ml-finance-python

python scripts for finance machine learning

git clone https://9o.is/git/ml-finance-python.git

edhec_risk_kit_105.py

(2436B)

      1 import pandas as pd
      2 
      3 def drawdown(return_series: pd.Series):
      4     """Takes a time series of asset returns.
      5        returns a DataFrame with columns for
      6        the wealth index, 
      7        the previous peaks, and 
      8        the percentage drawdown
      9     """
     10     wealth_index = 1000*(1+return_series).cumprod()
     11     previous_peaks = wealth_index.cummax()
     12     drawdowns = (wealth_index - previous_peaks)/previous_peaks
     13     return pd.DataFrame({"Wealth": wealth_index, 
     14                          "Previous Peak": previous_peaks, 
     15                          "Drawdown": drawdowns})
     16 
     17 
     18 def get_ffme_returns():
     19     """
     20     Load the Fama-French Dataset for the returns of the Top and Bottom Deciles by MarketCap
     21     """
     22     me_m = pd.read_csv("data/Portfolios_Formed_on_ME_monthly_EW.csv",
     23                        header=0, index_col=0, na_values=-99.99)
     24     rets = me_m[['Lo 10', 'Hi 10']]
     25     rets.columns = ['SmallCap', 'LargeCap']
     26     rets = rets/100
     27     rets.index = pd.to_datetime(rets.index, format="%Y%m").to_period('M')
     28     return rets
     29 
     30 
     31 def get_hfi_returns():
     32     """
     33     Load and format the EDHEC Hedge Fund Index Returns
     34     """
     35     hfi = pd.read_csv("data/edhec-hedgefundindices.csv",
     36                       header=0, index_col=0, parse_dates=True)
     37     hfi = hfi/100
     38     hfi.index = hfi.index.to_period('M')
     39     return hfi
     40 
     41 
     42 def skewness(r):
     43     """
     44     Alternative to scipy.stats.skew()
     45     Computes the skewness of the supplied Series or DataFrame
     46     Returns a float or a Series
     47     """
     48     demeaned_r = r - r.mean()
     49     # use the population standard deviation, so set dof=0
     50     sigma_r = r.std(ddof=0)
     51     exp = (demeaned_r**3).mean()
     52     return exp/sigma_r**3
     53 
     54 
     55 def kurtosis(r):
     56     """
     57     Alternative to scipy.stats.kurtosis()
     58     Computes the kurtosis of the supplied Series or DataFrame
     59     Returns a float or a Series
     60     """
     61     demeaned_r = r - r.mean()
     62     # use the population standard deviation, so set dof=0
     63     sigma_r = r.std(ddof=0)
     64     exp = (demeaned_r**4).mean()
     65     return exp/sigma_r**4
     66 
     67 import scipy.stats
     68 def is_normal(r, level=0.01):
     69     """
     70     Applies the Jarque-Bera test to determine if a Series is normal or not
     71     Test is applied at the 1% level by default
     72     Returns True if the hypothesis of normality is accepted, False otherwise
     73     """
     74     if isinstance(r, pd.DataFrame):
     75         return r.aggregate(is_normal)
     76     else:
     77         statistic, p_value = scipy.stats.jarque_bera(r)
     78         return p_value > level
     79