def import_vol_surface():
"""Import volatility surface.
Infer risk-free rate directly from data.
"""
zf = zipfile.ZipFile(path + "SPX_surface.zip", "r")
name = zf.namelist()[0]
df = pd.read_csv(zf.open(name), converters={"date": convert_dates})
df.loc[:, "weekday"] = df["date"].apply(lambda x: x.weekday())
# Apply some filters
df = df[df["weekday"] == 2]
df = df[df["days"] <= 365]
df = df.drop("weekday", axis=1)
surface = df # .set_index(['cp_flag', 'date', 'days']).sort_index()
cols = {"impl_volatility": "imp_vol", "impl_strike": "strike", "impl_premium": "premium"}
surface.rename(columns=cols, inplace=True)
# TODO : who term structure should be imported and merged!
riskfree = load_riskfree().reset_index()
dividends = load_dividends().reset_index()
spx = load_spx().reset_index()
surface = pd.merge(surface, riskfree)
surface = pd.merge(surface, spx)
surface = pd.merge(surface, dividends)
# Adjust riskfree by dividend yield
surface["riskfree"] -= surface["rate"]
# Remove percentage point
surface["riskfree"] /= 100
# Replace 'cp_flag' with True/False 'call' variable
surface.loc[:, "call"] = True
surface.loc[surface["cp_flag"] == "P", "call"] = False
# Normalize maturity to being a share of the year
surface["maturity"] = surface["days"] / 365
# Rename columns
surface.rename(columns={"spx": "price"}, inplace=True)
# Compute lf-moneyness
surface["moneyness"] = lfmoneyness(surface["price"], surface["strike"], surface["riskfree"], surface["maturity"])
# Compute option Delta normalized by current price
surface["delta"] = delta(surface["moneyness"], surface["maturity"], surface["imp_vol"], surface["call"])
# Compute option Vega normalized by current price
surface["vega"] = vega(surface["moneyness"], surface["maturity"], surface["imp_vol"])
# Sort index
surface.sort_index(by=["date", "maturity", "moneyness"], inplace=True)
print(surface.head())
surface.to_hdf(path + "surface.h5", "surface")
def import_vol_surface_simple():
"""Import volatility surface. Simple version.
"""
zf = zipfile.ZipFile(path + "SPX_surface.zip", "r")
name = zf.namelist()[0]
df = pd.read_csv(zf.open(name), converters={"date": convert_dates})
df.loc[:, "weekday"] = df["date"].apply(lambda x: x.weekday())
# Apply some filters
df = df[df["weekday"] == 2]
df = df[df["days"] <= 365]
surface = df.drop("weekday", axis=1)
cols = {"impl_volatility": "imp_vol", "impl_strike": "strike", "impl_premium": "premium"}
surface.rename(columns=cols, inplace=True)
spx = load_spx().reset_index()
standard_options = load_standard_options()[["forward"]].reset_index()
surface = pd.merge(surface, standard_options)
surface = pd.merge(surface, spx)
# Normalize maturity to being a share of the year
surface["maturity"] = surface["days"] / 365
surface["riskfree"] = np.log(surface["forward"] / surface["spx"])
surface["riskfree"] /= surface["maturity"]
# Remove percentage point
# Replace 'cp_flag' with True/False 'call' variable
surface.loc[:, "call"] = True
surface.loc[surface["cp_flag"] == "P", "call"] = False
# Rename columns
surface.rename(columns={"spx": "price"}, inplace=True)
# Compute lf-moneyness
surface["moneyness"] = lfmoneyness(surface["price"], surface["strike"], surface["riskfree"], surface["maturity"])
# Compute option Delta normalized by current price
surface["delta"] = delta(surface["moneyness"], surface["maturity"], surface["imp_vol"], surface["call"])
# Compute option Vega normalized by current price
surface["vega"] = vega(surface["moneyness"], surface["maturity"], surface["imp_vol"])
# Take out-of-the-money options
calls = surface["call"] & (surface["moneyness"] >= 0)
puts = np.logical_not(surface["call"]) & (surface["moneyness"] < 0)
surface = pd.concat([surface[calls], surface[puts]])
# Sort index
surface.sort_index(by=["date", "maturity", "moneyness"], inplace=True)
print(surface.head())
surface.to_hdf(path + "surface.h5", "surface")