如何量化识别股票底部形态（三）

import numpy as np
import pandas as pd
import akshare as ak
def find_v_bottom(df, min_drop=0.15, min_rise=0.15, volume_ratio=2.0, window=10):
    v_bottoms = []
    for i in range(window, len(df) - window):
        current_low = df['最低'].iloc[i]
        current_date = df.index[i]
        # 左侧：window天内跌幅超过min_drop
        left_df = df.iloc[i - window:i]
        left_drop = (left_df['开盘'].iloc[0] - current_low) / left_df['开盘'].iloc[0]
        # 右侧：window天内涨幅超过min_rise
        right_df = df.iloc[i + 1:i + 1 + window]
        if len(right_df) < window:
            continue
        right_rise = (right_df['收盘'].iloc[-1] - current_low) / current_low
        # 条件1：左侧跌幅+右侧涨幅达标
        if (left_drop < min_drop) or (right_rise < min_rise):
            continue
        # 条件2：最低点成交量放大（较前5日平均）
        current_volume = df['成交量'].iloc[i]
        avg_volume_5 = df['成交量'].iloc[i - 5:i].mean()
        if current_volume < avg_volume_5 * volume_ratio:
            continue
        # 条件3：左侧和右侧均线斜率陡峭（5日均线）
        left_df['MA5'] = left_df['收盘'].rolling(5).mean()
        right_df['MA5'] = right_df['收盘'].rolling(5).mean()
        # 左侧MA5斜率（负，且绝对值大）
        left_x = np.arange(len(left_df))
        left_ma5 = left_df['MA5'].dropna()
        if len(left_ma5) < 5:
            continue
        left_slope = np.polyfit(left_x[-len(left_ma5):], left_ma5, 1)[0]
        # 右侧MA5斜率（正，且大）
        right_x = np.arange(len(right_df))
        right_ma5 = right_df['MA5'].dropna()
        if len(right_ma5) < 5:
            continue
        right_slope = np.polyfit(right_x[:len(right_ma5)], right_ma5, 1)[0]
        if (left_slope > -0.5) or (right_slope < 0.5):  # 斜率阈值需根据股票调整
            continue
        v_bottoms.append({
            'V底日期': current_date,
            '最低价': current_low,
            '左侧跌幅': f"{left_drop:.2%}",
            '右侧涨幅': f"{right_rise:.2%}",
            '成交量倍数': current_volume / avg_volume_5
        })
    return v_bottoms
# 获取股票数据（以平安银行000001为例）
def get_stock_data(symbol="000001", start_date="20250101", end_date="20250921"):
    df = ak.stock_zh_a_hist(symbol=symbol, period="daily",
                            start_date=start_date, end_date=end_date, adjust="qfq")
    df['日期'] = pd.to_datetime(df['日期'])
    df.set_index('日期', inplace=True)
    return df[['开盘', '收盘', '最高', '最低', '成交量']]
df_ningde = get_stock_data(symbol="300750", start_date="20220101", end_date="20251231")
v_bottoms = find_v_bottom(df_ningde)
if v_bottoms:
    for vb in v_bottoms:
        print(f"V形底确认：\n日期：{vb['V底日期']}（最低价：{vb['最低价']:.2f}）")
        print(f"左侧跌幅：{vb['左侧跌幅']}，右侧涨幅：{vb['右侧涨幅']}")
        print(f"成交量倍数：{vb['成交量倍数']:.2f}倍\n")
else:
    print("未识别出符合条件的V形底形态")

import numpy as np
import pandas as pd
import akshare as ak
def find_triple_bottom(df, min_distance=15, price_tolerance=0.05, breakout_ratio=0.03):
    # 第一步：找出局部低点
    df['局部低点'] = False
    for i in range(5, len(df) - 5):
        if df['最低'].iloc[i] == df['最低'].iloc[i - 5:i + 5].min():
            df.at[df.index[i], '局部低点'] = True
    local_lows = df[df['局部低点']]
    triple_bottoms = []
    # 第二步：匹配三个连续低点
    for i in range(len(local_lows) - 2):
        low1 = local_lows.iloc[i]
        low2 = local_lows.iloc[i + 1]
        low3 = local_lows.iloc[i + 2]
        # 条件1：三个低点间隔足够
        days_12 = (low2.name - low1.name).days
        days_23 = (low3.name - low2.name).days
        if (days_12 < min_distance) or (days_23 < min_distance):
            continue
        # 条件2：三个低点价格相近
        price_diff_12 = abs(low2['最低'] - low1['最低']) / low1['最低']
        price_diff_23 = abs(low3['最低'] - low2['最低']) / low2['最低']
        if (price_diff_12 > price_tolerance) or (price_diff_23 > price_tolerance):
            continue
        # 条件3：计算两个颈线位（low1-low2之间的高点，low2-low3之间的高点）
        neckline1 = df[(df.index > low1.name) & (df.index < low2.name)]['最高'].max()
        neckline2 = df[(df.index > low2.name) & (df.index < low3.name)]['最高'].max()
        # 颈线位需平行（差异不超过5%）
        neckline_diff = abs(neckline2 - neckline1) / neckline1
        if neckline_diff > 0.05:
            continue
        # 条件4：突破第二个颈线位
        after_low3 = df[df.index > low3.name]
        if len(after_low3) == 0:
            continue
        breakout_close = neckline2 * (1 + breakout_ratio)
        breakout_signals = after_low3[after_low3['收盘'] > breakout_close]
        if len(breakout_signals) > 0:
            triple_bottoms.append({
                '第一个低点': low1.name,
                '第二个低点': low2.name,
                '第三个低点': low3.name,
                '颈线位1': neckline1,
                '颈线位2': neckline2,
                '突破日期': breakout_signals.index[0]
            })
    return triple_bottoms
def get_stock_data(symbol="000001", start_date="20250101", end_date="20250921"):
    df = ak.stock_zh_a_hist(symbol=symbol, period="daily",
                            start_date=start_date, end_date=end_date, adjust="qfq")
    df['日期'] = pd.to_datetime(df['日期'])
    df.set_index('日期', inplace=True)
    return df[['开盘', '收盘', '最高', '最低', '成交量']]
df_byd = get_stock_data(symbol="002594", start_date="20220101", end_date="20250921")
triple_bottoms = find_triple_bottom(df_byd)
if triple_bottoms:
    for tb in triple_bottoms:
        print(
            f"三重底确认：\n三个低点日期：{tb['第一个低点'].date()}、{tb['第二个低点'].date()}、{tb['第三个低点'].date()}")
        print(f"颈线位1：{tb['颈线位1']:.2f}，颈线位2：{tb['颈线位2']:.2f}")
        print(f"突破日期：{tb['突破日期'].date()}\n")
else:
    print("未识别出符合条件的股票")