def setUp(self):
ls_asset = ['GOOG']
self.df_stock_rets = get_price_data(
ls_asset, end_date=dt.datetime.today(),
look_back_mths=12).pct_change().fillna(0)
self.df_benchmark_rets = get_price_data(
['^GSPC'], end_date=dt.datetime.today(),
look_back_mths=12).pct_change().fillna(0)
def __init__(self,
asset_name_ls,
rets_hist_length_yrs=10,
end_date=dt.datetime.today()):
self.n = len(asset_name_ls) # number of assets
self.df_price_data = get_price_data(
asset_name_ls, end_date, look_back_mths=rets_hist_length_yrs * 12)
def risk_adversion_coefficient(benchmark_ticker, risk_free_rate=0.02, lookback_period_months=48):
"""
Calculate the risk adversion coefficient
Args:
benchmark_ticker: <str> ticker of the benchmark market asset
risk_free_rate: <float> the annual risk free rate
lookback_period_months: <int> number of months from today
Returns:
<float> risk adversion coefficient
"""
end_date = dt.datetime.today()
df_price_data = get_price_data([benchmark_ticker], end_date, look_back_mths=lookback_period_months)
# resample to annual data
rets = df_price_data.resample('Y').last().pct_change().apply(lambda x: np.log(1 + x))
exp_ret = rets.mean()
var = rets.var()
return ((exp_ret - risk_free_rate) / var)[0]
# Residual return = Excess return - (Benchmark's excess return * beta).
df = stock_excess_rets - (benchmark_excess_rets *
beta_stock_over_benchmark)
df.columns = [const.RESIDUAL_RETURN]
return {
const.RESIDUAL_RETURN: df,
const.RESIDUAL_RISK: df.std(),
const.EXP_RESIDUAL_RETURN: df.mean()
}
if __name__ == '__main__':
# Apple
df_stock_rets = get_price_data(['AAPL'],
end_date=dt.datetime.today(),
look_back_mths=12).pct_change().fillna(0)
# S&P 500
df_market_rets = get_price_data(['^GSPC'],
end_date=dt.datetime.today(),
look_back_mths=12).pct_change().fillna(0)
# iShares U.S. Technology ETF
df_benchmark_rets = get_price_data(
['IYW'], end_date=dt.datetime.today(),
look_back_mths=12).pct_change().fillna(0)
# compute beta over the market
beta_stock = calc_beta(df_stock_rets, df_market_rets)
beta_benchmark = calc_beta(df_benchmark_rets, df_market_rets)
# compute beta of the stock over the benchmark
def maximum_drawdown(df_price_data):
"""
Get the maximum drawdown from all given price data
Args:
df_price_data: <pd.DataFrame> for an individual stock's price history
Returns:
<pd.Series> containing the maximum drawdown experienced during the given price data history
"""
max_peaks = df_price_data.cummax()
daily_drawdowns = (df_price_data - max_peaks) / max_peaks
max_drawdown = daily_drawdowns.cummin()
# return the last element since this would be the cumulative minimum
return max_drawdown.iloc[-1]
if __name__ == '__main__':
ls_assets = ['AAPL', 'NKE', 'GOOGL', 'AMZN']
df = get_price_data(ls_assets, end_date=dt.datetime.today(), look_back_mths=24).pct_change().fillna(0)
print(volatility(df))
print(semi_deviation(df))
print(gaussian_VaR(df))
df_price_data = get_price_data(['AAPL'], end_date=dt.datetime.today(), look_back_mths=48)
print(rolling_maximum_drawdowns(df_price_data=df_price_data))
print(maximum_drawdown(df_price_data=df_price_data))