06. Numpy-数组运算

# 先把简单收益率转为增长因子，再沿axis=1方向连乘并减1
cumulative_returns = (1 + returns).cumprod(axis=1) - 1
print(f'复合累积收益率:\n{cumulative_returns}')
# 例如股票A第3列 = 1.05 × 1.03 × 0.98 - 1 = 0.05987

import numpy as np

# 5个交易日的股价数据
prices = np.array([10, 20, 30, 40, 50])

# 加法：价格调整
adjustment = np.array([1, 2, 3, 4, 5])
new_prices = prices + adjustment
print(f'调整后价格: {new_prices}')

# 减法：相邻日价差
price_diff = prices[1:] - prices[:-1]
print(f'价格变化: {price_diff}')

import numpy as np

# 中国建筑8月股价数据（18个交易日）
stock_p = np.array([4.98, 5.02, 4.95, 4.91, 4.98, 4.92, 4.88, 4.92,
                    4.88, 4.82, 4.85, 4.89, 4.91, 4.86, 4.84, 4.92,
                    5.01, 5.04])

# 计算日收益率
daily_returns = (stock_p[1:] - stock_p[:-1]) / stock_p[:-1]

# 基本统计量
mean_return = daily_returns.mean()
std_return = daily_returns.std()
# 年化波动率 = 日波动率 × √252
annualized_vol = std_return * np.sqrt(252)

print('=' * 50)
print('中国建筑8月风险收益分析')
print('=' * 50)
print(f'日平均收益率: {mean_return:.4%}')
print(f'日收益率标准差: {std_return:.4%}')
print(f'年化波动率: {annualized_vol:.4%}')

# 价格统计
max_price = stock_p.max()
min_price = stock_p.min()
price_range = max_price - min_price
print(f'最高价: {max_price:.2f}元')
print(f'最低价: {min_price:.2f}元')
print(f'价格区间: {price_range:.2f}元')

# 下行风险：只考虑下跌日
downside_returns = daily_returns[daily_returns < 0]
downside_risk = downside_returns.std() * np.sqrt(252)
print(f'下行波动率: {downside_risk:.4%}')

# 夏普比率（假设无风险利率3%）
sharpe_ratio = (mean_return * 252 - 0.03) / annualized_vol
print(f'夏普比率: {sharpe_ratio:.2f}')

import numpy as np

# 两个投资组合的持仓
portfolio_a = np.array(['600519.SH', '000858.SZ', '600036.SH', '000002.SZ'])
portfolio_b = np.array(['600036.SH', '601318.SH', '000858.SZ', '600000.SH'])

# 交集：共同持仓
common = np.intersect1d(portfolio_a, portfolio_b)
print(f'共同持仓: {common}')

# 并集：所有股票（去重）
all_stocks = np.union1d(portfolio_a, portfolio_b)
print(f'所有股票: {all_stocks}')

# 差集：独有持仓
unique_a = np.setdiff1d(portfolio_a, portfolio_b)
unique_b = np.setdiff1d(portfolio_b, portfolio_a)
print(f'A组合独有: {unique_a}')
print(f'B组合独有: {unique_b}')

import numpy as np

# 5只股票的价格
prices = np.array([10.5, 20.3, 15.8, 25.2, 18.6])

# 比较运算 → 返回布尔数组
above_20 = prices > 20
print(f'价格>20元: {above_20}')

# 布尔索引：提取满足条件的元素
expensive_stocks = prices[prices > 20]
print(f'高价股票: {expensive_stocks}')

import numpy as np

# 6个交易日的收益率
returns = np.array([0.05, -0.02, 0.03, 0.07, -0.01, 0.04])

# np.sort()：返回排序后的新数组
sorted_returns = np.sort(returns)
print(f'排序后: {sorted_returns}')

# np.argsort()：返回排序索引
indices = np.argsort(returns)
print(f'排序索引: {indices}')
print(f'按索引重新计算: {returns[indices]}')

# 搜索极值位置
max_idx = np.argmax(returns)
min_idx = np.argmin(returns)
print(f'最大值位置: {max_idx}, 值: {returns[max_idx]}')
print(f'最小值位置: {min_idx}, 值: {returns[min_idx]}')

# np.where：条件替换（三目运算）
adjusted = np.where(returns > 0, returns, 0)
print(f'负收益调整为0: {adjusted}')

import numpy as np
import time

n = 1000000
arr1 = np.random.randn(n)
arr2 = np.random.randn(n)

# Python循环
start = time.time()
result_slow = []
for i in range(n):
    result_slow.append(arr1[i] + arr2[i])
slow_time = time.time() - start

# NumPy向量化
start = time.time()
result_fast = arr1 + arr2
fast_time = time.time() - start

print(f'循环时间: {slow_time:.4f}秒')
print(f'向量化时间: {fast_time:.4f}秒')
print(f'加速比: {slow_time/fast_time:.1f}倍')