#!/usr/bin/env python3
# coding: utf-8
# @Author: ArthurBernard
# @Email: arthur.bernard.92@gmail.com
# @Date: 2019-11-05 15:50:33
# @Last modified by: ArthurBernard
# @Last modified time: 2020-05-08 09:22:17
from __future__ import annotations
# Built-in packages
# Third party packages
import numpy as np
from numpy.typing import NDArray
# Local packages
from fynance.features.metrics import accuracy, calmar, sharpe
__all__ = ['set_text_stats']
# =========================================================================== #
# Printer Tools #
# =========================================================================== #
[docs]
def set_text_stats(
underly: NDArray,
period: int = 252,
accur: bool = True,
perf: bool = True,
vol: bool = True,
sharp: bool = True,
calma: bool = True,
underlying: str = 'Underlying',
fees: float = 0,
**kwpred,
) -> str:
""" Set a table as string with different indicators (accuracy, perf, vol
and sharpe) for underlying and several strategies.
Parameters
----------
underly : np.ndarray[ndim=1, dtype=np.float64]
Series of underlying prices.
period : int, optional
Number of period per day, default is 252.
accur : bool, optional
If true compute accuracy else not, default is True.
perf : bool, optional
If true compute performance else not, default is True.
vol : bool, optional
If true compute volatility else not, default is True.
sharp : bool, optional
If true compute sharpe ratio else not, default is True.
calma : bool, optional
If true compute calmar ratio else not, default is True.
underlying : str, optional
Name of the underlying, default is 'Underlying'.
kwpred : dict of np.ndarray
Any strategies or predictions that you want to compare.
fees : float, optional
Fees to apply at the strategy performance.
Returns
-------
txt : str
Table of results.
See Also
--------
PlotBackTest, display_perf
"""
txt = ''
# Compute Accuracy
if accur:
txt += '+=============================+\n'
txt += '| Accuracy |\n'
txt += '+----------------+------------+\n'
for key, pred in kwpred.items():
accu_pred = accuracy(underly, pred)
txt += '| {:14} | {:10.2%} |\n'.format(key, accu_pred)
# Compute performance
if perf:
txt += '+=============================+\n'
txt += '| Performance |\n'
txt += '+----------------+------------+\n'
perf = np.exp(np.cumsum(underly))
perf_targ = np.sign(perf[-1] / perf[0]) * np.float_power(
np.abs(perf[-1] / perf[0]), period / perf.size) - 1.
txt += '| {:14} | {:10.2%} |\n'.format(underlying, perf_targ)
for key, pred in kwpred.items():
vect_fee = np.zeros(pred.shape)
vect_fee[1:] += (pred[1:] - pred[:-1]) * fees
perf = np.cumprod((np.exp(underly) - 1 - vect_fee) * pred + 1)
perf_pred = np.sign(perf[-1] / perf[0]) * np.float_power(
np.abs(perf[-1] / perf[0]), period / perf.size) - 1.
txt += '| {:14} | {:10.2%} |\n'.format(key, perf_pred)
# Compute volatility
if vol:
txt += '+=============================+\n'
txt += '| Volatility |\n'
txt += '+----------------+------------+\n'
perf = np.exp(np.cumsum(underly))
vol_targ = np.sqrt(period) * np.std(perf[1:] / perf[:-1] - 1)
txt += '| {:14} | {:10.2%} |\n'.format(underlying, vol_targ)
for key, pred in kwpred.items():
vect_fee = np.zeros(pred.shape)
vect_fee[1:] += (pred[1:] - pred[:-1]) * fees
perf = np.cumprod((np.exp(underly) - 1 - vect_fee) * pred + 1)
vol_pred = np.sqrt(period) * np.std(perf[1:] / perf[:-1] - 1)
txt += '| {:14} | {:10.2%} |\n'.format(key, vol_pred)
# Compute sharpe Ratio
if sharp:
txt += '+=============================+\n'
txt += '| Sharpe Ratio |\n'
txt += '+----------------+------------+\n'
sharpe_targ = sharpe(np.exp(np.cumsum(underly)), period=period)
txt += '| {:14} | {:10.2f} |\n'.format(underlying, sharpe_targ)
for key, pred in kwpred.items():
vect_fee = np.zeros(pred.shape)
vect_fee[1:] += (pred[1:] - pred[:-1]) * fees
perf = np.cumprod((np.exp(underly) - 1 - vect_fee) * pred + 1)
sharpe_pred = sharpe(perf, period=period)
txt += '| {:14} | {:10.2f} |\n'.format(key, sharpe_pred)
# Compute calmar
if calma:
txt += '+=============================+\n'
txt += '| Calmar Ratio |\n'
txt += '+----------------+------------+\n'
calmar_targ = calmar(np.exp(np.cumsum(underly)), period=period)
txt += '| {:14} | {:10.2f} |\n'.format(underlying, calmar_targ)
for key, pred in kwpred.items():
vect_fee = np.zeros(pred.shape)
vect_fee[1:] += (pred[1:] - pred[:-1]) * fees
perf = np.cumprod((np.exp(underly) - 1 - vect_fee) * pred + 1)
calmar_pred = calmar(perf, period=period)
txt += '| {:14} | {:10.2f} |\n'.format(key, calmar_pred)
txt += '+=============================+\n'
return txt
