mirror of
https://ghfast.top/https://github.com/aeroxw/easy-tdx.git
synced 2026-09-12 19:14:16 +08:00
Justin Gu
c15bd8232f
通达信服务端 QFQ 模式对长期重度除权股票（如 601088）深层历史页 返回负价格，导致回测总收益 -3087%、回撤 326.85%、年化 nan、 bollinger 崩溃、10 策略 invalid-value-in-scalar-power、MyTT divide-by-zero。 客户端兜底：检测 QFQ 负价时用 NONE+XDXR 本地重算前复权 （因子以除权日前一交易日含权收盘价为基准，保证除权日前后连续）。 同步+异步双路径一致修复，失败降级返回原值。 - 新增 src/easy_tdx/mac/adjust.py（纯函数 compute_forward_factor/ apply_forward_adjust/has_bad_prices） - MacClient/AsyncMacClient 触发本地重算，XDXR 按 (market,code) 缓存 - tests: +20 例（16 纯函数 + 4 集成），844 全绿，ruff/mypy 通过
210 lines
7.2 KiB
Python
210 lines
7.2 KiB
Python
"""本地前复权（QFQ）重算纯函数的单元测试。
|
||
|
||
覆盖 ``easy_tdx.mac.adjust`` 的因子计算与 OHLC 缩放，重点验证：
|
||
- 除权日前后价格连续（前复权的定义性性质）；
|
||
- 最新价锚定不动；
|
||
- 无事件 / 非法因子时安全降级。
|
||
|
||
公式见 ``examples/06_finance/xdxr_info.py`` 与 ``src/easy_tdx/mac/adjust.py``。
|
||
"""
|
||
|
||
from __future__ import annotations
|
||
|
||
import numpy as np
|
||
import pandas as pd
|
||
|
||
from easy_tdx.mac.adjust import (
|
||
apply_forward_adjust,
|
||
compute_forward_factor,
|
||
has_bad_prices,
|
||
)
|
||
|
||
|
||
def _kline(closes: list[float], start: str = "2024-01-01") -> pd.DataFrame:
|
||
"""构造最小 NONE K 线：OHLC 全等于给定 close 序列，逐日递增。"""
|
||
n = len(closes)
|
||
dates = pd.date_range(start, periods=n, freq="D")
|
||
arr = np.array(closes, dtype=float)
|
||
return pd.DataFrame(
|
||
{
|
||
"datetime": dates,
|
||
"open": arr,
|
||
"high": arr,
|
||
"low": arr,
|
||
"close": arr,
|
||
"vol": [100.0] * n,
|
||
}
|
||
)
|
||
|
||
|
||
def _xdxr_one(date: str, fenhong=0.0, peigujia=0.0, songzhuangu=0.0, peigu=0.0) -> pd.DataFrame:
|
||
"""构造单条 category==1 除权除息记录。"""
|
||
return pd.DataFrame(
|
||
[
|
||
{
|
||
"date": date,
|
||
"category": 1,
|
||
"fenhong": fenhong,
|
||
"peigujia": peigujia,
|
||
"songzhuangu": songzhuangu,
|
||
"peigu": peigu,
|
||
}
|
||
]
|
||
)
|
||
|
||
|
||
# --------------------------------------------------------------------------- #
|
||
# compute_forward_factor
|
||
# --------------------------------------------------------------------------- #
|
||
|
||
|
||
def test_factor_pure_cash_dividend():
|
||
"""纯现金分红：f = (P - fh) / P。"""
|
||
# P=10, fh=2 → f=0.8（前复权把含权价缩到除权后等价）
|
||
assert compute_forward_factor(10.0, 2.0, 0.0, 0.0, 0.0) == 0.8
|
||
|
||
|
||
def test_factor_songzhuangu_split():
|
||
"""10 送 10（songzhuangu=1.0）：f = P / (P*2) = 0.5。"""
|
||
assert compute_forward_factor(10.0, 0.0, 0.0, 1.0, 0.0) == 0.5
|
||
|
||
|
||
def test_factor_zero_cum_close_is_nan():
|
||
"""cum_close<=0 返回 NaN（不抛）。"""
|
||
assert np.isnan(compute_forward_factor(0.0, 1.0, 0.0, 0.0, 0.0))
|
||
assert np.isnan(compute_forward_factor(-1.0, 1.0, 0.0, 0.0, 0.0))
|
||
|
||
|
||
def test_factor_zero_denominator_is_nan():
|
||
"""分母 = P*(1+s+p) 为 0 时返回 NaN。"""
|
||
# P=10, s=-1, p=0 → denom=0
|
||
assert np.isnan(compute_forward_factor(10.0, 0.0, 0.0, -1.0, 0.0))
|
||
|
||
|
||
# --------------------------------------------------------------------------- #
|
||
# apply_forward_adjust
|
||
# --------------------------------------------------------------------------- #
|
||
|
||
|
||
def test_apply_pure_cash_dividend_scales_pre_event_only():
|
||
"""纯分红：除权日及之前价格乘 f，除权日之后不动。"""
|
||
# 3 根：cum-div close=10（含权），ex-date close=8（跌去 2 元分红），之后 9
|
||
# 事件在 day2（ex-date），cum-div bar 是 day1
|
||
df = _kline([10.0, 10.0, 8.0, 9.0])
|
||
xd = _xdxr_one("2024-01-03", fenhong=2.0) # ex-date = 第 3 天
|
||
out = apply_forward_adjust(df, xd)
|
||
# f = (10-2)/10 = 0.8 → day1/day2 (cum 及之前) *= 0.8
|
||
assert out["close"].tolist() == [8.0, 8.0, 8.0, 9.0]
|
||
|
||
|
||
def test_apply_latest_price_anchored():
|
||
"""最新价（最后一根）不被缩放，保持原值。"""
|
||
df = _kline([20.0, 10.0, 11.0])
|
||
xd = _xdxr_one("2024-01-02", fenhong=10.0) # ex-date=day2, cum-div=day1 close=20
|
||
out = apply_forward_adjust(df, xd)
|
||
# f=(20-10)/20=0.5 → day1*=0.5; day2/day3 不动
|
||
assert out["close"].iloc[-1] == 11.0
|
||
assert out["close"].iloc[0] == 10.0
|
||
|
||
|
||
def test_apply_no_events_returns_unchanged():
|
||
"""空 XDXR 或无 category==1 → 原样返回（值相等）。"""
|
||
df = _kline([10.0, 11.0, 12.0])
|
||
out = apply_forward_adjust(df, pd.DataFrame(columns=["date", "category"]))
|
||
assert out["close"].tolist() == [10.0, 11.0, 12.0]
|
||
|
||
|
||
def test_apply_empty_xdxr_df():
|
||
"""XDXR 为 None 或 empty → 原样返回。"""
|
||
df = _kline([10.0, 11.0])
|
||
assert apply_forward_adjust(df, pd.DataFrame()).equals(df)
|
||
assert apply_forward_adjust(df, None).equals(df) # type: ignore[arg-type]
|
||
|
||
|
||
def test_apply_nan_factor_event_skipped():
|
||
"""事件对应 cum_close<=0（因子非法）→ 跳过该事件，不抛异常。"""
|
||
# day1 close=0（非法 cum-div），事件在 day2
|
||
df = _kline([0.0, 5.0, 6.0])
|
||
xd = _xdxr_one("2024-01-02", fenhong=1.0)
|
||
out = apply_forward_adjust(df, xd) # 不应抛
|
||
# cum-div bar (day1) close=0 → 因子 NaN → 跳过，close 不变
|
||
assert out["close"].tolist() == [0.0, 5.0, 6.0]
|
||
|
||
|
||
def test_apply_multiple_events_cumulative():
|
||
"""两次连续分红：因子累乘。"""
|
||
# day1=20(cum1), day2 ex-date1, day3=12(cum2), day4 ex-date2, day5=10
|
||
df = _kline([20.0, 18.0, 12.0, 10.0, 10.0])
|
||
xd = pd.DataFrame(
|
||
[
|
||
{
|
||
"date": "2024-01-02",
|
||
"category": 1,
|
||
"fenhong": 2.0,
|
||
"peigujia": 0.0,
|
||
"songzhuangu": 0.0,
|
||
"peigu": 0.0,
|
||
},
|
||
{
|
||
"date": "2024-01-04",
|
||
"category": 1,
|
||
"fenhong": 2.0,
|
||
"peigujia": 0.0,
|
||
"songzhuangu": 0.0,
|
||
"peigu": 0.0,
|
||
},
|
||
]
|
||
)
|
||
out = apply_forward_adjust(df, xd)
|
||
# event1: cum1=day1=20, f1=(20-2)/20=0.9 → day1*=0.9 → 18.0
|
||
# event2: cum2=day3=12, f2=(12-2)/12=0.8333 → day1..day3 *= 0.8333
|
||
# day1: 18.0 * 0.8333 = 15.0 ; day3: 12.0*0.8333=10.0
|
||
# day4/day5 不动
|
||
assert round(out["close"].iloc[0], 4) == 15.0
|
||
assert round(out["close"].iloc[2], 4) == 10.0
|
||
assert out["close"].iloc[3] == 10.0
|
||
assert out["close"].iloc[4] == 10.0
|
||
|
||
|
||
def test_apply_songzhuangu_halves_pre_event_prices():
|
||
"""10 送 10：除权日前价格减半。"""
|
||
df = _kline([20.0, 20.0, 10.0, 11.0])
|
||
xd = _xdxr_one("2024-01-03", songzhuangu=1.0)
|
||
out = apply_forward_adjust(df, xd)
|
||
# cum-div=day2=20, f=20/(20*2)=0.5 → day1/day2 *= 0.5
|
||
assert out["close"].tolist() == [10.0, 10.0, 10.0, 11.0]
|
||
|
||
|
||
def test_apply_does_not_mutate_input():
|
||
"""apply_forward_adjust 不就地修改输入 DataFrame。"""
|
||
df = _kline([10.0, 10.0, 8.0])
|
||
original = df["close"].tolist()
|
||
xd = _xdxr_one("2024-01-03", fenhong=2.0)
|
||
apply_forward_adjust(df, xd)
|
||
assert df["close"].tolist() == original
|
||
|
||
|
||
# --------------------------------------------------------------------------- #
|
||
# has_bad_prices
|
||
# --------------------------------------------------------------------------- #
|
||
|
||
|
||
def test_has_bad_prices_detects_negative():
|
||
df = pd.DataFrame({"open": [1.0], "high": [1.0], "low": [-0.5], "close": [1.0]})
|
||
assert has_bad_prices(df) is True
|
||
|
||
|
||
def test_has_bad_prices_detects_zero():
|
||
df = pd.DataFrame({"open": [0.0], "high": [1.0], "low": [1.0], "close": [1.0]})
|
||
assert has_bad_prices(df) is True
|
||
|
||
|
||
def test_has_bad_prices_detects_nan():
|
||
df = pd.DataFrame({"open": [1.0], "high": [1.0], "low": [1.0], "close": [float("nan")]})
|
||
assert has_bad_prices(df) is True
|
||
|
||
|
||
def test_has_bad_prices_clean_returns_false():
|
||
df = pd.DataFrame({"open": [1.0], "high": [2.0], "low": [0.5], "close": [1.5]})
|
||
assert has_bad_prices(df) is False
|