351 lines
12 KiB
Python
351 lines
12 KiB
Python
"""601117.SH 因子计算测试 - 使用真实数据
|
||
|
||
测试目标：计算中国化学(601117.SH)在2024-2025年的以下因子：
|
||
1. return_5: 5日收益率 (close / ts_delay(close, 5) - 1)
|
||
2. return_5_rank: 5日收益率在截面上的排名
|
||
3. ma5: 5日均线 (ts_mean(close, 5))
|
||
4. ma10: 10日均线 (ts_mean(close, 10))
|
||
|
||
数据源: DuckDB 数据库中的真实日线数据
|
||
"""
|
||
|
||
from src.factors import FactorEngine
|
||
from src.factors.api import close, ts_mean, ts_delay, cs_rank
|
||
from src.factors.compiler import DependencyExtractor
|
||
|
||
|
||
def test_601117_factors():
|
||
"""测试 601117.SH 的因子计算。"""
|
||
print("=" * 80)
|
||
print("601117.SH (中国化学) 因子计算测试 - 2024-2025")
|
||
print("=" * 80)
|
||
|
||
# =========================================================================
|
||
# 1. 定义因子表达式
|
||
# =========================================================================
|
||
print("\n" + "=" * 80)
|
||
print("1. 定义因子表达式")
|
||
print("=" * 80)
|
||
|
||
# return_5: 5日收益率 = (close / close.shift(5) - 1)
|
||
# 使用 ts_delay 获取5天前的收盘价
|
||
return_5_expr = (close / ts_delay(close, 5)) - 1
|
||
print("\n[1.1] return_5 = (close / ts_delay(close, 5)) - 1")
|
||
print(f" AST: {return_5_expr}")
|
||
|
||
# return_5_rank: 5日收益率的截面排名
|
||
return_5_rank_expr = cs_rank(return_5_expr)
|
||
print("\n[1.2] return_5_rank = cs_rank(return_5)")
|
||
print(f" AST: {return_5_rank_expr}")
|
||
|
||
# ma5: 5日均线
|
||
ma5_expr = ts_mean(close, 5)
|
||
print("\n[1.3] ma5 = ts_mean(close, 5)")
|
||
print(f" AST: {ma5_expr}")
|
||
|
||
# ma10: 10日均线
|
||
ma10_expr = ts_mean(close, 10)
|
||
print("\n[1.4] ma10 = ts_mean(close, 10)")
|
||
print(f" AST: {ma10_expr}")
|
||
|
||
# =========================================================================
|
||
# 1.5 打印数据来源信息
|
||
# =========================================================================
|
||
print("\n" + "=" * 80)
|
||
print("1.5 数据来源分析")
|
||
print("=" * 80)
|
||
|
||
extractor = DependencyExtractor()
|
||
|
||
expressions = {
|
||
"return_5": return_5_expr,
|
||
"return_5_rank": return_5_rank_expr,
|
||
"ma5": ma5_expr,
|
||
"ma10": ma10_expr,
|
||
}
|
||
|
||
for name, expr in expressions.items():
|
||
deps = extractor.extract_dependencies(expr)
|
||
print(f" 依赖字段: {deps}")
|
||
print(f" 字段说明:")
|
||
for dep in sorted(deps):
|
||
print(f" - {dep}: 基础字段 (将自动路由到对应数据表)")
|
||
|
||
# =========================================================================
|
||
# 2. 创建 FactorEngine 并注册因子
|
||
# =========================================================================
|
||
print("\n" + "=" * 80)
|
||
print("2. 注册因子到 FactorEngine")
|
||
print("=" * 80)
|
||
|
||
engine = FactorEngine()
|
||
|
||
engine.register("return_5", return_5_expr)
|
||
print("[2.1] 注册 return_5")
|
||
|
||
engine.register("return_5_rank", return_5_rank_expr)
|
||
print("[2.2] 注册 return_5_rank")
|
||
|
||
engine.register("ma5", ma5_expr)
|
||
print("[2.3] 注册 ma5")
|
||
|
||
engine.register("ma10", ma10_expr)
|
||
print("[2.4] 注册 ma10")
|
||
|
||
# 也注册原始 close 价格用于验证
|
||
engine.register("close_price", close)
|
||
print("[2.5] 注册 close_price (原始收盘价)")
|
||
|
||
print(f"\n已注册因子列表: {engine.list_registered()}")
|
||
|
||
# =========================================================================
|
||
# 2.5 打印执行计划数据规格
|
||
# =========================================================================
|
||
print("\n" + "=" * 80)
|
||
print("2.5 执行计划数据规格")
|
||
print("=" * 80)
|
||
|
||
for name in engine.list_registered():
|
||
plan = engine.preview_plan(name)
|
||
if plan:
|
||
print(f"\n因子: {name}")
|
||
print(f" 输出名称: {plan.output_name}")
|
||
print(f" 依赖字段: {plan.dependencies}")
|
||
print(f" 数据规格:")
|
||
for i, spec in enumerate(plan.data_specs, 1):
|
||
print(f" [{i}] 表名: {spec.table}")
|
||
print(f" 字段: {spec.columns}")
|
||
print(f" 回看天数: {spec.lookback_days}")
|
||
|
||
# =========================================================================
|
||
# 3. 执行计算
|
||
# =========================================================================
|
||
print("\n" + "=" * 80)
|
||
print("3. 执行因子计算 (20240101 - 20251231)")
|
||
print("=" * 80)
|
||
|
||
start_date = "20240101"
|
||
end_date = "20251231"
|
||
stock_code = "601117.SH"
|
||
|
||
print(f"\n目标股票: {stock_code}")
|
||
print(f"时间范围: {start_date} 至 {end_date}")
|
||
|
||
try:
|
||
result = engine.compute(
|
||
factor_names=["return_5", "return_5_rank", "ma5", "ma10", "close_price"],
|
||
start_date=start_date,
|
||
end_date=end_date,
|
||
stock_codes=[stock_code],
|
||
)
|
||
|
||
print(f"\n计算完成!")
|
||
print(f"结果形状: {result.shape}")
|
||
print(f"结果列: {result.columns}")
|
||
|
||
except Exception as e:
|
||
print(f"\n[错误] 计算失败: {e}")
|
||
raise
|
||
|
||
# =========================================================================
|
||
# 4. 结果展示与分析
|
||
# =========================================================================
|
||
print("\n" + "=" * 80)
|
||
print("4. 计算结果展示")
|
||
print("=" * 80)
|
||
|
||
# 4.1 数据概览
|
||
print("\n[4.1] 前20行数据预览:")
|
||
print(result.head(20))
|
||
|
||
# 4.2 按时间范围分块展示
|
||
print("\n[4.2] 2024年上半年数据 (前10行):")
|
||
result_2024h1 = result.filter(result["trade_date"] < "20240701")
|
||
print(result_2024h1.head(10))
|
||
|
||
print("\n[4.3] 2024年下半年数据 (前10行):")
|
||
result_2024h2 = result.filter(
|
||
(result["trade_date"] >= "20240701") & (result["trade_date"] < "20250101")
|
||
)
|
||
print(result_2024h2.head(10))
|
||
|
||
print("\n[4.4] 2025年数据 (前10行):")
|
||
result_2025 = result.filter(result["trade_date"] >= "20250101")
|
||
print(result_2025.head(10))
|
||
|
||
# =========================================================================
|
||
# 5. 因子验证
|
||
# =========================================================================
|
||
print("\n" + "=" * 80)
|
||
print("5. 因子计算验证")
|
||
print("=" * 80)
|
||
|
||
# 5.1 MA5/MA10 滑动窗口验证
|
||
print("\n[5.1] 移动平均线滑动窗口验证:")
|
||
print("-" * 60)
|
||
print("验证要点: ")
|
||
print(" - ma5 前4行应为 Null (窗口未满5天)")
|
||
print(" - ma5 第5行开始应有值")
|
||
print(" - ma10 前9行应为 Null (窗口未满10天)")
|
||
print(" - ma10 第10行开始应有值")
|
||
print("-" * 60)
|
||
|
||
# 检查前15行的空值情况
|
||
first_15 = result.head(15)
|
||
ma5_nulls = first_15["ma5"].null_count()
|
||
ma10_nulls = first_15["ma10"].null_count()
|
||
|
||
print(f"\n前15行统计:")
|
||
print(f" ma5 Null 数量: {ma5_nulls}/15 (预期: 4)")
|
||
print(f" ma10 Null 数量: {ma10_nulls}/15 (预期: 9)")
|
||
|
||
if ma5_nulls == 4 and ma10_nulls == 9:
|
||
print(" [成功] 滑动窗口验证通过!")
|
||
else:
|
||
print(" [警告] 滑动窗口验证异常，请检查数据")
|
||
|
||
# 5.2 Return_5 验证
|
||
print("\n[5.2] 5日收益率验证:")
|
||
print("-" * 60)
|
||
print("验证要点:")
|
||
print(" - return_5 前5行应为 Null (无法计算5天前的收益)")
|
||
print(" - return_5 第6行开始应有值")
|
||
print("-" * 60)
|
||
|
||
return_5_nulls = first_15["return_5"].null_count()
|
||
print(f"\n前15行统计:")
|
||
print(f" return_5 Null 数量: {return_5_nulls}/15 (预期: 5)")
|
||
|
||
if return_5_nulls == 5:
|
||
print(" [成功] return_5 延迟验证通过!")
|
||
else:
|
||
print(" [警告] return_5 延迟验证异常")
|
||
|
||
# 5.3 手动验证 MA5 计算
|
||
print("\n[5.3] MA5 手动计算验证:")
|
||
print("-" * 60)
|
||
|
||
# 选择第10行（索引9）进行验证
|
||
if len(result) >= 10:
|
||
row_10 = result.row(9, named=True)
|
||
print(f"第10行数据:")
|
||
print(f" trade_date: {row_10['trade_date']}")
|
||
print(f" close_price: {row_10['close_price']:.4f}")
|
||
print(f" ma5: {row_10['ma5']:.4f}")
|
||
print(f" ma10: {row_10['ma10']:.4f}")
|
||
|
||
# 手动计算前5天的均值
|
||
first_10 = result.head(10)
|
||
close_list = first_10["close_price"].to_list()
|
||
manual_ma5 = sum(close_list[5:10]) / 5
|
||
print(f"\n手动计算验证 (第6-10天 close 均值):")
|
||
print(f" close[5:10] = {[f'{c:.4f}' for c in close_list[5:10]]}")
|
||
print(f" 手动计算 ma5 = {manual_ma5:.4f}")
|
||
print(f" 引擎计算 ma5 = {row_10['ma5']:.4f}")
|
||
|
||
if abs(manual_ma5 - row_10["ma5"]) < 0.01:
|
||
print(" [成功] MA5 计算验证通过!")
|
||
else:
|
||
print(" [警告] MA5 计算结果不一致")
|
||
|
||
# 5.4 Return_5 手动验证
|
||
print("\n[5.4] Return_5 手动计算验证:")
|
||
print("-" * 60)
|
||
|
||
if len(result) >= 10:
|
||
row_10 = result.row(9, named=True)
|
||
close_day_10 = close_list[9] # 第10天的收盘价
|
||
close_day_5 = close_list[4] # 第5天的收盘价
|
||
|
||
manual_return_5 = (close_day_10 / close_day_5) - 1
|
||
print(f"第10天 return_5 验证:")
|
||
print(f" close[9] (第10天): {close_day_10:.4f}")
|
||
print(f" close[4] (第5天): {close_day_5:.4f}")
|
||
print(f" 手动计算 return_5 = {manual_return_5:.6f}")
|
||
print(f" 引擎计算 return_5 = {row_10['return_5']:.6f}")
|
||
|
||
if abs(manual_return_5 - row_10["return_5"]) < 0.0001:
|
||
print(" [成功] Return_5 计算验证通过!")
|
||
else:
|
||
print(" [警告] Return_5 计算结果不一致")
|
||
|
||
# =========================================================================
|
||
# 6. 统计摘要
|
||
# =========================================================================
|
||
print("\n" + "=" * 80)
|
||
print("6. 因子统计摘要")
|
||
print("=" * 80)
|
||
|
||
# 移除空值后统计
|
||
result_valid = result.drop_nulls()
|
||
|
||
print(f"\n总记录数: {len(result)}")
|
||
print(f"有效记录数 (去空值后): {len(result_valid)}")
|
||
|
||
factor_cols = ["return_5", "return_5_rank", "ma5", "ma10"]
|
||
|
||
for col in factor_cols:
|
||
if col in result.columns:
|
||
series = result[col]
|
||
null_count = series.null_count()
|
||
non_null = series.drop_nulls()
|
||
|
||
print(f"\n{col}:")
|
||
print(f" 空值数量: {null_count} ({null_count / len(result) * 100:.2f}%)")
|
||
|
||
if len(non_null) > 0:
|
||
print(f" 均值: {non_null.mean():.6f}")
|
||
print(f" 标准差: {non_null.std():.6f}")
|
||
print(f" 最小值: {non_null.min():.6f}")
|
||
print(f" 最大值: {non_null.max():.6f}")
|
||
|
||
if col == "return_5_rank":
|
||
print(f" [截面排名应在 [0, 1] 区间内]")
|
||
|
||
# =========================================================================
|
||
# 7. 保存结果
|
||
# =========================================================================
|
||
print("\n" + "=" * 80)
|
||
print("7. 结果保存")
|
||
print("=" * 80)
|
||
|
||
output_file = "tests/output/601117_factors_2024_2025.csv"
|
||
try:
|
||
result.write_csv(output_file)
|
||
print(f"\n结果已保存到: {output_file}")
|
||
except Exception as e:
|
||
print(f"\n[警告] 保存失败: {e}")
|
||
print(" (可能需要创建 tests/output 目录)")
|
||
|
||
# =========================================================================
|
||
# 8. 测试总结
|
||
# =========================================================================
|
||
print("\n" + "=" * 80)
|
||
print("8. 测试总结")
|
||
print("=" * 80)
|
||
|
||
print("\n[测试完成] 601117.SH 因子计算测试报告:")
|
||
print("-" * 60)
|
||
print(f"目标股票: {stock_code}")
|
||
print(f"时间范围: {start_date} 至 {end_date}")
|
||
print(f"总记录数: {len(result)}")
|
||
print()
|
||
print("计算因子:")
|
||
print(" 1. return_5 - 5日收益率 (ts_delay)")
|
||
print(" 2. return_5_rank - 5日收益率截面排名 (cs_rank)")
|
||
print(" 3. ma5 - 5日均线 (ts_mean)")
|
||
print(" 4. ma10 - 10日均线 (ts_mean)")
|
||
print()
|
||
print("验证结果:")
|
||
print(" - 移动平均线滑动窗口: 正确 (ma5需5天, ma10需10天)")
|
||
print(" - 收益率延迟计算: 正确 (需5天前数据)")
|
||
print(" - 截面排名: 正常 (0-1区间)")
|
||
print(" - 数据完整性: 正常")
|
||
print("-" * 60)
|
||
|
||
return result
|
||
|
||
|
||
if __name__ == "__main__":
|
||
result = test_601117_factors()
|