的记录
|
|
|
+ if current_dominant is not None and period_start is not None:
|
|
|
+ results.append({
|
|
|
+ 'symbol': symbol,
|
|
|
+ 'dominant_contract': current_dominant,
|
|
|
+ 'start_date': period_start,
|
|
|
+ 'end_date': current_date - datetime.timedelta(days=1)
|
|
|
+ })
|
|
|
+
|
|
|
+ # 开始新合约的记录
|
|
|
+ current_dominant = dominant_contract
|
|
|
+ period_start = current_date
|
|
|
+
|
|
|
+ if self.verbose_logging:
|
|
|
+ print(f" {current_date}: 主力合约变更为 {dominant_contract}")
|
|
|
+
|
|
|
+ except Exception as e:
|
|
|
+ if self.verbose_logging:
|
|
|
+ print(f" 获取 {current_date} 的主力合约时出错: {str(e)}")
|
|
|
+
|
|
|
+ current_date += datetime.timedelta(days=1)
|
|
|
+
|
|
|
+ # 处理最后一个合约
|
|
|
+ if current_dominant is not None and period_start is not None:
|
|
|
+ results.append({
|
|
|
+ 'symbol': symbol,
|
|
|
+ 'dominant_contract': current_dominant,
|
|
|
+ 'start_date': period_start,
|
|
|
+ 'end_date': end_date
|
|
|
+ })
|
|
|
+
|
|
|
+ return results
|
|
|
+
|
|
|
+ def get_dominant_contracts_history(self, symbol_list, start_date, end_date):
|
|
|
+ """
|
|
|
+ 获取指定时间范围内多个标的的主力合约历史变化
|
|
|
+
|
|
|
+ 参数:
|
|
|
+ symbol_list (list): 标的编码列表
|
|
|
+ start_date (date): 开始日期
|
|
|
+ end_date (date): 结束日期
|
|
|
+
|
|
|
+ 返回:
|
|
|
+ pandas.DataFrame: 包含主力合约变化历史的数据框
|
|
|
+ """
|
|
|
+ if self.verbose_logging:
|
|
|
+ print(f"获取主力合约数据...")
|
|
|
+ print(f"标的列表: {symbol_list}")
|
|
|
+ print(f"时间范围: {start_date} 至 {end_date}")
|
|
|
+
|
|
|
+ # 存储所有结果的列表
|
|
|
+ all_results = []
|
|
|
+
|
|
|
+ # 逐个处理每个标的
|
|
|
+ for symbol in symbol_list:
|
|
|
+ if self.verbose_logging:
|
|
|
+ print(f"\n处理标的: {symbol}")
|
|
|
+
|
|
|
+ try:
|
|
|
+ # 获取该标的在指定时间范围内的主力合约历史
|
|
|
+ symbol_results = self._get_symbol_dominant_history(symbol, start_date, end_date)
|
|
|
+ all_results.extend(symbol_results)
|
|
|
+
|
|
|
+ except Exception as e:
|
|
|
+ if self.verbose_logging:
|
|
|
+ print(f"处理标的 {symbol} 时发生错误: {str(e)}")
|
|
|
+ continue
|
|
|
+
|
|
|
+ # 将结果转换为DataFrame
|
|
|
+ if all_results:
|
|
|
+ result_df = pd.DataFrame(all_results)
|
|
|
+ result_df = result_df.sort_values(['symbol', 'start_date']).reset_index(drop=True)
|
|
|
+
|
|
|
+ if self.verbose_logging:
|
|
|
+ print(f"\n成功获取 {len(result_df)} 条主力合约记录")
|
|
|
+ return result_df
|
|
|
+ else:
|
|
|
+ if self.verbose_logging:
|
|
|
+ print("\n未获取到任何主力合约数据")
|
|
|
+ return pd.DataFrame(columns=['symbol', 'dominant_contract', 'start_date', 'end_date'])
|
|
|
+
|
|
|
+ def build_contract_structure(self):
|
|
|
+ """
|
|
|
+ 构建期货主力合约数据结构
|
|
|
+ 返回结构：Key=完整合约代码, Value=[start_date, end_date]
|
|
|
+ """
|
|
|
+ if self.verbose_logging:
|
|
|
+ print("\n=== 步骤1: 构建期货主力合约数据结构 ===")
|
|
|
+
|
|
|
+ # 分析时间范围
|
|
|
+ analysis_start = self.analysis_start_date
|
|
|
+ analysis_end = self.analysis_end_date
|
|
|
+
|
|
|
+ if self.verbose_logging:
|
|
|
+ print(f"分析时间范围: {analysis_start.strftime('%Y-%m-%d')} 至 {analysis_end.strftime('%Y-%m-%d')}")
|
|
|
+ print(f"合约类型: 主力合约")
|
|
|
+
|
|
|
+ # 构建主力合约数据结构
|
|
|
+ contract_dict = self._build_dominant_contract_structure(analysis_start, analysis_end)
|
|
|
+
|
|
|
+ self.contract_data = contract_dict
|
|
|
+ if self.verbose_logging:
|
|
|
+ print(f"构建完成，共{len(contract_dict)}个期货合约")
|
|
|
+ return contract_dict
|
|
|
+
|
|
|
+ def _build_dominant_contract_structure(self, analysis_start, analysis_end):
|
|
|
+ """构建基于主力合约的数据结构"""
|
|
|
+ if self.verbose_logging:
|
|
|
+ print("使用主力合约获取逻辑...")
|
|
|
+
|
|
|
+ # 获取要分析的期货品种列表
|
|
|
+ symbol_list = list(self.exchange_map.keys())[:self.max_analysis_contracts]
|
|
|
+
|
|
|
+ # 获取主力合约历史数据
|
|
|
+ dominant_df = self.get_dominant_contracts_history(
|
|
|
+ symbol_list,
|
|
|
+ analysis_start.date(),
|
|
|
+ analysis_end.date()
|
|
|
+ )
|
|
|
+
|
|
|
+ contract_dict = {}
|
|
|
+
|
|
|
+ if not dominant_df.empty:
|
|
|
+ # 将DataFrame转换为字典结构
|
|
|
+ for _, row in dominant_df.iterrows():
|
|
|
+ contract_code = row['dominant_contract']
|
|
|
+ start_date = row['start_date'] if isinstance(row['start_date'], datetime.date) else row['start_date'].date()
|
|
|
+ end_date = row['end_date'] if isinstance(row['end_date'], datetime.date) else row['end_date'].date()
|
|
|
+
|
|
|
+ contract_dict[contract_code] = [start_date, end_date]
|
|
|
+
|
|
|
+ if self.verbose_logging:
|
|
|
+ print(f" 添加主力合约: {contract_code} ({start_date} 至 {end_date})")
|
|
|
+
|
|
|
+ return contract_dict
|
|
|
+
|
|
|
+ def collect_extended_data(self):
|
|
|
+ """
|
|
|
+ 数据收集：为每个合约收集扩展时间范围的数据
|
|
|
+ 扩展期限：从(start_date-365天)到(end_date+100天)
|
|
|
+ """
|
|
|
+ if self.verbose_logging:
|
|
|
+ print("\n=== 步骤2: 收集扩展时间范围的数据 ===")
|
|
|
+
|
|
|
+ extended_data = {}
|
|
|
+
|
|
|
+ for contract_code, period in self.contract_data.items():
|
|
|
+ start_date_1, end_date_1 = period
|
|
|
+
|
|
|
+ # 计算扩展期限
|
|
|
+ start_date_2 = start_date_1 - datetime.timedelta(days=self.historical_days)
|
|
|
+ end_date_2 = end_date_1 + datetime.timedelta(days=self.forward_days)
|
|
|
+
|
|
|
+ if self.verbose_logging:
|
|
|
+ print(f"获取 {contract_code} 数据")
|
|
|
+ print(f" 原始期间: {start_date_1} 至 {end_date_1}")
|
|
|
+ print(f" 扩展期间: {start_date_2} 至 {end_date_2}")
|
|
|
+
|
|
|
+ try:
|
|
|
+ # 获取日线数据
|
|
|
+ data = get_price(
|
|
|
+ contract_code,
|
|
|
+ start_date=start_date_2,
|
|
|
+ end_date=end_date_2,
|
|
|
+ frequency='daily',
|
|
|
+ fields=['open', 'close', 'high', 'low', 'volume'],
|
|
|
+ skip_paused=False,
|
|
|
+ panel=False
|
|
|
+ )
|
|
|
+
|
|
|
+ if data is not None and len(data) > 0:
|
|
|
+ # 计算移动平均线
|
|
|
+ data['K5'] = data['close'].rolling(window=self.ma_period).mean()
|
|
|
+
|
|
|
+ extended_data[contract_code] = {
|
|
|
+ 'data': data,
|
|
|
+ 'original_start': start_date_1,
|
|
|
+ 'original_end': end_date_1,
|
|
|
+ 'extended_start': start_date_2,
|
|
|
+ 'extended_end': end_date_2
|
|
|
+ }
|
|
|
+
|
|
|
+ if self.verbose_logging:
|
|
|
+ print(f" 成功获取{len(data)}条数据记录")
|
|
|
+ else:
|
|
|
+ if self.verbose_logging:
|
|
|
+ print(f" 未获取到数据")
|
|
|
+
|
|
|
+ except Exception as e:
|
|
|
+ if self.verbose_logging:
|
|
|
+ print(f" 获取数据失败: {str(e)}")
|
|
|
+ continue
|
|
|
+
|
|
|
+ if self.verbose_logging:
|
|
|
+ print(f"数据收集完成，共{len(extended_data)}个合约有效数据")
|
|
|
+ return extended_data
|
|
|
+
|
|
|
+ def calculate_k5_distances(self, extended_data):
|
|
|
+ """
|
|
|
+ 计算距离：每个交易日价格相对K5的上下偏离距离
|
|
|
+ """
|
|
|
+ if self.verbose_logging:
|
|
|
+ print("\n=== 步骤3: 计算K5偏离距离 ===")
|
|
|
+
|
|
|
+ distance_data = {}
|
|
|
+
|
|
|
+ for contract_code, contract_info in extended_data.items():
|
|
|
+ data = contract_info['data']
|
|
|
+ if self.verbose_logging:
|
|
|
+ print(f"计算 {contract_code} 的K5距离")
|
|
|
+
|
|
|
+ # 初始化距离列
|
|
|
+ data['upper_distance'] = np.nan
|
|
|
+ data['lower_distance'] = np.nan
|
|
|
+
|
|
|
+ # 计算每日的上下偏离距离
|
|
|
+ for idx, row in data.iterrows():
|
|
|
+ if pd.isna(row['K5']): # K5不可用则跳过
|
|
|
+ continue
|
|
|
+
|
|
|
+ K5 = row['K5']
|
|
|
+ high = row['high']
|
|
|
+ low = row['low']
|
|
|
+
|
|
|
+ # 计算K5与高低价的差值
|
|
|
+ k5_minus_high = K5 - high
|
|
|
+ k5_minus_low = K5 - low
|
|
|
+
|
|
|
+ if k5_minus_high > 0 and k5_minus_low > 0:
|
|
|
+ # 高低都低于K5
|
|
|
+ data.loc[idx, 'upper_distance'] = np.nan
|
|
|
+ data.loc[idx, 'lower_distance'] = abs(K5 - low)
|
|
|
+
|
|
|
+ elif k5_minus_high < 0 and k5_minus_low < 0:
|
|
|
+ # 高低都在K5以上
|
|
|
+ data.loc[idx, 'upper_distance'] = abs(K5 - high)
|
|
|
+ data.loc[idx, 'lower_distance'] = np.nan
|
|
|
+
|
|
|
+ elif k5_minus_high > 0 and k5_minus_low < 0:
|
|
|
+ # 高低于K5，低高于K5（K5介于高与低之间）
|
|
|
+ data.loc[idx, 'upper_distance'] = abs(K5 - high)
|
|
|
+ data.loc[idx, 'lower_distance'] = abs(K5 - low)
|
|
|
+
|
|
|
+ # 处理零情况
|
|
|
+ if abs(k5_minus_high) < 1e-6: # high == K5
|
|
|
+ data.loc[idx, 'upper_distance'] = 0.0
|
|
|
+ if abs(k5_minus_low) < 1e-6: # low == K5
|
|
|
+ data.loc[idx, 'lower_distance'] = 0.0
|
|
|
+
|
|
|
+ distance_data[contract_code] = contract_info
|
|
|
+
|
|
|
+ # 统计有效距离数据
|
|
|
+ valid_upper = data['upper_distance'].notna().sum()
|
|
|
+ valid_lower = data['lower_distance'].notna().sum()
|
|
|
+ if self.verbose_logging:
|
|
|
+ print(f" 上偏离距离有效数据: {valid_upper}条, 下偏离距离有效数据: {valid_lower}条")
|
|
|
+
|
|
|
+ if self.verbose_logging:
|
|
|
+ print("K5距离计算完成")
|
|
|
+ return distance_data
|
|
|
+
|
|
|
+ def determine_thresholds(self, distance_data):
|
|
|
+ """
|
|
|
+ 阈值确定：使用Range_1数据计算百分位数阈值
|
|
|
+ Range_1: start_date_2 <= date < start_date_1
|
|
|
+ """
|
|
|
+ if self.verbose_logging:
|
|
|
+ print(f"\n=== 步骤4: 确定{self.deviation_percentile}分位数阈值 ===")
|
|
|
+
|
|
|
+ threshold_data = {}
|
|
|
+
|
|
|
+ for contract_code, contract_info in distance_data.items():
|
|
|
+ data = contract_info['data']
|
|
|
+ start_date_1 = contract_info['original_start']
|
|
|
+ start_date_2 = contract_info['extended_start']
|
|
|
+
|
|
|
+ if self.verbose_logging:
|
|
|
+ print(f"计算 {contract_code} 的阈值")
|
|
|
+ print(f" Range_1期间: {start_date_2} 至 {start_date_1}")
|
|
|
+
|
|
|
+ # 筛选Range_1数据
|
|
|
+ range_1_mask = (data.index.date >= start_date_2) & (data.index.date < start_date_1)
|
|
|
+ range_1_data = data[range_1_mask]
|
|
|
+
|
|
|
+ if len(range_1_data) == 0:
|
|
|
+ if self.verbose_logging:
|
|
|
+ print(f" Range_1无有效数据，跳过")
|
|
|
+ continue
|
|
|
+
|
|
|
+ # 计算上下偏离距离的分布和阈值
|
|
|
+ upper_distances = range_1_data['upper_distance'].dropna()
|
|
|
+ lower_distances = range_1_data['lower_distance'].dropna()
|
|
|
+
|
|
|
+ # 百分位数阈值
|
|
|
+ percentile_value = self.deviation_percentile / 100.0
|
|
|
+ upper_threshold = upper_distances.quantile(percentile_value) if len(upper_distances) > 0 else np.nan
|
|
|
+ lower_threshold = lower_distances.quantile(percentile_value) if len(lower_distances) > 0 else np.nan
|
|
|
+
|
|
|
+ threshold_data[contract_code] = {
|
|
|
+ 'contract_info': contract_info,
|
|
|
+ 'upper_threshold': upper_threshold,
|
|
|
+ 'lower_threshold': lower_threshold,
|
|
|
+ 'range_1_stats': {
|
|
|
+ 'total_days': len(range_1_data),
|
|
|
+ 'upper_valid_days': len(upper_distances),
|
|
|
+ 'lower_valid_days': len(lower_distances),
|
|
|
+ 'upward_deviation_days': len(lower_distances[lower_distances >= lower_threshold]) if not pd.isna(lower_threshold) else 0,
|
|
|
+ 'downward_deviation_days': len(upper_distances[upper_distances >= upper_threshold]) if not pd.isna(upper_threshold) else 0
|
|
|
+ }
|
|
|
+ }
|
|
|
+
|
|
|
+ if self.verbose_logging:
|
|
|
+ print(f" 上偏离{self.deviation_percentile}分位数阈值: {upper_threshold:.4f}" if not pd.isna(upper_threshold) else f" 上偏离阈值: 无效")
|
|
|
+ print(f" 下偏离{self.deviation_percentile}分位数阈值: {lower_threshold:.4f}" if not pd.isna(lower_threshold) else f" 下偏离阈值: 无效")
|
|
|
+
|
|
|
+ if self.verbose_logging:
|
|
|
+ print("阈值确定完成")
|
|
|
+ return threshold_data
|
|
|
+
|
|
|
+ def detect_signals_and_analyze(self, threshold_data):
|
|
|
+ """
|
|
|
+ 信号检测与分析：在Range_2期间扫描极端偏差并分析后续走势
|
|
|
+ Range_2: start_date_1 <= date <= end_date_2
|
|
|
+ """
|
|
|
+ if self.verbose_logging:
|
|
|
+ print("\n=== 步骤5: 信号检测与前瞻性分析 ===")
|
|
|
+
|
|
|
+ signal_dict = {}
|
|
|
+
|
|
|
+ for contract_code, threshold_info in threshold_data.items():
|
|
|
+ contract_info = threshold_info['contract_info']
|
|
|
+ data = contract_info['data']
|
|
|
+ start_date_1 = contract_info['original_start']
|
|
|
+ end_date_2 = contract_info['extended_end']
|
|
|
+
|
|
|
+ upper_threshold = threshold_info['upper_threshold']
|
|
|
+ lower_threshold = threshold_info['lower_threshold']
|
|
|
+
|
|
|
+ if self.verbose_logging:
|
|
|
+ print(f"\n分析 {contract_code} 的信号")
|
|
|
+ print(f" Range_2期间: {start_date_1} 至 {end_date_2}")
|
|
|
+
|
|
|
+ # 筛选Range_2数据
|
|
|
+ range_2_mask = (data.index.date >= start_date_1) & (data.index.date <= end_date_2)
|
|
|
+ range_2_data = data[range_2_mask]
|
|
|
+
|
|
|
+ contract_signals = {}
|
|
|
+
|
|
|
+ # 扫描极端偏差天数
|
|
|
+ for idx, (signal_date, row) in enumerate(range_2_data.iterrows()):
|
|
|
+ upper_distance = row['upper_distance']
|
|
|
+ lower_distance = row['lower_distance']
|
|
|
+
|
|
|
+ # 检查是否触发阈值
|
|
|
+ trigger_upper = not pd.isna(upper_distance) and not pd.isna(upper_threshold) and upper_distance >= upper_threshold
|
|
|
+ trigger_lower = not pd.isna(lower_distance) and not pd.isna(lower_threshold) and lower_distance >= lower_threshold
|
|
|
+
|
|
|
+ if trigger_upper or trigger_lower:
|
|
|
+ # 确定偏离方向和参考价格
|
|
|
+ if trigger_lower: # lower_distance触发表示向上偏离（价格在K5之上）
|
|
|
+ deviation_direction = "up"
|
|
|
+ deviation_distance = lower_distance
|
|
|
+ reference_category = "low"
|
|
|
+ reference_price = row['low']
|
|
|
+ else: # upper_distance触发表示向下偏离（价格在K5之下）
|
|
|
+ deviation_direction = "down"
|
|
|
+ deviation_distance = upper_distance
|
|
|
+ reference_category = "high"
|
|
|
+ reference_price = row['high']
|
|
|
+
|
|
|
+ if self.verbose_logging:
|
|
|
+ print(f" 发现信号: {signal_date.strftime('%Y-%m-%d')} {deviation_direction}偏离 距离{deviation_distance:.4f}")
|
|
|
+
|
|
|
+ # 前瞻性分析
|
|
|
+ forward_analysis = self._analyze_forward_performance(
|
|
|
+ range_2_data, idx, deviation_direction, reference_price, self.forward_monitor_days
|
|
|
+ )
|
|
|
+
|
|
|
+ # 计算最大利润和风险
|
|
|
+ max_profit_info = self._find_max_profit(forward_analysis)
|
|
|
+ max_risk_info = self._find_max_risk(forward_analysis)
|
|
|
+
|
|
|
+ contract_signals[signal_date] = {
|
|
|
+ "deviation_direction": deviation_direction,
|
|
|
+ "deviation_distance": float(deviation_distance),
|
|
|
+ "reference_category": reference_category,
|
|
|
+ "reference_price": float(reference_price),
|
|
|
+ "forward_analysis": forward_analysis,
|
|
|
+ "max_profit": max_profit_info,
|
|
|
+ "max_risk": max_risk_info
|
|
|
+ }
|
|
|
+
|
|
|
+ if contract_signals:
|
|
|
+ signal_dict[contract_code] = contract_signals
|
|
|
+ if self.verbose_logging:
|
|
|
+ print(f" 检测到{len(contract_signals)}个有效信号")
|
|
|
+ else:
|
|
|
+ if self.verbose_logging:
|
|
|
+ print(f" 未检测到有效信号")
|
|
|
+
|
|
|
+ if self.verbose_logging:
|
|
|
+ print(f"\n信号检测完成，共{sum(len(signals) for signals in signal_dict.values())}个信号")
|
|
|
+ self.signal_results = signal_dict
|
|
|
+ return signal_dict
|
|
|
+
|
|
|
+ def _analyze_forward_performance(self, data, signal_idx, deviation_direction, reference_price, monitor_days):
|
|
|
+ """分析信号后续N个交易日的表现"""
|
|
|
+ forward_analysis = {}
|
|
|
+
|
|
|
+ for i in range(1, monitor_days + 1):
|
|
|
+ future_idx = signal_idx + i
|
|
|
+ if future_idx < len(data):
|
|
|
+ future_row = data.iloc[future_idx]
|
|
|
+ future_date = data.index[future_idx]
|
|
|
+
|
|
|
+ # 根据偏离方向选择对应价格
|
|
|
+ if deviation_direction == "up":
|
|
|
+ corresponding_price = future_row['low']
|
|
|
+ corresponding_return = (reference_price - corresponding_price) / reference_price
|
|
|
+ reverse_return = (future_row['high'] - reference_price) / reference_price
|
|
|
+ else: # down
|
|
|
+ corresponding_price = future_row['high']
|
|
|
+ corresponding_return = (corresponding_price - reference_price) / reference_price
|
|
|
+ reverse_return = (reference_price - future_row['low']) / reference_price
|
|
|
+
|
|
|
+ forward_analysis[future_date] = {
|
|
|
+ "date": future_date.strftime('%Y-%m-%d'),
|
|
|
+ "corresponding_price": float(corresponding_price),
|
|
|
+ "corresponding_return": float(corresponding_return),
|
|
|
+ "reverse_category": "high" if deviation_direction == "up" else "low",
|
|
|
+ "reverse_return": float(reverse_return)
|
|
|
+ }
|
|
|
+
|
|
|
+ return forward_analysis
|
|
|
+
|
|
|
+ def _find_max_profit(self, forward_analysis):
|
|
|
+ """找到最大利润日期和收益率"""
|
|
|
+ if not forward_analysis:
|
|
|
+ return {"date": None, "return": 0.0}
|
|
|
+
|
|
|
+ max_return = max(day_data["corresponding_return"] for day_data in forward_analysis.values())
|
|
|
+ max_date = None
|
|
|
+
|
|
|
+ for date, day_data in forward_analysis.items():
|
|
|
+ if day_data["corresponding_return"] == max_return:
|
|
|
+ max_date = date.strftime('%Y-%m-%d')
|
|
|
+ break
|
|
|
+
|
|
|
+ return {"date": max_date, "return": float(max_return)}
|
|
|
+
|
|
|
+ def _find_max_risk(self, forward_analysis):
|
|
|
+ """找到最大风险日期和收益率"""
|
|
|
+ if not forward_analysis:
|
|
|
+ return {"date": None, "return": 0.0}
|
|
|
+
|
|
|
+ max_risk = max(day_data["reverse_return"] for day_data in forward_analysis.values())
|
|
|
+ max_risk_date = None
|
|
|
+
|
|
|
+ for date, day_data in forward_analysis.items():
|
|
|
+ if day_data["reverse_return"] == max_risk:
|
|
|
+ max_risk_date = date.strftime('%Y-%m-%d')
|
|
|
+ break
|
|
|
+
|
|
|
+ return {"date": max_risk_date, "return": float(max_risk)}
|
|
|
+
|
|
|
+ def generate_csv_outputs(self, threshold_data, signal_dict):
|
|
|
+ """生成两个CSV文件输出"""
|
|
|
+ if self.verbose_logging:
|
|
|
+ print("\n=== 步骤6: 生成CSV输出文件 ===")
|
|
|
+
|
|
|
+ timestamp = datetime.datetime.now().strftime('%Y%m%d_%H%M%S')
|
|
|
+
|
|
|
+ # CSV 1: Range_1偏差统计
|
|
|
+ self._generate_range1_stats_csv(threshold_data, timestamp)
|
|
|
+
|
|
|
+ # CSV 2: 信号分析结果
|
|
|
+ self._generate_signal_analysis_csv(signal_dict, timestamp)
|
|
|
+
|
|
|
+ if self.verbose_logging:
|
|
|
+ print("CSV文件生成完成")
|
|
|
+
|
|
|
+ def _generate_range1_stats_csv(self, threshold_data, timestamp):
|
|
|
+ """生成Range_1偏差统计CSV"""
|
|
|
+ stats_data = []
|
|
|
+
|
|
|
+ for contract_code, threshold_info in threshold_data.items():
|
|
|
+ stats = threshold_info['range_1_stats']
|
|
|
+ total_days = stats['total_days']
|
|
|
+
|
|
|
+ upward_ratio = stats['upward_deviation_days'] / total_days if total_days > 0 else 0
|
|
|
+ downward_ratio = stats['downward_deviation_days'] / total_days if total_days > 0 else 0
|
|
|
+
|
|
|
+ stats_data.append({
|
|
|
+ 'contract_code': contract_code,
|
|
|
+ 'upward_deviation_ratio': round(upward_ratio, 4),
|
|
|
+ 'downward_deviation_ratio': round(downward_ratio, 4)
|
|
|
+ })
|
|
|
+
|
|
|
+ if stats_data:
|
|
|
+ stats_df = pd.DataFrame(stats_data)
|
|
|
+ filename = f'range1_deviation_stats_{timestamp}.csv'
|
|
|
+ stats_df.to_csv(filename, index=False, encoding=self.output_encoding)
|
|
|
+ if self.verbose_logging:
|
|
|
+ print(f"Range_1偏差统计保存至: {filename}")
|
|
|
+
|
|
|
+ def _generate_signal_analysis_csv(self, signal_dict, timestamp):
|
|
|
+ """生成信号分析结果CSV"""
|
|
|
+ signal_data = []
|
|
|
+
|
|
|
+ for contract_code, contract_signals in signal_dict.items():
|
|
|
+ for signal_date, signal_info in contract_signals.items():
|
|
|
+ signal_data.append({
|
|
|
+ 'contract_code': contract_code,
|
|
|
+ 'signal_date': signal_date.strftime('%Y-%m-%d'),
|
|
|
+ 'deviation_direction': signal_info['deviation_direction'],
|
|
|
+ 'max_profit_date': signal_info['max_profit']['date'],
|
|
|
+ 'max_profit_return': round(signal_info['max_profit']['return'], 4),
|
|
|
+ 'max_risk_date': signal_info['max_risk']['date'],
|
|
|
+ 'max_risk_return': round(signal_info['max_risk']['return'], 4)
|
|
|
+ })
|
|
|
+
|
|
|
+ if signal_data:
|
|
|
+ signal_df = pd.DataFrame(signal_data)
|
|
|
+ filename = f'signal_analysis_results_{timestamp}.csv'
|
|
|
+ signal_df.to_csv(filename, index=False, encoding=self.output_encoding)
|
|
|
+ if self.verbose_logging:
|
|
|
+ print(f"信号分析结果保存至: {filename}")
|
|
|
+
|
|
|
+ def run_complete_analysis(self):
|
|
|
+ """执行完整的分析流程"""
|
|
|
+ if self.verbose_logging:
|
|
|
+ print("开始执行期货5日移动平均线偏离分析")
|
|
|
+ print("=" * 60)
|
|
|
+
|
|
|
+ try:
|
|
|
+ # 步骤1: 构建合约数据结构
|
|
|
+ contract_structure = self.build_contract_structure()
|
|
|
+
|
|
|
+ # 步骤2: 收集扩展数据
|
|
|
+ extended_data = self.collect_extended_data()
|
|
|
+ if not extended_data:
|
|
|
+ if self.verbose_logging:
|
|
|
+ print("未获取到有效数据，分析终止")
|
|
|
+ return
|
|
|
+
|
|
|
+ # 步骤3: 计算K5距离
|
|
|
+ distance_data = self.calculate_k5_distances(extended_data)
|
|
|
+
|
|
|
+ # 步骤4: 确定阈值
|
|
|
+ threshold_data = self.determine_thresholds(distance_data)
|
|
|
+ if not threshold_data:
|
|
|
+ if self.verbose_logging:
|
|
|
+ print("未能确定有效阈值，分析终止")
|
|
|
+ return
|
|
|
+
|
|
|
+ # 步骤5: 信号检测与分析
|
|
|
+ signal_dict = self.detect_signals_and_analyze(threshold_data)
|
|
|
+
|
|
|
+ # 步骤6: 生成输出文件
|
|
|
+ self.generate_csv_outputs(threshold_data, signal_dict)
|
|
|
+
|
|
|
+ # 分析汇总
|
|
|
+ total_contracts = len(contract_structure)
|
|
|
+ valid_contracts = len(threshold_data)
|
|
|
+ total_signals = sum(len(signals) for signals in signal_dict.values())
|
|
|
+
|
|
|
+ if self.verbose_logging:
|
|
|
+ print("\n" + "=" * 60)
|
|
|
+ print("分析完成汇总:")
|
|
|
+ print(f"总合约数: {total_contracts}")
|
|
|
+ print(f"有效合约数: {valid_contracts}")
|
|
|
+ print(f"检测信号数: {total_signals}")
|
|
|
+
|
|
|
+ return {
|
|
|
+ 'contract_structure': contract_structure,
|
|
|
+ 'threshold_data': threshold_data,
|
|
|
+ 'signal_results': signal_dict,
|
|
|
+ 'summary': {
|
|
|
+ 'total_contracts': total_contracts,
|
|
|
+ 'valid_contracts': valid_contracts,
|
|
|
+ 'total_signals': total_signals
|
|
|
+ }
|
|
|
+ }
|
|
|
+
|
|
|
+ except Exception as e:
|
|
|
+ if self.verbose_logging:
|
|
|
+ print(f"分析过程中出现错误: {str(e)}")
|
|
|
+ import traceback
|
|
|
+ traceback.print_exc()
|
|
|
+ return None
|
|
|
+
|
|
|
+
|
|
|
+# =====================================================================================
|
|
|
+# 主程序入口
|
|
|
+# =====================================================================================
|
|
|
+
|
|
|
+def run_ma5_deviation_analysis(config=None):
|
|
|
+ """运行期货5日移动平均线偏离分析"""
|
|
|
+ if config is None:
|
|
|
+ config = AnalysisConfig
|
|
|
+
|
|
|
+ # 打印配置信息
|
|
|
+ config.print_config()
|
|
|
+
|
|
|
+ # 创建分析器并运行
|
|
|
+ analyzer = FutureMA5DeviationAnalyzer(config)
|
|
|
+ results = analyzer.run_complete_analysis()
|
|
|
+ return results
|
|
|
+
|
|
|
+# 执行分析
|
|
|
+if __name__ == "__main__":
|
|
|
+ print("期货5日移动平均线偏离分析工具")
|
|
|
+ print("研究期货合约在大幅偏离K5后的反转走势规律")
|
|
|
+ print("使用动态获取的主力合约进行精准分析")
|
|
|
+ print("适用于聚宽在线研究平台")
|
|
|
+
|
|
|
+ results = run_ma5_deviation_analysis()
|
|
|
+
|
|
|
+ if results:
|
|
|
+ print("\n✅ 分析执行成功！")
|
|
|
+ summary = results['summary']
|
|
|
+ print(f"📊 结果摘要:")
|
|
|
+ print(f" - 分析合约数: {summary['total_contracts']}")
|
|
|
+ print(f" - 有效数据合约: {summary['valid_contracts']}")
|
|
|
+ print(f" - 检测信号总数: {summary['total_signals']}")
|
|
|
+ else:
|
|
|
+ print("\n❌ 分析执行失败，请检查错误信息")
|
maxfeng