Source code for fynance.features.money_management
#!/usr/bin/env python3
# coding: utf-8
# @Author: ArthurBernard
# @Email: arthur.bernard.92@gmail.com
# @Date: 2019-03-15 12:23:04
# @Last modified by: ArthurBernard
# @Last modified time: 2019-11-05 15:56:01
""" Module with function to compute some money management coefficients. """
# Built-in packages
# External packages
import numpy as np
# Internal packages
from fynance.features.momentums import ema
__all__ = ['iso_vol']
# =========================================================================== #
# Money-management Tools #
# =========================================================================== #
[docs]
def iso_vol(series, target_vol=0.20, leverage=1., period=252, half_life=11):
""" Make an iso-vol vector to apply to signal vector.
Parameters
----------
series : np.ndarray[ndim=1, dtype=np.float64]
Series of price of underlying.
target_vol : float (default 20 %)
Volatility to target.
leverage : float (default 1)
Max leverage to use.
period : int (default 250)
Number of period per year.
half_life : int (default 11)
Half-life of exponential moving average used to compute volatility.
Returns
-------
iv : np.ndarray[ndim=1, dtype=np.float64]
Series of iso-vol coefficient.
Examples
--------
>>> series = np.array([95, 100, 85, 105, 110, 90]).astype(np.float64)
>>> iso_vol(series, target_vol=0.5, leverage=2, period=12, half_life=3)
array([1. , 1. , 2. , 1.28407693, 0.78278978,
1.07186485])
"""
# Coerce input to a 1d float array so list/tuple inputs work too.
series = np.asarray(series, dtype=np.float64).reshape(-1)
# Set iso-vol vector
iv = np.ones([series.size])
# Compute squared daily return vector (standard return s_t / s_{t-1} - 1)
ret2 = np.square(series[1:] / series[:-1] - 1)
# Compute volatility vector
vol = np.sqrt(period * ema(ret2, w=half_life))
vol[vol <= 0.] = 1e-8
# Compute iso-vol coefficient (iv[t] uses only vol[t - 2], i.e. data up to
# series[t - 1] -- strictly causal w.r.t. the signal applied at t)
iv[2:] = target_vol / vol[:-1]
# Cap with the max leverage available
iv[iv > leverage] = leverage
return iv
