# 生成过去一年的工作日（不含周末）日期序列
try:
# 使用上海时区作为本地时间
today = pd.Timestamp.now(tz='Asia/Shanghai').normalize()
except Exception:
# 回退到本地无时区的今天
today = pd.Timestamp.today().normalize()
start = today - pd.offsets.DateOffset(years=1)
dates = pd.bdate_range(start=start, end=today) # B 表示工作日频率
len(dates), dates.min(), dates.max()(261,
Timestamp('2024-10-04 00:00:00+0800', tz='Asia/Shanghai'),
Timestamp('2025-10-03 00:00:00+0800', tz='Asia/Shanghai'))# 定义 GBM 模拟函数
def simulate_gbm(S0: float, mu: float, sigma: float, n_steps: int, dt: float = 1.0/252, seed: int | None = None):
if seed is not None:
rng = np.random.default_rng(seed)
z = rng.standard_normal(n_steps)
else:
z = np.random.normal(size=n_steps)
dr = (mu - 0.5 * sigma**2) * dt + sigma * np.sqrt(dt) * z
path = S0 * np.exp(np.cumsum(dr))
return path
n = len(dates)
# 纳指参数（示例）
nasdaq_S0 = 16000.0
nasdaq_mu = 0.10
nasdaq_sigma = 0.25
# 标普500参数（示例）
spx_S0 = 5600.0
spx_mu = 0.08
spx_sigma = 0.20
nasdaq_prices = simulate_gbm(nasdaq_S0, nasdaq_mu, nasdaq_sigma, n)
spx_prices = simulate_gbm(spx_S0, spx_mu, spx_sigma, n)
df = pd.DataFrame({
'NASDAQ_Close': nasdaq_prices,
'SPX_Close': spx_prices,
}, index=dates)
# 标准化至100（首日=100）
df_norm = df / df.iloc[0] * 100.0
df.head()| NASDAQ_Close | SPX_Close | |
|---|---|---|
| 2024-10-04 00:00:00+08:00 | 15936.279209 | 5615.695306 |
| 2024-10-07 00:00:00+08:00 | 15966.970430 | 5715.919353 |
| 2024-10-08 00:00:00+08:00 | 16034.465514 | 5755.178875 |
| 2024-10-09 00:00:00+08:00 | 15806.840474 | 5821.073011 |
| 2024-10-10 00:00:00+08:00 | 15953.006046 | 5737.760714 |
# 图1：标准化至100的相对表现
plt.figure(figsize=(12, 5))
df_norm['NASDAQ_Close'].plot(label='纳斯达克（归一化）', lw=1.6, color='tab:blue')
df_norm['SPX_Close'].plot(label='标普500（归一化）', lw=1.6, color='tab:orange')
plt.title('纳指100 vs 标普500 - 伪行情（过去一年，相对表现=100）', fontname='Microsoft YaHei', fontsize=14)
plt.xlabel('日期', fontname='Microsoft YaHei')
plt.ylabel('相对指数（首日=100）', fontname='Microsoft YaHei')
plt.grid(True, ls='--', alpha=0.4)
plt.legend(prop={'family':'Microsoft YaHei'})
plt.tight_layout()
plt.show()
# 图2：原始指数水平对比（双轴可读性更好）
fig, ax1 = plt.subplots(figsize=(12, 5))
color1 = 'tab:blue'
color2 = 'tab:orange'
l1 = ax1.plot(df.index, df['NASDAQ_Close'], color=color1, lw=1.6, label='纳斯达克')
ax1.set_xlabel('日期', fontname='Microsoft YaHei')
ax1.set_ylabel('纳指水平', color=color1, fontname='Microsoft YaHei')
ax1.tick_params(axis='y', labelcolor=color1)
ax1.grid(True, ls='--', alpha=0.4)
ax2 = ax1.twinx()
l2 = ax2.plot(df.index, df['SPX_Close'], color=color2, lw=1.6, label='标普500')
ax2.set_ylabel('标普水平', color=color2, fontname='Microsoft YaHei')
ax2.tick_params(axis='y', labelcolor=color2)
lines = l1 + l2
labels = [l.get_label() for l in lines]
plt.title('纳指 vs 标普500 - 伪行情（过去一年，原始水平）', fontname='Microsoft YaHei', fontsize=14)
ax1.legend(lines, labels, loc='upper left', prop={'family':'Microsoft YaHei'})
fig.tight_layout()
plt.show()