在 Python 中计算 RSI
Calculating RSI in Python
我正在尝试计算数据帧的 RSI
df = pd.DataFrame({"Close": [100,101,102,103,104,105,106,105,103,102,103,104,103,105,106,107,108,106,105,107,109]})
df["Change"] = df["Close"].diff()
df["Gain"] = np.where(df["Change"]>0,df["Change"],0)
df["Loss"] = np.where(df["Change"]<0,abs(df["Change"]),0 )
df["Index"] = [x for x in range(len(df))]
print(df)
Close Change Gain Loss Index
0 100 NaN 0.0 0.0 0
1 101 1.0 1.0 0.0 1
2 102 1.0 1.0 0.0 2
3 103 1.0 1.0 0.0 3
4 104 1.0 1.0 0.0 4
5 105 1.0 1.0 0.0 5
6 106 1.0 1.0 0.0 6
7 105 -1.0 0.0 1.0 7
8 103 -2.0 0.0 2.0 8
9 102 -1.0 0.0 1.0 9
10 103 1.0 1.0 0.0 10
11 104 1.0 1.0 0.0 11
12 103 -1.0 0.0 1.0 12
13 105 2.0 2.0 0.0 13
14 106 1.0 1.0 0.0 14
15 107 1.0 1.0 0.0 15
16 108 1.0 1.0 0.0 16
17 106 -2.0 0.0 2.0 17
18 105 -1.0 0.0 1.0 18
19 107 2.0 2.0 0.0 19
20 109 2.0 2.0 0.0 20
RSI_length = 7
现在卡在计算中了"Avg Gain"。这里平均收益的逻辑是，索引 6 的第一个平均收益将是 "Gain" 的 RSI_length 周期的平均值。对于连续的 "Avg Gain" 它应该是
(Previous Avg Gain * (RSI_length - 1) + "Gain") / RSI_length
我尝试了以下但没有按预期工作
df["Avg Gain"] = np.nan
df["Avg Gain"] = np.where(df["Index"]==(RSI_length-1),df["Gain"].rolling(window=RSI_length).mean(),\
np.where(df["Index"]>(RSI_length-1),(df["Avg Gain"].iloc[df["Index"]-1]*(RSI_length-1)+df["Gain"]) / RSI_length,np.nan))
这段代码的输出是：
print(df)
Close Change Gain Loss Index Avg Gain
0 100 NaN 0.0 0.0 0 NaN
1 101 1.0 1.0 0.0 1 NaN
2 102 1.0 1.0 0.0 2 NaN
3 103 1.0 1.0 0.0 3 NaN
4 104 1.0 1.0 0.0 4 NaN
5 105 1.0 1.0 0.0 5 NaN
6 106 1.0 1.0 0.0 6 0.857143
7 105 -1.0 0.0 1.0 7 NaN
8 103 -2.0 0.0 2.0 8 NaN
9 102 -1.0 0.0 1.0 9 NaN
10 103 1.0 1.0 0.0 10 NaN
11 104 1.0 1.0 0.0 11 NaN
12 103 -1.0 0.0 1.0 12 NaN
13 105 2.0 2.0 0.0 13 NaN
14 106 1.0 1.0 0.0 14 NaN
15 107 1.0 1.0 0.0 15 NaN
16 108 1.0 1.0 0.0 16 NaN
17 106 -2.0 0.0 2.0 17 NaN
18 105 -1.0 0.0 1.0 18 NaN
19 107 2.0 2.0 0.0 19 NaN
20 109 2.0 2.0 0.0 20 NaN
期望的输出是：
Close Change Gain Loss Index Avg Gain
0 100 NaN 0 0 0 NaN
1 101 1.0 1 0 1 NaN
2 102 1.0 1 0 2 NaN
3 103 1.0 1 0 3 NaN
4 104 1.0 1 0 4 NaN
5 105 1.0 1 0 5 NaN
6 106 1.0 1 0 6 0.857143
7 105 -1.0 0 1 7 0.734694
8 103 -2.0 0 2 8 0.629738
9 102 -1.0 0 1 9 0.539775
10 103 1.0 1 0 10 0.605522
11 104 1.0 1 0 11 0.661876
12 103 -1.0 0 1 12 0.567322
13 105 2.0 2 0 13 0.771990
14 106 1.0 1 0 14 0.804563
15 107 1.0 1 0 15 0.832483
16 108 1.0 1 0 16 0.856414
17 106 -2.0 0 2 17 0.734069
18 105 -1.0 0 1 18 0.629202
19 107 2.0 2 0 19 0.825030
20 109 2.0 2 0 20 0.992883
（已编辑）
这是您的公式的一个实现。
RSI_LENGTH = 7
rolling_gain = df["Gain"].rolling(RSI_LENGTH).mean()
df.loc[RSI_LENGTH-1, "RSI"] = rolling_gain[RSI_LENGTH-1]
for inx in range(RSI_LENGTH, len(df)):
df.loc[inx, "RSI"] = (df.loc[inx-1, "RSI"] * (RSI_LENGTH -1) + df.loc[inx, "Gain"]) / RSI_LENGTH
结果是：
Close Change Gain Loss Index RSI
0 100 NaN 0.0 0.0 0 NaN
1 101 1.0 1.0 0.0 1 NaN
2 102 1.0 1.0 0.0 2 NaN
3 103 1.0 1.0 0.0 3 NaN
4 104 1.0 1.0 0.0 4 NaN
5 105 1.0 1.0 0.0 5 NaN
6 106 1.0 1.0 0.0 6 0.857143
7 105 -1.0 0.0 1.0 7 0.734694
8 103 -2.0 0.0 2.0 8 0.629738
9 102 -1.0 0.0 1.0 9 0.539775
10 103 1.0 1.0 0.0 10 0.605522
11 104 1.0 1.0 0.0 11 0.661876
12 103 -1.0 0.0 1.0 12 0.567322
13 105 2.0 2.0 0.0 13 0.771990
14 106 1.0 1.0 0.0 14 0.804563
15 107 1.0 1.0 0.0 15 0.832483
16 108 1.0 1.0 0.0 16 0.856414
17 106 -2.0 0.0 2.0 17 0.734069
18 105 -1.0 0.0 1.0 18 0.629202
19 107 2.0 2.0 0.0 19 0.825030
20 109 2.0 2.0 0.0 20 0.992883
我正在尝试计算数据帧的 RSI
df = pd.DataFrame({"Close": [100,101,102,103,104,105,106,105,103,102,103,104,103,105,106,107,108,106,105,107,109]})
df["Change"] = df["Close"].diff()
df["Gain"] = np.where(df["Change"]>0,df["Change"],0)
df["Loss"] = np.where(df["Change"]<0,abs(df["Change"]),0 )
df["Index"] = [x for x in range(len(df))]
print(df)
Close Change Gain Loss Index
0 100 NaN 0.0 0.0 0
1 101 1.0 1.0 0.0 1
2 102 1.0 1.0 0.0 2
3 103 1.0 1.0 0.0 3
4 104 1.0 1.0 0.0 4
5 105 1.0 1.0 0.0 5
6 106 1.0 1.0 0.0 6
7 105 -1.0 0.0 1.0 7
8 103 -2.0 0.0 2.0 8
9 102 -1.0 0.0 1.0 9
10 103 1.0 1.0 0.0 10
11 104 1.0 1.0 0.0 11
12 103 -1.0 0.0 1.0 12
13 105 2.0 2.0 0.0 13
14 106 1.0 1.0 0.0 14
15 107 1.0 1.0 0.0 15
16 108 1.0 1.0 0.0 16
17 106 -2.0 0.0 2.0 17
18 105 -1.0 0.0 1.0 18
19 107 2.0 2.0 0.0 19
20 109 2.0 2.0 0.0 20
RSI_length = 7
现在卡在计算中了"Avg Gain"。这里平均收益的逻辑是，索引 6 的第一个平均收益将是 "Gain" 的 RSI_length 周期的平均值。对于连续的 "Avg Gain" 它应该是
(Previous Avg Gain * (RSI_length - 1) + "Gain") / RSI_length
我尝试了以下但没有按预期工作
df["Avg Gain"] = np.nan
df["Avg Gain"] = np.where(df["Index"]==(RSI_length-1),df["Gain"].rolling(window=RSI_length).mean(),\
np.where(df["Index"]>(RSI_length-1),(df["Avg Gain"].iloc[df["Index"]-1]*(RSI_length-1)+df["Gain"]) / RSI_length,np.nan))
这段代码的输出是：
print(df)
Close Change Gain Loss Index Avg Gain
0 100 NaN 0.0 0.0 0 NaN
1 101 1.0 1.0 0.0 1 NaN
2 102 1.0 1.0 0.0 2 NaN
3 103 1.0 1.0 0.0 3 NaN
4 104 1.0 1.0 0.0 4 NaN
5 105 1.0 1.0 0.0 5 NaN
6 106 1.0 1.0 0.0 6 0.857143
7 105 -1.0 0.0 1.0 7 NaN
8 103 -2.0 0.0 2.0 8 NaN
9 102 -1.0 0.0 1.0 9 NaN
10 103 1.0 1.0 0.0 10 NaN
11 104 1.0 1.0 0.0 11 NaN
12 103 -1.0 0.0 1.0 12 NaN
13 105 2.0 2.0 0.0 13 NaN
14 106 1.0 1.0 0.0 14 NaN
15 107 1.0 1.0 0.0 15 NaN
16 108 1.0 1.0 0.0 16 NaN
17 106 -2.0 0.0 2.0 17 NaN
18 105 -1.0 0.0 1.0 18 NaN
19 107 2.0 2.0 0.0 19 NaN
20 109 2.0 2.0 0.0 20 NaN
期望的输出是：
Close Change Gain Loss Index Avg Gain
0 100 NaN 0 0 0 NaN
1 101 1.0 1 0 1 NaN
2 102 1.0 1 0 2 NaN
3 103 1.0 1 0 3 NaN
4 104 1.0 1 0 4 NaN
5 105 1.0 1 0 5 NaN
6 106 1.0 1 0 6 0.857143
7 105 -1.0 0 1 7 0.734694
8 103 -2.0 0 2 8 0.629738
9 102 -1.0 0 1 9 0.539775
10 103 1.0 1 0 10 0.605522
11 104 1.0 1 0 11 0.661876
12 103 -1.0 0 1 12 0.567322
13 105 2.0 2 0 13 0.771990
14 106 1.0 1 0 14 0.804563
15 107 1.0 1 0 15 0.832483
16 108 1.0 1 0 16 0.856414
17 106 -2.0 0 2 17 0.734069
18 105 -1.0 0 1 18 0.629202
19 107 2.0 2 0 19 0.825030
20 109 2.0 2 0 20 0.992883
（已编辑）
这是您的公式的一个实现。
RSI_LENGTH = 7
rolling_gain = df["Gain"].rolling(RSI_LENGTH).mean()
df.loc[RSI_LENGTH-1, "RSI"] = rolling_gain[RSI_LENGTH-1]
for inx in range(RSI_LENGTH, len(df)):
df.loc[inx, "RSI"] = (df.loc[inx-1, "RSI"] * (RSI_LENGTH -1) + df.loc[inx, "Gain"]) / RSI_LENGTH
结果是：
Close Change Gain Loss Index RSI
0 100 NaN 0.0 0.0 0 NaN
1 101 1.0 1.0 0.0 1 NaN
2 102 1.0 1.0 0.0 2 NaN
3 103 1.0 1.0 0.0 3 NaN
4 104 1.0 1.0 0.0 4 NaN
5 105 1.0 1.0 0.0 5 NaN
6 106 1.0 1.0 0.0 6 0.857143
7 105 -1.0 0.0 1.0 7 0.734694
8 103 -2.0 0.0 2.0 8 0.629738
9 102 -1.0 0.0 1.0 9 0.539775
10 103 1.0 1.0 0.0 10 0.605522
11 104 1.0 1.0 0.0 11 0.661876
12 103 -1.0 0.0 1.0 12 0.567322
13 105 2.0 2.0 0.0 13 0.771990
14 106 1.0 1.0 0.0 14 0.804563
15 107 1.0 1.0 0.0 15 0.832483
16 108 1.0 1.0 0.0 16 0.856414
17 106 -2.0 0.0 2.0 17 0.734069
18 105 -1.0 0.0 1.0 18 0.629202
19 107 2.0 2.0 0.0 19 0.825030
20 109 2.0 2.0 0.0 20 0.992883