update

quant/nlinebreak.py

@@ -3,6 +3,7 @@
 import pandas as pd
 import baostock as bs
 from tqdm import tqdm
+import matplotlib.pyplot as plt
 # %%
 lg = bs.login()
 
@@ -34,3 +35,130 @@
 # %%
 all_stocks.head()
 # %%
+bs.logout()
+# %%
+
+
+def download_data(start_date, end_date, stocks_list):
+    """
+    start_date: start date
+    end_date: end date, must be workday
+    stocks_list: stocks code lists
+    """
+    bs.login()
+    data_df = pd.DataFrame()
+    for code in tqdm(stocks_list):
+        k_rs = bs.query_history_k_data_plus(code,
+                                            'date,code,open,high,low,close,preclose,volume,amount,adjustflag,turn,tradestatus,pctChg,isST',
+                                            start_date, end_date)
+        data_df = data_df.append(k_rs.get_data())
+    bs.logout()
+    data_df['open'] = data_df['open'].astype('float')
+    data_df['high'] = data_df['high'].astype('float')
+    data_df['low'] = data_df['low'].astype('float')
+    data_df['close'] = data_df['close'].astype('float')
+    return data_df
+
+
+# %%
+data = download_data('2020-01-01', '2020-09-14', all_stocks['code'])
+# %%
+data.head()
+# %%
+data = data.merge(all_stocks[['code', 'code_name']],
+                  how='left', left_on='code', right_on='code')
+# %%
+
+
+def nlinebreak(date, close, n):
+    height = []
+    top = []
+    bottom = []
+    color = []
+    datetime = []
+    for i in range(1, len(close)):
+        if not height:
+            if close[i] > close[0]:
+                top.append(close[i])
+                bottom.append(close[0])
+                height.append(top[-1] - bottom[-1])
+                color.append('red')
+                datetime.append(date[i])
+            else:
+                top.append(close[0])
+                bottom.append(close[i])
+                height.append(top[-1] - bottom[-1])
+                color.append('green')
+                datetime.append(date[i])
+        else:
+            if close[i] > max(top[-n:]):
+                bottom.append(top[-1])
+                top.append(close[i])
+                height.append(top[-1] - bottom[-1])
+                color.append('red')
+                datetime.append(date[i])
+            elif close[i] < min(bottom[-n:]):
+                top.append(bottom[-1])
+                bottom.append(close[i])
+                height.append(top[-1] - bottom[-1])
+                color.append('green')
+                datetime.append(date[i])
+    return datetime, height, top, bottom, color
+
+
+# %%
+qzmt = data.loc[data['code_name'] == '钱江摩托']
+# %%
+datetime, height, top, bottom, color = nlinebreak(
+    qzmt['date'].tolist(), qzmt['close'].tolist(), 3)
+# %%
+plt.figure(figsize=(16, 9))
+plt.bar(datetime, height=height, width=1, bottom=bottom, color=color)
+plt.xticks(rotation=90)
+plt.show()
+# %%
+
+
+def find_PositiveStar_NegativeStar(df, stockcode='sh.600000'):
+    """找出启明星和黄昏之星"""
+    # 获取历史K线数据
+    hisdata = df.loc[df['code'] == stockcode]
+    hisdata = hisdata.reset_index()
+    highlist = hisdata['high'].astype('float')
+    lowlist = hisdata['low'].astype('float')
+    openlist = hisdata['open'].astype('float')
+    closelist = hisdata['close'].astype('float')
+    zdflist = hisdata['pctChg'].astype('float')
+    datelist = hisdata['date']
+    for i in range(len(datelist)):
+        if i > 2 and i < len(datelist) - 2:
+            # 启明星 前一日是中阳线，或者连续两天阳线
+            if zdflist[i-1] < -5 or (zdflist[i-1] < -4 and zdflist[i-1] + zdflist[i-2] < -6):
+                # 当日高开或者平开，但是有上下影线,且上影线长度超过了K线实体的长度
+                # or zdflist[i] < 0
+                if (min(openlist[i], closelist[i]) - lowlist[i] > (abs(openlist[i] - closelist[i]))):
+                    # 次日下跌，且跌到之前上涨阳线的一半以下
+                    if zdflist[i + 1] > 0 and closelist[i + 1] > 0.5*(openlist[i - 1] + closelist[i - 1]):
+                        print("at %s，%s出现了启明星，前日涨幅%f，当日涨幅%f，次日涨幅 %f，形态出现后涨幅%f" % (
+                            datelist[i], stockcode, zdflist[i-1], zdflist[i], zdflist[i+1], zdflist[i+2]))
+                        return stockcode, datelist[i], zdflist[i+2]
+    return None, None, None
+
+
+# %%
+find_PositiveStar_NegativeStar(data, 'sz.000913')
+# %%
+codelist, datelist, zdf = [], [], []
+for code in tqdm(all_stocks['code']):
+    c, d, z = find_PositiveStar_NegativeStar(data, code)
+    if c and d and z:
+        codelist.append(c)
+        datelist.append(d)
+        zdf.append(z)
+# %%
+res = pd.DataFrame({'stockcode': codelist, 'date': datelist, 'zdf': zdf})
+# %%
+res['pos'] = res['zdf'].apply(lambda x: 1 if x > 0 else 0)
+# %%
+sum(res['pos']) / res.shape[0]
+# %%