Techical Analysis of Stock & Commodities
Z-score: How to use it in Trading

Z-score 计算公式
Z = (X - μ) / σ
μ 均值
σ 标准差，也称波动率

假设 X 服从正态分布 
3个标准差（σ）原则，或者68-95-99.7原则

pandas 滚动计算  Z-score

def rolling_zscore(s, win=20):
    ma = s.rolling(window=win).mean()
    std = s.rolling(window=win).std()
    return (s - ma)/std

rolling_mean = close_prices.rolling(window=period).mean()
rolling_std = close_prices.rolling(window=period).std()
z_scores = (close_prices - rolling_mean) / rolling_std


计算最近 250 天的 zscore， 买点 zscore 小于-2  ，卖点 zscore 大于 2 
布林通道 上下轨  均值 ± 2个标准差

z-score 因子的有效性来源于正态分布的假设
然而，股价的波动并不符合正态分布





import pandas as pd
import matplotlib.pyplot as plt

# 常量
Z_THRESH = 2
PERIODS = [30, 60, 90]
TICKER_SYMBOL = "hs300"
START_DATE = '2020-1-1'
# END_DATE 暂不使用
END_DATE = '2023-12-25'


def fetch_data(ticker_symbol, start_date, end_date):
    """Fetches historical data for a given ticker symbol."""
    # ticker_data = yf.Ticker(ticker_symbol)
    # 不使用 yfinance ，读取本地文件
    # /demo/myweb2020/docs/Q/data/hs300
    ticker_data = pd.read_csv('/Users/dugang/tmp/hs300.txt')
    # return ticker_data.history(period='1d', start=start_date, end=end_date)
    ticker_data = ticker_data[ticker_data['date']>=start_date]
    return ticker_data

def calculate_z_scores(close_prices, periods):
    """Calculates Z-scores for given periods."""
    z_scores_dict = {}
    for period in periods:
        # 计算给定周期的滚动平均值
        rolling_mean = close_prices.rolling(window=period).mean()      
        # 计算给定周期的滚动标准差
        rolling_std = close_prices.rolling(window=period).std()       
        # 计算收盘价的Z值
        z_scores = (close_prices - rolling_mean) / rolling_std      
        # 将Z值存储在以周期为关键字的字典中
        z_scores_dict[period] = z_scores
    return z_scores_dict

def plot_data(close_prices, z_scores_data):
    """Plots close prices and z-scores."""   
    
    # 为收盘价和Z值创建子图
    fig, (ax1, ax2) = plt.subplots(2, sharex=True, figsize=(20, 8))   
    
    # 在第一个子图上绘制收盘价
    ax1.plot(close_prices.index, close_prices, label='Close Prices')
    for period, z_scores in z_scores_data.items():
        # 在第二个子图上绘制每个时期的Z值
        ax2.plot(z_scores.index, z_scores, label=f'Z-Scores {period} days', alpha=0.7)       
        # 如果周期是列表中的第一个，则在第一个子图上绘制买入/卖出信号
        if period == PERIODS[0]:
            buy_signals = (z_scores < -Z_THRESH)
            sell_signals = (z_scores > Z_THRESH)
            ax1.plot(close_prices[buy_signals].index, close_prices[buy_signals], 'o', color='g', label='Buy Signal')
            ax1.plot(close_prices[sell_signals].index, close_prices[sell_signals], 'o', color='r', label='Sell Signal')
    # 为收盘价子图设置y标签和图例
    ax1.set_ylabel('Close Prices')
    ax1.legend(loc="upper left")
    ax1.grid(True)
    
    # 在Z值子图上绘制表示Z值阈值的水平线
    ax2.axhline(-Z_THRESH, color='red', linestyle='--')
    ax2.axhline(Z_THRESH, color='red', linestyle='--')   
    # 设置Z值子图的Y标签和图例
    ax2.set_ylabel('Z-Scores')
    ax2.legend(loc="upper left")
    ax2.grid(True)

    # 为整个绘图设置主标题
    plt.suptitle(f'{TICKER_SYMBOL} Close Prices and Z-Scores {Z_THRESH} Treshold')
    # 显示图表
    plt.show()

# 获取股票代码的历史数据
ticker_data = fetch_data(TICKER_SYMBOL, START_DATE, END_DATE)

# 计算指定时期的Z值
z_scores_data = calculate_z_scores(ticker_data['close'], PERIODS)

# 绘制收盘价和Z值
plot_data(ticker_data['close'], z_scores_data)



日线数据接口 由歪枣网提供 
http://waizaowang.com/api/detail/1007

date,open,close,high,low,a,v
2005-01-04,994.77,982.79,994.77,980.66,7412870.0,4431980032
2005-01-05,981.58,992.56,997.32,979.88,7119110.0,4529209856
2005-01-06,993.33,983.17,993.79,980.33,6288030.0,3921019904

沪深300
http://api.waizaowang.com/doc/getIndexDayKLine?code=000300&ktype=101&startDate=1990-01-01&endDate=2100-01-01&fields=tdate,open,close,high,low,cjl,cje&export=0&token=xxx

token 注册账号 登录后获取