Hands-On Machine Learning for Algorithmic Trading

How to use the VAR model for macro fundamentals forecasts

We will extend the univariate example of a single time series of monthly data on industrial production and add a monthly time series on consumer sentiment, both provided by the Federal Reserve's data service. We will use the familiar pandas-datareader library to retrieve data from 1970 through 2017:

df = web.DataReader(['UMCSENT', 'IPGMFN'], 'fred', '1970', '2017-12').dropna()df.columns = ['sentiment', 'ip']

Log-transforming the industrial production series and seasonal differencing using lag 12 of both series yields stationary results:

df_transformed = pd.DataFrame({'ip': np.log(df.ip).diff(12),                              'sentiment': df.sentiment.diff(12)}).dropna()test_unit_root(df_transformed) # see notebook ...