无法通过 python pandas 计算 MACD
声明：本页面是StackOverFlow热门问题的中英对照翻译，遵循CC BY-SA 4.0协议，如果您需要使用它，必须同样遵循CC BY-SA许可，注明原文地址和作者信息，同时你必须将它归于原作者（不是我）：StackOverFlow
原文地址: http://stackoverflow.com/questions/38270524/
Warning: these are provided under cc-by-sa 4.0 license. You are free to use/share it, But you must attribute it to the original authors (not me):
StackOverFlow
cannot calculate MACD via python pandas
提问by Excaliburst
I have made a script to get stock-info for a list of stocks. For the stocks involved (groups in groupby), I need to calculate MACD.
我制作了一个脚本来获取股票列表的股票信息。对于所涉及的股票（groupby 中的组），我需要计算 MACD。
So as to not mix the price for one stock with the other I use a pandas groupby.
为了不将一只股票的价格与另一只股票的价格混在一起，我使用了 Pandas groupby。
# -*- coding: utf-8 -*-
import pandas as pd
from pandas.io.data import DataReader
import numpy as np
import time
from io import StringIO
runstart = time.time() # Start script timer
stocklist = ['nflx','mmm']
tickers = []
days_backtest=102 # MA98 kr?ver 102 d for at virke!
end = pd.Timestamp.utcnow()
start = end - days_backtest * pd.tseries.offsets.BDay()
# Fetch stockinfo
def GetStock(stocklist, start, end, csv_file_all='alltickers_ohlc.csv'):
'''
Fetches stock-info for analysis of each ticker in stocklist
'''
print('\nGetting Stock-info from Yahoo-Finance')
for ticker in stocklist:
r = DataReader(ticker, "yahoo",
start = start, end = end)
# add a symbol column
r['Ticker'] = ticker
tickers.append(r)
# concatenate all the dfs
df_all = pd.concat(tickers)
# add col without space in adj close
df_all['Adj_Close'] = df_all['Adj Close']
#define df with the columns that i need These can be put back in df_all
df_all = df_all[['Ticker','Adj_Close','Volume']] #'Adj Close','Open','High','Low',
# round to 2 dig.
# df_all['Open'] = np.round(df_all['Open'], decimals=2)
# df_all['High'] = np.round(df_all['High'], decimals=2)
# df_all['Low'] = np.round(df_all['Low'], decimals=2)
# df_all['Adj Close'] = np.round(df_all['Adj Close'], decimals=2)
df_all['Adj_Close'] = np.round(df_all['Adj_Close'], decimals=2)
# # Test the first 3 rows of each group for 'Difference' col transgress groups...
# df_all_test = df_all.groupby('Ticker').head(27).reset_index().set_index('Date')
# print ('\n df_all_test (27d summary from df) (Output)\n',df_all_test,'\n')
# saving to a csv #
df_all.reset_index().sort(['Ticker', 'Date'], ascending=[1,1]).set_index('Ticker').to_csv(csv_file_all, date_format='%Y/%m/%d')
# df_all.sort_index(inplace=True) # Sorts rows from date, mingling tickers - not wanted
print('========= Picked up new stockinfo (df_all) \n')
# print ('df_all.tail (Input)\n',df_all.tail(6),'\n')
print(70 * '-')
# print(df_all)
return df_all
def moving_average(group, n=9, type='simple'):
"""
compute an n period moving average.
type is 'simple' | 'exponential'
"""
group = np.asarray(df_['Adj_Close'])
if type == 'simple':
weights = np.ones(n)
else:
weights = np.exp(np.linspace(-1., 0., n))
weights /= weights.sum()
a = np.convolve(group, weights, mode='full')[:len(group)]
a[:n] = a[n]
return a
# return pd.DataFrame({'MCD_Sign':a})
def moving_average_convergence(group, nslow=26, nfast=12):
"""
compute the MACD (Moving Average Convergence/Divergence) using a fast and slow exponential moving avg'
return value is emaslow, emafast, macd which are len(x) arrays
"""
emaslow = moving_average(group, nslow, type='exponential')
emafast = moving_average(group, nfast, type='exponential')
# return emaslow, emafast, emafast - emaslow
return pd.DataFrame({'emaSlw': emaslow,
'emaFst': emafast,
'MACD': emafast - emaslow})
if __name__ == '__main__':
### Getstocks
df_all = GetStock(stocklist, start, end)
### Sort DF
df_all.reset_index().sort(['Ticker', 'Date'], ascending=[1,1]).set_index('Ticker')
### groupby screeener (filtering to only rel ticker group)
df_ = df_all.set_index('Ticker', append=True)
''' Calculating all the KPIs via groupby (filtering pr ticker)'''
grouped = df_.groupby(level=1).Adj_Close
nslow = 26
nfast = 12
nema = 9
df_[['emaSlw', 'emaFst', 'MACD']] = df_.groupby(level=1).Adj_Close.apply(moving_average_convergence)
df_['MCD_Sign'] = df_.groupby(level=1).Adj_Close.apply(moving_average)
print ('(Output df)\n',df_,'\n')
df = df_.reset_index('Ticker')
# Test the last row of each group for new numbers pr group...
df_test = df.groupby('Ticker').tail(1).reset_index().set_index('Date')
print ('df_test (summary from df) (Output)\n',df_test,'\n')
Apparantly I get no results in the columns for all MACD numbers. So somewhere the calculation goes south. I have no clue as to what is going wrong...
显然，我在所有 MACD 数字的列中都没有得到结果。所以在某处计算向南。我不知道出了什么问题......
The output line pr stock-ticker:
输出行 pr 股票行情：
df_test (summary from df) (Output)
Ticker Adj_Close Volume emaSlw emaFst MACD MCD_Sign
Date
2016-07-07 nflx 95.10 9902700 NaN NaN NaN NaN
2016-07-07 mmm 174.87 1842300 NaN NaN NaN NaN
Any of you guys... A tip!?
你们中的任何人......提示！？
回答by Grr
So it looks to me like you did more work than you really needed to here. The answer is a little simpler. You don't need to define your own moving average function, and this is in fact what is causing your problem.
所以在我看来，你做了比你真正需要的更多的工作。答案要简单一些。您不需要定义自己的移动平均函数，这实际上是导致您出现问题的原因。
Change moving_averageto:
更改moving_average为：
def moving_average(group, n=9):
sma = pd.rolling_mean(group, n)
return sma
Change moving_average_converganceto:
更改moving_average_convergance为：
def moving_average_convergence(group, nslow=26, nfast=12):
emaslow = pd.ewma(group, span=nslow, min_periods=1)
emafast = pd.ewma(group, span=nfast, min_periods=1)
result = pd.DataFrame({'MACD': emafast-emaslow, 'emaSlw': emaslow, 'emaFst': emafast})
return result
N.B. I have put 'MACD' first here because despite how you have it listed, the DataFrame reorders the columns alphabetically.
注意，我将“MACD”放在首位，因为不管您如何列出它，DataFrame 都会按字母顺序对列重新排序。
Finally change:
最后改变：
df_[['emaSlw', 'emaFst', 'MACD']] = df_.groupby(level=1).Adj_Close.apply(moving_average_convergence)
to:
到：
df_[['MACD', 'emaSlw', 'emaFst']] = df_.groupby(level=1).Adj_Close.apply(moving_average_convergence)
That should do it.
那应该这样做。
