aszer/etf
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

fix(core): 修复计算与数据对齐等多处逻辑问题

Browse Source 
- 修正CAGR计算，去除NaN并检查起始值有效性以避免异常结果
- 优化混合数据源的数据对齐逻辑，使用配置结束日期与A股最新数据日期的较早者
- 计算因子时对齐A股交易日历，重新基于对齐价格计算日收益率，改进因子对齐准确度
- 轮动策略中跳过空信号，避免空信号影响持仓和调仓逻辑
- 调整信号处理，过滤空字符串和NaN，保证轮动信号数据有效性
- 多品种轮动持仓中加入空信号判断，避免无效信号导致错误
- 调整调仓明细和品种汇总保存逻辑，增加空文件创建以保证输出路径文件稳定生成
- 完善多处打印信息和注释，增强代码可读性与调试便利性
This commit is contained in:
aszerW
2026-03-26 22:21:38 +08:00
parent 2faea1517f
commit 70bb69fd98
5 changed files with 114 additions and 49 deletions
Download Patch File Download Diff File Expand all files Collapse all files

15
core/common/utils.py
View File

@@ -72,7 +72,20 @@ def calculate_cagr(
Returns:
float: CAGR值
"""
total_return = nav_series.iloc[-1] / nav_series.iloc[0]
# 去除NaN值
nav_series = nav_series.dropna()
if len(nav_series) < 2:
return 0.0
start_val = nav_series.iloc[0]
end_val = nav_series.iloc[-1]
# 检查起始值是否有效
if pd.isna(start_val) or pd.isna(end_val) or start_val <= 0:
return 0.0
total_return = end_val / start_val
if method == "natural_days":
days = (nav_series.index[-1] - nav_series.index[0]).days

23
core/datasource/hybrid_source.py
View File

@@ -639,18 +639,31 @@ class HybridDataSource:
aggfunc='first'
)
# 数据对齐策略：使用各标的能获取到的最新数据
# 以A股最新数据日期为基准，其他市场数据对齐到该日期
# 数据对齐策略：使用配置的日期范围，确保回测区间与配置一致
# 以A股最新数据日期或配置的end_date（取较早者）为基准
# 获取A股最新数据日期
china_codes = [c for c in valid_codes if self._is_china_index(c)]
if china_codes:
a_share_latest = index_data[china_codes].dropna().index.max()
a_share_data_latest = index_data[china_codes].dropna().index.max()
else:
# 如果没有A股，使用所有数据的最早最新日期
a_share_latest = index_data.dropna().index.max()
a_share_data_latest = index_data.dropna().index.max()
print(f" A股最新数据日期: {a_share_latest.strftime('%Y-%m-%d')}")
# 使用配置的end_date，但不超过数据的最新日期
config_end = pd.Timestamp(end_date)
a_share_latest = min(a_share_data_latest, config_end)
print(f" A股数据最新日期: {a_share_data_latest.strftime('%Y-%m-%d')}")
print(f" 配置结束日期: {config_end.strftime('%Y-%m-%d')}")
print(f" 实际使用日期: {a_share_latest.strftime('%Y-%m-%d')}")
# 打印各标的的数据时间范围（用于调试）
print(f"\n 各标的数据时间范围:")
for code in valid_codes:
code_data = index_data[code].dropna()
if len(code_data) > 0:
print(f" {code}: {code_data.index.min().strftime('%Y-%m-%d')} ~ {code_data.index.max().strftime('%Y-%m-%d')} ({len(code_data)}条)")
# 获取A股交易日历（从start_date到a_share_latest）
start_str = pd.Timestamp(start_date).strftime('%Y%m%d')

25
core/factors/momentum.py
View File

@@ -142,7 +142,7 @@ def compute_factors(
valid_codes.remove(code)
continue
# 按照该标的自己的交易日历计算指标
# 按照该标的自己的交易日历计算指标（使用指数数据）
if factor_type == "momentum":
factor_series = calculate_momentum(price_series, n)
elif factor_type == "slope_r2":
@@ -150,14 +150,18 @@ def compute_factors(
else:
raise ValueError(f"不支持的因子类型: {factor_type}")
# 计算日收益率
return_series = calculate_daily_return(price_series)
# 对齐到A股交易日历：价格使用ffill，指标使用ffill
# 但日收益率需要基于对齐后的价格重新计算，而不是直接ffill
price_aligned = price_series.reindex(a_share_dates, method='ffill')
factor_aligned = factor_series.reindex(a_share_dates, method='ffill')
# 对齐到A股交易日历：取离A股交易日最近的有效数据（不使用未来数据）
# 使用reindex + method='ffill'，确保T日使用T日或之前的数据
result[code] = price_series.reindex(a_share_dates, method='ffill')
result[f"得分_{code}"] = factor_series.reindex(a_share_dates, method='ffill')
result[f"日收益率_{code}"] = return_series.reindex(a_share_dates, method='ffill')
# 基于对齐后的价格重新计算日收益率
# 这样如果T日没有交易（价格被ffill），日收益率为0
return_aligned = calculate_daily_return(price_aligned)
result[code] = price_aligned
result[f"得分_{code}"] = factor_aligned
result[f"日收益率_{code}"] = return_aligned
# 过滤掉缺失值过多的指数（基于A股交易日历）
total_rows = len(result)
@@ -170,9 +174,10 @@ def compute_factors(
else:
final_valid_codes.append(code)
# 按得分列做 dropna（确保所有标的同时有数据）
# 注意：不做dropna，保留所有A股交易日
# 非A股标的在没有数据的日子，得分和日收益率会保持NaN或前向填充值
# 这是正常的横截面策略行为：T日只交易有数据的标的
score_cols = [f"得分_{code}" for code in final_valid_codes]
result = result.dropna(subset=score_cols)
print("\n因子计算完成:")
print(f" 因子类型: {factor_type}")

43
strategies/rotation/engine.py
View File

@@ -115,6 +115,10 @@ class RotationStrategy(BacktestStrategy):
target = daily_target.iloc[i]
if current_held is None:
# 跳过空信号，直到找到第一个有效信号
if not target:
held_signals.append(None) # 添加None占位，保持长度一致
continue
current_held = target
last_rebalance_idx = i
held_signals.append(current_held)
@@ -122,20 +126,24 @@ class RotationStrategy(BacktestStrategy):
days_since = i - last_rebalance_idx
if days_since >= rebalance_days:
should = self._check_rebalance(
result.iloc[i], current_held, target,
select_num, rebalance_threshold
)
if should:
current_held = target
last_rebalance_idx = i
# 目标信号为空时不调仓
if target: # 只在目标有效时才检查是否调仓
should = self._check_rebalance(
result.iloc[i], current_held, target,
select_num, rebalance_threshold
)
if should:
current_held = target
last_rebalance_idx = i
held_signals.append(current_held)
result["信号_raw"] = held_signals
result["信号"] = result["信号_raw"].shift(1)
result = result.drop(columns=["信号_raw"])
# 删除信号为 NaN 或空字符串的行
result = result.dropna(subset=["信号"])
result = result[result["信号"] != ""]
self.signals = result
self._print_signal_stats(result, select_num, rebalance_days, rebalance_threshold)
@@ -152,12 +160,14 @@ class RotationStrategy(BacktestStrategy):
return (new_score / old_score - 1) >= threshold
return new_score > 0
else:
new_codes = target.split(",")
old_codes = current_held.split(",")
new_codes = [c for c in target.split(",") if c] # 过滤空字符串
old_codes = [c for c in current_held.split(",") if c] # 过滤空字符串
if not new_codes or not old_codes:
return True # 有空持仓，需要调仓
if set(new_codes) == set(old_codes):
return False
new_total = sum(float(row[f"得分_{c}"]) for c in new_codes)
old_total = sum(float(row[f"得分_{c}"]) for c in old_codes)
new_total = sum(float(row.get(f"得分_{c}", 0)) for c in new_codes)
old_total = sum(float(row.get(f"得分_{c}", 0)) for c in old_codes)
if old_total > 0:
return (new_total / old_total - 1) >= threshold
return new_total > 0
@@ -206,12 +216,17 @@ class RotationStrategy(BacktestStrategy):
# 计算策略日收益率
if select_num == 1:
def calc_return(row):
return row[f"日收益率_{row['信号']}"]
signal = row['信号']
if not signal or pd.isna(signal):
return 0.0
return row.get(f"日收益率_{signal}", 0.0)
result["轮动策略日收益率"] = result.apply(calc_return, axis=1)
else:
def calc_multi_return(row):
codes = row["信号"].split(",")
returns = [row[f"日收益率_{c}"] for c in codes]
codes = [c for c in row["信号"].split(",") if c] # 过滤空字符串
if not codes:
return 0.0
returns = [row.get(f"日收益率_{c}", 0.0) for c in codes]
return np.mean(returns)
result["轮动策略日收益率"] = result.apply(calc_multi_return, axis=1)

57
strategies/rotation/portfolio.py
View File

@@ -95,7 +95,10 @@ def track_positions(
# 多品种等权轮动
current_signal = signals[0]
entry_date = dates[0]
codes = current_signal.split(",")
codes = [c for c in current_signal.split(",") if c] # 过滤空字符串
if not codes:
# 空信号，返回空结果
return pd.DataFrame(trades), pd.DataFrame()
weight = 1.0 / len(codes)
entry_prices = {c: data.loc[entry_date, c] for c in codes}
entry_nav = data.loc[entry_date, "轮动策略净值"]
@@ -131,7 +134,9 @@ def track_positions(
current_signal = today_signal
entry_date = dates[i]
codes = current_signal.split(",")
codes = [c for c in current_signal.split(",") if c] # 过滤空字符串
if not codes:
break # 空信号，结束循环
weight = 1.0 / len(codes)
entry_prices = {c: data.loc[entry_date, c] for c in codes}
entry_nav = data.loc[entry_date, "轮动策略净值"]
@@ -211,23 +216,37 @@ def save_trades(
import os
os.makedirs(os.path.dirname(save_path) if os.path.dirname(save_path) else ".", exist_ok=True)
trades_out = trades_df.copy()
trades_out["持仓收益"] = trades_out["持仓收益"].apply(lambda x: f"{x:.2%}")
trades_out["进场日期"] = trades_out["进场日期"].apply(
lambda x: x.strftime("%Y-%m-%d") if hasattr(x, "strftime") else str(x)[:10]
)
trades_out["出场日期"] = trades_out["出场日期"].apply(
lambda x: x.strftime("%Y-%m-%d") if hasattr(x, "strftime") else str(x)[:10]
)
# 保存调仓明细
trades_path = f"{save_path}_trades.csv"
trades_out.to_csv(trades_path, index=False, encoding="utf-8-sig")
print(f"\n调仓明细已保存: {trades_path}")
summary_out = summary_df.copy()
for col in ["胜率", "平均收益", "累计收益", "最大单次收益", "最大单次亏损"]:
summary_out[col] = summary_out[col].apply(lambda x: f"{x:.2%}")
if not trades_df.empty:
trades_out = trades_df.copy()
if "持仓收益" in trades_out.columns:
trades_out["持仓收益"] = trades_out["持仓收益"].apply(lambda x: f"{x:.2%}")
if "进场日期" in trades_out.columns:
trades_out["进场日期"] = trades_out["进场日期"].apply(
lambda x: x.strftime("%Y-%m-%d") if hasattr(x, "strftime") else str(x)[:10]
)
if "出场日期" in trades_out.columns:
trades_out["出场日期"] = trades_out["出场日期"].apply(
lambda x: x.strftime("%Y-%m-%d") if hasattr(x, "strftime") else str(x)[:10]
)
trades_out.to_csv(trades_path, index=False, encoding="utf-8-sig")
print(f"\n调仓明细已保存: {trades_path}")
else:
# 创建空文件
pd.DataFrame().to_csv(trades_path, index=False, encoding="utf-8-sig")
print(f"\n调仓明细为空: {trades_path}")
# 保存品种汇总
summary_path = f"{save_path}_summary.csv"
summary_out.to_csv(summary_path, index=False, encoding="utf-8-sig")
print(f"品种汇总已保存: {summary_path}")
if not summary_df.empty:
summary_out = summary_df.copy()
for col in ["胜率", "平均收益", "累计收益", "最大单次收益", "最大单次亏损"]:
if col in summary_out.columns:
summary_out[col] = summary_out[col].apply(lambda x: f"{x:.2%}")
summary_out.to_csv(summary_path, index=False, encoding="utf-8-sig")
print(f"品种汇总已保存: {summary_path}")
else:
# 创建空文件
pd.DataFrame().to_csv(summary_path, index=False, encoding="utf-8-sig")
print(f"品种汇总为空: {summary_path}")