我想专门使用 Quandl 的 Google Finance 数据库来下载股票价格以对策略进行回测。原因是,与 Quandl 的 WIKI 和 Yahoo 数据库相比,谷歌金融拥有针对拆分调整的股票等的干净数据。如此处所示,最后一个链接将显示调整后的股票拆分:
https://www.quandl.com/WIKI/AAPL-Apple-Inc-AAPL
https://www.quandl.com/YAHOO/AAPL-AAPL-Apple-Inc
https://www.quandl.com/GOOG/NASDAQ_AAPL-Apple-Inc-AAPL
然而,Quandl 的谷歌数据库标签是 GOOG/NYSE_IBM 或 GOOG/NASDAQ_AAPL 的形式,这与 WIKI/IBM、YAHOO/IBM 等标签不同。
由于手动为这些交易所上市的股票数量添加纽约证券交易所或纳斯达克标签是不可行的,有没有一种有效的方法可以从 Quandl 下载股票数据给定 csv 或 pandas 数据框中的股票列表?
这是我的代码 FWIW:
nyseList = pd.read_csv('dowjonesIA.csv') # read csv
masterList = pd.DataFrame(nyseList.Ticker) # save symbols only into another df
for index, rows in masterList.iterrows():
ticker = masterList.loc[index] # this will not work for passing element
stock = Quandl.get(ticker, trim_start="2000-01-01", trim_end="2015-01-01")
#stock = Quandl.get("GOOG/NASDAQ_AAPL", trim_start="2000-01-01", trim_end="2015-01-01") #this is the actual format that works
# lags data for signal
stock['diff'] = (stock.Open - stock.Close.shift(1))/stock.Close.shift(1)
lowerBound = -0.08
upperBound = 0.08
#generate signal based on 8% rule
stock['signal'] = np.where(stock['diff'] >= upperBound, 1.0, np.where (stock['diff'] <= lowerBound, -1.0, 0.0))
initialCapital = 100000.0
accountLimit = 0.05
#calculate size based on account risk and price
stock['position'] = (stock.signal*initialCapital*accountLimit)/stock.Open
#shows if there is a position open
stock['open trade'] = np.where(stock['position'] > 0, 1.0, np.where(stock['position'] < 0, -1.0, 0.0))
#determine profit/loss
stock['pnl'] = (stock.position*stock.Close) - (stock.position*stock.Open)
#sums up results to starting acct capital
stock['equity curve'] = initialCapital + stock.pnl.cumsum()
print(stock.head(20)) # is dataframe
# plots test results
stock['equity curve'].plot()
plt.show()
我曾尝试使用内置于远程数据访问中的 pandas,并且在将字符串作为股票符号传递给 args 时也会出现问题。此外,任何以矢量化方式执行循环的建议都值得赞赏,而不是迭代,以及一般逻辑流程。提前致谢。