Day19 趨勢、位階、型態（３）

2023 iThome 鐵人賽

DAY 19

AI & Data

用Python程式進行股票技術分析系列第 19 篇

15th鐵人賽

不明

2023-09-19 05:00:15

617 瀏覽

分享至

經過昨日Day18的「由多翻空的過程」的程式實作，應該發現技術分析的主觀部份真的很依賴人的視覺以及經驗；跟著股票老師學這個課題的時候也是這樣，股票老師會給案例（兩張圖，多頭跟空頭；請參考我在6月2日的股票投資筆記）；在老師說明完講義上的範例之後就得自行在案例上繪圖練習，因此沒有太多的文字描述可供參照。今天會接續著說明「由空翻多的過程」。

由空翻多的過程

起漲→回檔/盤整→續漲→止漲→盤頭→做頭→起跌→短線翻空→中線翻空→長線翻空

程式實作：人工方式標註空翻多過程，以瞭解多空之間的各種現象及其特徵

以下將會介紹過程中各階段的特徵並進行程式實作，使用Day1所提的「主觀性質客觀化」的方式一（將在看盤軟體繪製的圖形與線段以人工方式轉換成數據）來將視覺觀察的現象轉換為客觀數據。

一、多頭確立

空頭落底後翻多的兩個工具：

道氏理論：N字型的現象低點越來越高，高點越來越高
移動平均線理論 20日線站上且20日線上揚
程式實作如下：

### 人工方式標示多頭確立 ###
# 方法一：道氏理論 N字型的現象：低點越來越高，高點越來越高
bulls_dow_points_len = len(np.array(df_k_line['Close']))
bulls_dow_points = np.array([np.nan]*bulls_dow_points_len)
idx = date_to_index(df_k_line,'2019-08-06')
bulls_dow_points[idx] = df_k_line.iloc[idx]['Open']
idx = date_to_index(df_k_line,'2019-08-26')
bulls_dow_points[idx] = df_k_line.iloc[idx]['Close']
idx = date_to_index(df_k_line,'2019-08-08')
bulls_dow_points[idx] = df_k_line.iloc[idx]['Close']
idx = date_to_index(df_k_line,'2019-08-22')
bulls_dow_points[idx] = df_k_line.iloc[idx]['Open']
# 方法二：移動平均線理論 20日線站上且20日線上揚
bulls_ma20_points_len = len(np.array(df_k_line['Close']))
bulls_ma20_points = np.array([np.nan]*bulls_ma20_points_len)
idx = date_to_index(df_k_line,'2019-08-30')
bulls_ma20_points[idx] = df_k_line.iloc[idx]['Open']

程式執行結果如下：
Imgur

二、多頭起漲

起漲就是最低的三個點位

第一個：波段最低點
第二個：上漲回檔再續漲
第三個：出現迷你底型或盤整區間突破
其中第一個低點只能後來驗證，當下並不知道。接下來用人工方式標註三個點的數據，程式碼如下所示：

### 人工方式標示多頭起漲 ###
start_rise_points_len = len(np.array(df_k_line['Close']))
start_rise_points = np.array([np.nan]*start_rise_points_len)
# 起漲就是最低的三個點位：
# 第一個：最低點 
idx = date_to_index(df_k_line,'2019-08-06')
start_rise_points[idx] = df_k_line.iloc[idx]['Close']
# 第二個：上漲回檔再續漲
idx = date_to_index(df_k_line,'2019-08-30')
start_rise_points[idx] = df_k_line.iloc[idx]['Close']
# 第三個：出現迷你底型或盤整區間突破
idx = date_to_index(df_k_line,'2019-09-04')
start_rise_points[idx] = df_k_line.iloc[idx]['Close']

程式執行結果如下：
Imgur

三、多頭回檔

多頭格局中的拉回叫做回檔，程式碼如下所示：

### 多頭回檔 ###

range_lines = []
# 範圍一
corner1 = ('2019-08-22',df_k_line.loc['2019-08-22']['Open'])
corner2 = ('2019-08-28',df_k_line.loc['2019-08-22']['Open'])
corner3 = ('2019-08-22',df_k_line.loc['2019-08-27']['Close'])
corner4 = ('2019-08-28',df_k_line.loc['2019-08-27']['Close'])
range_lines.append([corner1,corner2])
range_lines.append([corner1,corner3])
range_lines.append([corner3,corner4])
range_lines.append([corner4,corner2])

# 範圍二
corner1 = ('2019-11-06',df_k_line.loc['2019-11-06']['Open'])
corner2 = ('2019-11-15',df_k_line.loc['2019-11-06']['Open'])
corner3 = ('2019-11-06',df_k_line.loc['2019-11-11']['Close'])
corner4 = ('2019-11-15',df_k_line.loc['2019-11-11']['Close'])
range_lines.append([corner1,corner2])
range_lines.append([corner1,corner3])
range_lines.append([corner3,corner4])
range_lines.append([corner4,corner2])


# 範圍三
corner1 = ('2019-11-27',df_k_line.loc['2019-11-27']['Close'])
corner2 = ('2019-12-04',df_k_line.loc['2019-11-27']['Close'])
corner3 = ('2019-11-27',df_k_line.loc['2019-12-03']['Open'])
corner4 = ('2019-12-04',df_k_line.loc['2019-12-03']['Open'])
range_lines.append([corner1,corner2])
range_lines.append([corner1,corner3])
range_lines.append([corner3,corner4])
range_lines.append([corner4,corner2])

# 範圍四
corner1 = ('2020-01-02',df_k_line.loc['2020-01-02']['Close'])
corner2 = ('2020-01-09',df_k_line.loc['2020-01-02']['Close'])
corner3 = ('2020-01-02',df_k_line.loc['2020-01-09']['Open'])
corner4 = ('2020-01-09',df_k_line.loc['2020-01-09']['Open'])
range_lines.append([corner1,corner2])
range_lines.append([corner1,corner3])
range_lines.append([corner3,corner4])
range_lines.append([corner4,corner2])

程式執行結果如下：
Imgur

四、多頭續漲

程式碼如下所示：

### 多頭續漲 ###

range_lines = []
# 範圍一
corner1 = ('2019-08-15',df_k_line.loc['2019-08-15']['Close'])
corner2 = ('2019-08-22',df_k_line.loc['2019-08-15']['Close'])
corner3 = ('2019-08-15',df_k_line.loc['2019-08-22']['Open'])
corner4 = ('2019-08-22',df_k_line.loc['2019-08-22']['Open'])
range_lines.append([corner1,corner2])
range_lines.append([corner1,corner3])
range_lines.append([corner3,corner4])
range_lines.append([corner4,corner2])

# 範圍二
corner1 = ('2019-08-26',df_k_line.loc['2019-08-26']['Close'])
corner2 = ('2019-09-19',df_k_line.loc['2019-08-26']['Close'])
corner3 = ('2019-08-26',df_k_line.loc['2019-09-19']['Open'])
corner4 = ('2019-09-19',df_k_line.loc['2019-09-19']['Open'])
range_lines.append([corner1,corner2])
range_lines.append([corner1,corner3])
range_lines.append([corner3,corner4])
range_lines.append([corner4,corner2])

# 範圍三
corner1 = ('2019-10-03',df_k_line.loc['2019-10-03']['Open'])
corner2 = ('2019-11-06',df_k_line.loc['2019-10-03']['Open'])
corner3 = ('2019-10-03',df_k_line.loc['2019-11-06']['Close'])
corner4 = ('2019-11-06',df_k_line.loc['2019-11-06']['Close'])
range_lines.append([corner1,corner2])
range_lines.append([corner1,corner3])
range_lines.append([corner3,corner4])
range_lines.append([corner4,corner2])

# 範圍四
corner1 = ('2019-11-13',df_k_line.loc['2019-11-14']['Close'])
corner2 = ('2019-11-20',df_k_line.loc['2019-11-14']['Close'])
corner3 = ('2019-11-13',df_k_line.loc['2019-11-19']['Close'])
corner4 = ('2019-11-20',df_k_line.loc['2019-11-19']['Close'])
range_lines.append([corner1,corner2])
range_lines.append([corner1,corner3])
range_lines.append([corner3,corner4])
range_lines.append([corner4,corner2])

# 範圍五
corner1 = ('2019-12-03',df_k_line.loc['2019-12-03']['Open'])
corner2 = ('2019-12-18',df_k_line.loc['2019-12-03']['Open'])
corner3 = ('2019-12-03',df_k_line.loc['2019-12-18']['Close'])
corner4 = ('2019-12-18',df_k_line.loc['2019-12-18']['Close'])
range_lines.append([corner1,corner2])
range_lines.append([corner1,corner3])
range_lines.append([corner3,corner4])
range_lines.append([corner4,corner2])

程式執行結果如下：
Imgur

五、多頭盤整

程式碼如下所示：

### 多頭盤整 ###

range_lines = []
# 範圍一
corner1 = ('2019-09-12',df_k_line.loc['2019-10-01']['Close'])
corner2 = ('2019-10-08',df_k_line.loc['2019-10-01']['Close'])
corner3 = ('2019-09-12',df_k_line.loc['2019-09-16']['Open'])
corner4 = ('2019-10-08',df_k_line.loc['2019-09-16']['Open'])
range_lines.append([corner1,corner2])
range_lines.append([corner1,corner3])
range_lines.append([corner3,corner4])
range_lines.append([corner4,corner2])

# 範圍二
corner1 = ('2019-11-05',df_k_line.loc['2019-12-09']['Close'])
corner2 = ('2019-12-11',df_k_line.loc['2019-12-09']['Close'])
corner3 = ('2019-11-05',df_k_line.loc['2019-11-11']['Close'])
corner4 = ('2019-12-11',df_k_line.loc['2019-11-11']['Close'])
range_lines.append([corner1,corner2])
range_lines.append([corner1,corner3])
range_lines.append([corner3,corner4])
range_lines.append([corner4,corner2])

# 範圍三
corner1 = ('2019-12-13',df_k_line.loc['2020-01-03']['Open'])
corner2 = ('2020-01-30',df_k_line.loc['2020-01-03']['Open'])
corner3 = ('2019-12-13',df_k_line.loc['2020-01-07']['Close'])
corner4 = ('2020-01-30',df_k_line.loc['2020-01-07']['Close'])
range_lines.append([corner1,corner2])
range_lines.append([corner1,corner3])
range_lines.append([corner3,corner4])
range_lines.append([corner4,corner2])

程式執行結果如下：
Imgur

六、多頭止漲

漲到最高點就是多頭止漲，程式碼如下所示：

### 人工方式標示多頭止漲 ###
stop_rise_points_len = len(np.array(df_k_line['Close']))
stop_rise_points = np.array([np.nan]*stop_rise_points_len)

# 最高點
idx = date_to_index(df_k_line,'2020-01-03')
stop_rise_points[idx] = df_k_line.iloc[idx]['Open']

程式執行結果如下：
Imgur

七、多頭盤頭與空頭做頭

盤頭跌破就是做頭，程式碼如下所示：

### 多頭盤頭 與 空頭做頭 ####

range_lines = []
# 頭部範圍
corner1 = ('2019-12-13',df_k_line.loc['2020-01-03']['Open'])
corner2 = ('2020-01-30',df_k_line.loc['2020-01-03']['Open'])
corner3 = ('2019-12-13',df_k_line.loc['2020-01-07']['Close'])
corner4 = ('2020-01-30',df_k_line.loc['2020-01-07']['Close'])
range_lines.append([corner1,corner2])
range_lines.append([corner1,corner3])
range_lines.append([corner3,corner4])
range_lines.append([corner4,corner2])

# 跌破點
breakthrough_points_len = len(np.array(df_k_line['Close']))
breakthrough_points = np.array([np.nan]*breakthrough_points_len)
breakthrough_points[date_to_index(df_k_line,'2020-01-30')] = df_k_line.loc['2020-01-07']['Close']

程式執行結果如下：
Imgur
接下來又循環地回到昨天Day18的「空頭起跌」。
完整的程式碼請參照「第十九天：趨勢、位階、型態（３）.ipynb」。