def test_downside_risk_noisy(self, noise, flat_line):
noisy_returns_1 = noise[0:250].add(flat_line[250:], fill_value=0)
noisy_returns_2 = noise[0:500].add(flat_line[500:], fill_value=0)
noisy_returns_3 = noise[0:750].add(flat_line[750:], fill_value=0)
dr_1 = empyrical.downside_risk(noisy_returns_1, flat_line)
dr_2 = empyrical.downside_risk(noisy_returns_2, flat_line)
dr_3 = empyrical.downside_risk(noisy_returns_3, flat_line)
assert dr_1 <= dr_2
assert dr_2 <= dr_3
def test_downside_risk_std(self, smaller_std, larger_std):
less_noise = pd.Series(
[random.gauss(0, smaller_std) for i in range(1000)],
index=pd.date_range('2000-1-30', periods=1000, freq='D'))
more_noise = pd.Series(
[random.gauss(0, larger_std) for i in range(1000)],
index=pd.date_range('2000-1-30', periods=1000, freq='D'))
assert empyrical.downside_risk(less_noise) < \
empyrical.downside_risk(more_noise)
def test_downside_risk(self, test_required_return):
res_a = empyrical.downside_risk(ret['a'], required_return=test_required_return)
res_b = empyrical.downside_risk(ret['b'], required_return=test_required_return)
res_c = empyrical.downside_risk(ret['c'], required_return=test_required_return)
assert isclose(ret['a'].vbt.returns.downside_risk(required_return=test_required_return), res_a)
pd.testing.assert_series_equal(
ret.vbt.returns.downside_risk(required_return=test_required_return),
pd.Series([res_a, res_b, res_c], index=ret.columns).rename('downside_risk')
)
def get_annual_downside_risk(self, data): # 2.15%
data = copy.deepcopy(data)
rh = self.rft_ret / self.q
downside_risk = empyrical.downside_risk(data.rets.dropna(),
required_return=rh,
period='weekly')
return downside_risk
def downside_risk(returns, required_return=0, period=DAILY):
"""
Determines the downside deviation below a threshold
Parameters
----------
returns : pd.Series or pd.DataFrame
Daily returns of the strategy, noncumulative.
- See full explanation in :func:`~pyfolio.timeseries.cum_returns`.
required_return: float / series
minimum acceptable return
period : str, optional
Defines the periodicity of the 'returns' data for purposes of
annualizing. Can be 'monthly', 'weekly', or 'daily'.
- Defaults to 'daily'.
Returns
-------
depends on input type
series ==> float
DataFrame ==> np.array
Annualized downside deviation
"""
return empyrical.downside_risk(returns,
required_return=required_return,
period=period)
def calculate_metrics(self):
self.benchmark_period_returns = \
cum_returns(self.benchmark_returns).iloc[-1]
self.algorithm_period_returns = \
cum_returns(self.algorithm_returns).iloc[-1]
if not self.algorithm_returns.index.equals(
self.benchmark_returns.index):
message = "Mismatch between benchmark_returns ({bm_count}) and \
algorithm_returns ({algo_count}) in range {start} : {end}"
message = message.format(bm_count=len(self.benchmark_returns),
algo_count=len(self.algorithm_returns),
start=self._start_session,
end=self._end_session)
raise Exception(message)
self.num_trading_days = len(self.benchmark_returns)
self.mean_algorithm_returns = (
self.algorithm_returns.cumsum() /
np.arange(1, self.num_trading_days + 1, dtype=np.float64))
self.benchmark_volatility = annual_volatility(self.benchmark_returns)
self.algorithm_volatility = annual_volatility(self.algorithm_returns)
self.treasury_period_return = choose_treasury(
self.treasury_curves,
self._start_session,
self._end_session,
self.trading_calendar,
)
self.sharpe = sharpe_ratio(self.algorithm_returns, )
# The consumer currently expects a 0.0 value for sharpe in period,
# this differs from cumulative which was np.nan.
# When factoring out the sharpe_ratio, the different return types
# were collapsed into `np.nan`.
# TODO: Either fix consumer to accept `np.nan` or make the
# `sharpe_ratio` return type configurable.
# In the meantime, convert nan values to 0.0
if pd.isnull(self.sharpe):
self.sharpe = 0.0
self.downside_risk = downside_risk(self.algorithm_returns.values)
self.sortino = sortino_ratio(
self.algorithm_returns.values,
_downside_risk=self.downside_risk,
)
self.information = information_ratio(
self.algorithm_returns.values,
self.benchmark_returns.values,
)
self.alpha, self.beta = alpha_beta_aligned(
self.algorithm_returns.values,
self.benchmark_returns.values,
)
self.excess_return = self.algorithm_period_returns - \
self.treasury_period_return
self.max_drawdown = max_drawdown(self.algorithm_returns.values)
self.max_leverage = self.calculate_max_leverage()
def evaluation(self):
ap.sound(f'entry: create_df')
mdd = empyrical.max_drawdown(self.df.eac_stgy_rt)
stgy_ret_an = empyrical.annual_return(self.df.eac_stgy_rt, annualization=self.cls.annualization)
bcmk_ret_an = empyrical.annual_return(self.df.eac_bcmk_rt, annualization=self.cls.annualization)
stgy_vlt_an = empyrical.annual_volatility(self.df.eac_stgy_rt, annualization=self.cls.annualization)
bcmk_vlt_an = empyrical.annual_volatility(self.df.eac_bcmk_rt, annualization=self.cls.annualization)
calmar = empyrical.calmar_ratio(self.df.eac_stgy_rt, annualization=self.cls.annualization)
omega = empyrical.omega_ratio(self.df.eac_stgy_rt, risk_free=self.cls.rf, annualization=self.cls.annualization)
sharpe = qp.sharpe_ratio(stgy_ret_an, self.df.cum_stgy_rt, self.cls.rf)
sortino = empyrical.sortino_ratio(self.df.eac_stgy_rt, annualization=self.cls.annualization)
dsrk = empyrical.downside_risk(self.df.eac_stgy_rt, annualization=self.cls.annualization)
information = empyrical.information_ratio(self.df.eac_stgy_rt, factor_returns=self.df.eac_bcmk_rt)
beta = empyrical.beta(self.df.eac_stgy_rt, factor_returns=self.df.eac_bcmk_rt, risk_free=self.cls.rf)
tail_rt = empyrical.tail_ratio(self.df.eac_stgy_rt)
alpha = qp.alpha_ratio(stgy_ret_an, bcmk_ret_an, self.cls.rf, beta)
stgy_ttrt_rt = (self.cls.yd.ttas[-1] - self.cls.yd.ttas[0]) / self.cls.yd.ttas[0]
bcmk_ttrt_rt = (self.cls.pc.close[-1] - self.cls.pc.close[0]) / self.cls.pc.close[0]
car_rt = stgy_ttrt_rt - bcmk_ttrt_rt
car_rt_an = stgy_ret_an - bcmk_ret_an
self.cls.df_output = pd.DataFrame(
{'sgty_ttrt_rt': [stgy_ttrt_rt], 'bcmk_ttrt_rt': [bcmk_ttrt_rt], 'car_rt': [car_rt],
'stgy_ret_an': [stgy_ret_an], 'bcmk_ret_an': [bcmk_ret_an], 'car_rt_an': [car_rt_an],
'stgy_vlt_an': [stgy_vlt_an], 'bcmk_vlt_an': [bcmk_vlt_an], 'mdd': [mdd],
'sharpe': [sharpe], 'alpha': [alpha], 'beta': [beta], 'information': [information],
'tail_rt': [tail_rt], 'calmar': [calmar], 'omega': [omega], 'sortino': [sortino], 'dsrk': [dsrk]})
print(f'feedback: \n{self.cls.df_output.T}')
def downside_risk(returns, required_return=0, period=DAILY):
"""
Determines the downside deviation below a threshold
Parameters
----------
returns : pd.Series or pd.DataFrame
Daily returns of the strategy, noncumulative.
- See full explanation in :func:`~pyfolio.timeseries.cum_returns`.
required_return: float / series
minimum acceptable return
period : str, optional
Defines the periodicity of the 'returns' data for purposes of
annualizing. Can be 'monthly', 'weekly', or 'daily'.
- Defaults to 'daily'.
Returns
-------
depends on input type
series ==> float
DataFrame ==> np.array
Annualized downside deviation
"""
return ep.downside_risk(returns,
required_return=required_return,
period=period)
def test_downside_risk(self, returns, required_return, period, expected):
downside_risk = empyrical.downside_risk(
returns, required_return=required_return, period=period)
if isinstance(downside_risk, float):
assert_almost_equal(downside_risk, expected, DECIMAL_PLACES)
else:
for i in range(downside_risk.size):
assert_almost_equal(downside_risk[i], expected[i],
DECIMAL_PLACES)
def plot(self):
# show a plot of portfolio vs mean market performance
df_info = pd.DataFrame(self.infos)
df_info.set_index('current step', inplace=True)
# df_info.set_index('date', inplace=True)
rn = np.asarray(df_info['portfolio return'])
try:
spf = df_info['portfolio value'].iloc[1] # Start portfolio value
epf = df_info['portfolio value'].iloc[-1] # End portfolio value
pr = (epf - spf) / spf
except:
pr = 0
try:
sr = sharpe_ratio(rn)
except:
sr = 0
try:
sor = sortino_ratio(rn)
except:
sor = 0
try:
mdd = max_drawdown(rn)
except:
mdd = 0
try:
cr = calmar_ratio(rn)
except:
cr = 0
try:
om = omega_ratio(rn)
except:
om = 0
try:
dr = downside_risk(rn)
except:
dr = 0
print("First portfolio value: ",
np.round(df_info['portfolio value'].iloc[1]))
print("Last portfolio value: ",
np.round(df_info['portfolio value'].iloc[-1]))
title = self.strategy_name + ': ' + 'profit={: 2.2%} sharpe={: 2.2f} sortino={: 2.2f} max drawdown={: 2.2%} calmar={: 2.2f} omega={: 2.2f} downside risk={: 2.2f}'.format(
pr, sr, sor, mdd, cr, om, dr)
# df_info[['market value', 'portfolio value']].plot(title=title, fig=plt.gcf(), figsize=(15,10), rot=30)
df_info[['portfolio value']].plot(title=title,
fig=plt.gcf(),
figsize=(15, 10),
rot=30)
def _get_reward(self, current_prices, next_prices):
if self.compute_reward == compute_reward.profit:
returns_rate = next_prices / current_prices
# pip_value = self._calculate_pip_value_in_account_currency(account_currency.USD, next_prices)
# returns_rate = np.multiply(returns_rate, pip_value)
log_returns = np.log(returns_rate)
last_weight = self.current_weights
securities_value = self.current_portfolio_values[:-1] * returns_rate
self.current_portfolio_values[:-1] = securities_value
self.current_weights = self.current_portfolio_values / np.sum(
self.current_portfolio_values)
reward = last_weight[:-1] * log_returns
elif self.compute_reward == compute_reward.sharpe:
try:
sr = sharpe_ratio(np.asarray(self.returns))
except:
sr = 0
reward = sr
elif self.compute_reward == compute_reward.sortino:
try:
sr = sortino_ratio(np.asarray(self.returns))
except:
sr = 0
reward = sr
elif self.compute_reward == compute_reward.max_drawdown:
try:
mdd = max_drawdown(np.asarray(self.returns))
except:
mdd = 0
reward = mdd
elif self.compute_reward == compute_reward.calmar:
try:
cr = calmar_ratio(np.asarray(self.returns))
except:
cr = 0
reward = cr
elif self.compute_reward == compute_reward.omega:
try:
om = omega_ratio(np.asarray(self.returns))
except:
om = 0
reward = om
elif self.compute_reward == compute_reward.downside_risk:
try:
dr = downside_risk(np.asarray(self.returns))
except:
dr = 0
reward = dr
try:
reward = reward.mean()
except:
reward = reward
return reward
def downside_risk(close, required_return=0, period='daily',
annualization=None):
try:
rets = daily_returns(close)
dr_data = empyrical.downside_risk(rets,
required_return=required_return,
period=period,
annualization=annualization)
return dr_data
except Exception as e:
raise (e)
def downside_risk(daily_returns,
required_return=0,
period='daily',
annualization=None):
"""Downside Risk"""
try:
logger.info('Calculating Downside Risk...')
dr_data = empyrical.downside_risk(daily_returns,
required_return=required_return,
period=period,
annualization=annualization)
return dr_data
except Exception as exception:
logger.error('Oops! An Error Occurred ⚠️')
raise exception
def getStats(cls, returns, benchmark_returns):
_alpha, _beta = alpha_beta_aligned(
returns,
benchmark_returns,
)
_sharpe = sharpe_ratio(
returns
)
_downside_risk = downside_risk(
returns
)
_max_drawdown = max_drawdown(
returns
)
_annual_volatility = annual_volatility(
returns
)
_benchmark_volatility = annual_volatility(
benchmark_returns
)
_annual_return = annual_return(
returns
)
_cum_return = cum_returns(
returns
)
return {
'cum_return' : _cum_return,
'annual_return' : _annual_return,
'annual_volatility' : _annual_volatility,
'benchmark_volatility' : _benchmark_volatility,
'max_drawdown' : _max_drawdown,
'downside_risk' : _downside_risk,
'sharpe ratio' : _sharpe,
'alpha' : _alpha,
'beta' : _beta,
}
def getCombinedIndex(assets, start_date, end_date, factor_weight, **kwargs):
daily_return_data = getDailyIndexData(assets, start_date, end_date, "CHG_PCT")
monthly_index_data = getMonthlyIndexData(assets, start_date, end_date)
monthly_return_data = getMonthlyReturnData(monthly_index_data)
#计算动量因子
momentum = getMomentum(monthly_return_data)
#计算波动率因子
#volatility = getAnnualizedVolatility(daily_return_data)
volatility = pd.DataFrame(empyrical.downside_risk(daily_return_data, required_return=0, period='daily'))
#计算相关性因子
corr_factor = getCorrelationFactor(daily_return_data)
#计算资产的三因子排名
momentum_rank = getRank(momentum, False)
volatility_rank = getRank(volatility, True)
corr_factor_rank = getRank(corr_factor, True)
combined_index = factor_weight["momentum"] * momentum_rank + factor_weight["volatility"] * volatility_rank + factor_weight["corr"] * corr_factor_rank
#过滤掉动量因子为负的资产
positive_momentum_asset = []
for asset in momentum.index:
if momentum.loc[asset, 0] > 0:
positive_momentum_asset.append(asset)
if 'bond' in kwargs.keys():
bond_assets = kwargs['bond']
for bond_asset in bond_assets:
if bond_asset not in positive_momentum_asset:
positive_momentum_asset.append(bond_asset)
combined_index = combined_index.loc[positive_momentum_asset]
return combined_index
def risk_metric_period(
cls,
start_session,
end_session,
algorithm_returns,
benchmark_returns,
algorithm_leverages,
):
"""
Creates a dictionary representing the state of the risk report.
Parameters
----------
start_session : pd.Timestamp
Start of period (inclusive) to produce metrics on
end_session : pd.Timestamp
End of period (inclusive) to produce metrics on
algorithm_returns : pd.Series(pd.Timestamp -> float)
Series of algorithm returns as of the end of each session
benchmark_returns : pd.Series(pd.Timestamp -> float)
Series of benchmark returns as of the end of each session
algorithm_leverages : pd.Series(pd.Timestamp -> float)
Series of algorithm leverages as of the end of each session
Returns
-------
risk_metric : dict[str, any]
Dict of metrics that with fields like:
{
'algorithm_period_return': 0.0,
'benchmark_period_return': 0.0,
'treasury_period_return': 0,
'excess_return': 0.0,
'alpha': 0.0,
'beta': 0.0,
'sharpe': 0.0,
'sortino': 0.0,
'period_label': '1970-01',
'trading_days': 0,
'algo_volatility': 0.0,
'benchmark_volatility': 0.0,
'max_drawdown': 0.0,
'max_leverage': 0.0,
}
"""
algorithm_returns = algorithm_returns[
(algorithm_returns.index >= start_session)
& (algorithm_returns.index <= end_session)]
# Benchmark needs to be masked to the same dates as the algo returns
benchmark_returns = benchmark_returns[
(benchmark_returns.index >= start_session)
& (benchmark_returns.index <= algorithm_returns.index[-1])]
benchmark_period_returns = ep.cum_returns(benchmark_returns).iloc[-1]
algorithm_period_returns = ep.cum_returns(algorithm_returns).iloc[-1]
alpha, beta = ep.alpha_beta_aligned(
algorithm_returns.values,
benchmark_returns.values,
)
benchmark_volatility = ep.annual_volatility(benchmark_returns)
sharpe = ep.sharpe_ratio(algorithm_returns)
# The consumer currently expects a 0.0 value for sharpe in period,
# this differs from cumulative which was np.nan.
# When factoring out the sharpe_ratio, the different return types
# were collapsed into `np.nan`.
# TODO: Either fix consumer to accept `np.nan` or make the
# `sharpe_ratio` return type configurable.
# In the meantime, convert nan values to 0.0
if pd.isnull(sharpe):
sharpe = 0.0
sortino = ep.sortino_ratio(
algorithm_returns.values,
_downside_risk=ep.downside_risk(algorithm_returns.values),
)
rval = {
"algorithm_period_return": algorithm_period_returns,
"benchmark_period_return": benchmark_period_returns,
"treasury_period_return": 0,
"excess_return": algorithm_period_returns,
"alpha": alpha,
"beta": beta,
"sharpe": sharpe,
"sortino": sortino,
"period_label": end_session.strftime("%Y-%m"),
"trading_days": len(benchmark_returns),
"algo_volatility": ep.annual_volatility(algorithm_returns),
"benchmark_volatility": benchmark_volatility,
"max_drawdown": ep.max_drawdown(algorithm_returns.values),
"max_leverage": algorithm_leverages.max(),
}
# check if a field in rval is nan or inf, and replace it with None
# except period_label which is always a str
return {
k: (None if k != "period_label" and not np.isfinite(v) else v)
for k, v in rval.items()
}
max_drawdown,
sharpe_ratio,
sortino_ratio,
calmar_ratio,
omega_ratio,
tail_ratio
)
import pandas as pd
returns = pd.Series(
index=pd.date_range('2017-03-10', '2017-03-19'),
data=(-0.012143, 0.045350, 0.030957, 0.004902, 0.002341, -0.02103, 0.00148, 0.004820, -0.00023, 0.01201)
)
benchmark_returns = pd.Series(
index=pd.date_range('2017-03-10', '2017-03-19'),
data=(-0.031940, 0.025350, -0.020957, -0.000902, 0.007341, -0.01103, 0.00248, 0.008820, -0.00123, 0.01091)
)
creturns = cum_returns(returns)
max_drawdown(returns)
annual_return(returns)
annual_volatility(returns, period='daily')
calmar_ratio(returns)
omega_ratio(returns=returns, risk_free=0.01)
sharpe_ratio(returns=returns, risk_free=0.01)
sortino_ratio(returns=returns)
downside_risk(returns=returns)
alpha(returns=returns, factor_returns=benchmark_returns, risk_free=0.01)
beta(returns=returns, factor_returns=benchmark_returns, risk_free=0.01)
tail_ratio(returns=returns)
def test_downside_risk_trans(self, returns, required_return):
dr_0 = empyrical.downside_risk(returns, -required_return)
dr_1 = empyrical.downside_risk(returns, 0)
dr_2 = empyrical.downside_risk(returns, required_return)
assert dr_0 <= dr_1
assert dr_1 <= dr_2
def getDownsideRisk(self, required_return=0, period='daily', annualization=None):
return empyrical.downside_risk(self.returns, required_return, period, annualization)
def risk_metric_period(cls,
start_session,
end_session,
algorithm_returns,
benchmark_returns,
algorithm_leverages):
"""
Creates a dictionary representing the state of the risk report.
Parameters
----------
start_session : pd.Timestamp
Start of period (inclusive) to produce metrics on
end_session : pd.Timestamp
End of period (inclusive) to produce metrics on
algorithm_returns : pd.Series(pd.Timestamp -> float)
Series of algorithm returns as of the end of each session
benchmark_returns : pd.Series(pd.Timestamp -> float)
Series of benchmark returns as of the end of each session
algorithm_leverages : pd.Series(pd.Timestamp -> float)
Series of algorithm leverages as of the end of each session
Returns
-------
risk_metric : dict[str, any]
Dict of metrics that with fields like:
{
'algorithm_period_return': 0.0,
'benchmark_period_return': 0.0,
'treasury_period_return': 0,
'excess_return': 0.0,
'alpha': 0.0,
'beta': 0.0,
'sharpe': 0.0,
'sortino': 0.0,
'period_label': '1970-01',
'trading_days': 0,
'algo_volatility': 0.0,
'benchmark_volatility': 0.0,
'max_drawdown': 0.0,
'max_leverage': 0.0,
}
"""
algorithm_returns = algorithm_returns[
(algorithm_returns.index >= start_session) &
(algorithm_returns.index <= end_session)
]
# Benchmark needs to be masked to the same dates as the algo returns
benchmark_returns = benchmark_returns[
(benchmark_returns.index >= start_session) &
(benchmark_returns.index <= algorithm_returns.index[-1])
]
benchmark_period_returns = ep.cum_returns(benchmark_returns).iloc[-1]
algorithm_period_returns = ep.cum_returns(algorithm_returns).iloc[-1]
alpha, beta = ep.alpha_beta_aligned(
algorithm_returns.values,
benchmark_returns.values,
)
sharpe = ep.sharpe_ratio(algorithm_returns)
# The consumer currently expects a 0.0 value for sharpe in period,
# this differs from cumulative which was np.nan.
# When factoring out the sharpe_ratio, the different return types
# were collapsed into `np.nan`.
# TODO: Either fix consumer to accept `np.nan` or make the
# `sharpe_ratio` return type configurable.
# In the meantime, convert nan values to 0.0
if pd.isnull(sharpe):
sharpe = 0.0
sortino = ep.sortino_ratio(
algorithm_returns.values,
_downside_risk=ep.downside_risk(algorithm_returns.values),
)
rval = {
'algorithm_period_return': algorithm_period_returns,
'benchmark_period_return': benchmark_period_returns,
'treasury_period_return': 0,
'excess_return': algorithm_period_returns,
'alpha': alpha,
'beta': beta,
'sharpe': sharpe,
'sortino': sortino,
'period_label': end_session.strftime("%Y-%m"),
'trading_days': len(benchmark_returns),
'algo_volatility': ep.annual_volatility(algorithm_returns),
'benchmark_volatility': ep.annual_volatility(benchmark_returns),
'max_drawdown': ep.max_drawdown(algorithm_returns.values),
'max_leverage': algorithm_leverages.max(),
}
# check if a field in rval is nan or inf, and replace it with None
# except period_label which is always a str
return {
k: (
None
if k != 'period_label' and not np.isfinite(v) else
v
)
for k, v in iteritems(rval)
}
def update(self, dt, algorithm_returns, benchmark_returns, leverage):
# Keep track of latest dt for use in to_dict and other methods
# that report current state.
self.latest_dt = dt
dt_loc = self.cont_index.get_loc(dt)
self.latest_dt_loc = dt_loc
self.algorithm_returns_cont[dt_loc] = algorithm_returns
self.algorithm_returns = self.algorithm_returns_cont[:dt_loc + 1]
self.num_trading_days = len(self.algorithm_returns)
if self.create_first_day_stats:
if len(self.algorithm_returns) == 1:
self.algorithm_returns = np.append(0.0, self.algorithm_returns)
self.algorithm_cumulative_returns[dt_loc] = cum_returns(
self.algorithm_returns
)[-1]
algo_cumulative_returns_to_date = \
self.algorithm_cumulative_returns[:dt_loc + 1]
self.mean_returns_cont[dt_loc] = \
algo_cumulative_returns_to_date[dt_loc] / self.num_trading_days
self.mean_returns = self.mean_returns_cont[:dt_loc + 1]
self.annualized_mean_returns_cont[dt_loc] = \
self.mean_returns_cont[dt_loc] * 252
self.annualized_mean_returns = \
self.annualized_mean_returns_cont[:dt_loc + 1]
if self.create_first_day_stats:
if len(self.mean_returns) == 1:
self.mean_returns = np.append(0.0, self.mean_returns)
self.annualized_mean_returns = np.append(
0.0, self.annualized_mean_returns)
self.benchmark_returns_cont[dt_loc] = benchmark_returns
self.benchmark_returns = self.benchmark_returns_cont[:dt_loc + 1]
if self.create_first_day_stats:
if len(self.benchmark_returns) == 1:
self.benchmark_returns = np.append(0.0, self.benchmark_returns)
self.benchmark_cumulative_returns[dt_loc] = cum_returns(
self.benchmark_returns
)[-1]
benchmark_cumulative_returns_to_date = \
self.benchmark_cumulative_returns[:dt_loc + 1]
self.mean_benchmark_returns_cont[dt_loc] = \
benchmark_cumulative_returns_to_date[dt_loc] / \
self.num_trading_days
self.mean_benchmark_returns = self.mean_benchmark_returns_cont[:dt_loc]
self.annualized_mean_benchmark_returns_cont[dt_loc] = \
self.mean_benchmark_returns_cont[dt_loc] * 252
self.annualized_mean_benchmark_returns = \
self.annualized_mean_benchmark_returns_cont[:dt_loc + 1]
self.algorithm_cumulative_leverages_cont[dt_loc] = leverage
self.algorithm_cumulative_leverages = \
self.algorithm_cumulative_leverages_cont[:dt_loc + 1]
if self.create_first_day_stats:
if len(self.algorithm_cumulative_leverages) == 1:
self.algorithm_cumulative_leverages = np.append(
0.0,
self.algorithm_cumulative_leverages)
if not len(self.algorithm_returns) and len(self.benchmark_returns):
message = "Mismatch between benchmark_returns ({bm_count}) and \
algorithm_returns ({algo_count}) in range {start} : {end} on {dt}"
message = message.format(
bm_count=len(self.benchmark_returns),
algo_count=len(self.algorithm_returns),
start=self.start_session,
end=self.end_session,
dt=dt
)
raise Exception(message)
self.update_current_max()
self.benchmark_volatility[dt_loc] = annual_volatility(
self.benchmark_returns
)
self.algorithm_volatility[dt_loc] = annual_volatility(
self.algorithm_returns
)
# caching the treasury rates for the minutely case is a
# big speedup, because it avoids searching the treasury
# curves on every minute.
# In both minutely and daily, the daily curve is always used.
treasury_end = dt.replace(hour=0, minute=0)
if np.isnan(self.daily_treasury[treasury_end]):
treasury_period_return = choose_treasury(
self.treasury_curves,
self.start_session,
treasury_end,
self.trading_calendar,
)
self.daily_treasury[treasury_end] = treasury_period_return
self.treasury_period_return = self.daily_treasury[treasury_end]
self.excess_returns[dt_loc] = (
self.algorithm_cumulative_returns[dt_loc] -
self.treasury_period_return)
self.alpha[dt_loc], self.beta[dt_loc] = alpha_beta_aligned(
self.algorithm_returns,
self.benchmark_returns,
)
self.sharpe[dt_loc] = sharpe_ratio(
self.algorithm_returns,
)
self.downside_risk[dt_loc] = downside_risk(
self.algorithm_returns
)
self.sortino[dt_loc] = sortino_ratio(
self.algorithm_returns,
_downside_risk=self.downside_risk[dt_loc]
)
self.information[dt_loc] = information_ratio(
self.algorithm_returns,
self.benchmark_returns,
)
self.max_drawdown = max_drawdown(
self.algorithm_returns
)
self.max_drawdowns[dt_loc] = self.max_drawdown
self.max_leverage = self.calculate_max_leverage()
self.max_leverages[dt_loc] = self.max_leverage
def calculate_metrics(self):
self.benchmark_period_returns = \
cum_returns(self.benchmark_returns).iloc[-1]
self.algorithm_period_returns = \
cum_returns(self.algorithm_returns).iloc[-1]
if not self.algorithm_returns.index.equals(
self.benchmark_returns.index
):
message = "Mismatch between benchmark_returns ({bm_count}) and \
algorithm_returns ({algo_count}) in range {start} : {end}"
message = message.format(
bm_count=len(self.benchmark_returns),
algo_count=len(self.algorithm_returns),
start=self._start_session,
end=self._end_session
)
raise Exception(message)
self.num_trading_days = len(self.benchmark_returns)
self.mean_algorithm_returns = (
self.algorithm_returns.cumsum() /
np.arange(1, self.num_trading_days + 1, dtype=np.float64)
)
self.benchmark_volatility = annual_volatility(self.benchmark_returns)
self.algorithm_volatility = annual_volatility(self.algorithm_returns)
self.treasury_period_return = choose_treasury(
self.treasury_curves,
self._start_session,
self._end_session,
self.trading_calendar,
)
self.sharpe = sharpe_ratio(
self.algorithm_returns,
)
# The consumer currently expects a 0.0 value for sharpe in period,
# this differs from cumulative which was np.nan.
# When factoring out the sharpe_ratio, the different return types
# were collapsed into `np.nan`.
# TODO: Either fix consumer to accept `np.nan` or make the
# `sharpe_ratio` return type configurable.
# In the meantime, convert nan values to 0.0
if pd.isnull(self.sharpe):
self.sharpe = 0.0
self.downside_risk = downside_risk(
self.algorithm_returns.values
)
self.sortino = sortino_ratio(
self.algorithm_returns.values,
_downside_risk=self.downside_risk,
)
self.information = information_ratio(
self.algorithm_returns.values,
self.benchmark_returns.values,
)
self.alpha, self.beta = alpha_beta_aligned(
self.algorithm_returns.values,
self.benchmark_returns.values,
)
self.excess_return = self.algorithm_period_returns - \
self.treasury_period_return
self.max_drawdown = max_drawdown(self.algorithm_returns.values)
self.max_leverage = self.calculate_max_leverage()
def update(self, dt, algorithm_returns, benchmark_returns, leverage):
# Keep track of latest dt for use in to_dict and other methods
# that report current state.
self.latest_dt = dt
dt_loc = self.cont_index.get_loc(dt)
self.latest_dt_loc = dt_loc
self.algorithm_returns_cont[dt_loc] = algorithm_returns
self.algorithm_returns = self.algorithm_returns_cont[:dt_loc + 1]
self.num_trading_days = len(self.algorithm_returns)
if self.create_first_day_stats:
if len(self.algorithm_returns) == 1:
self.algorithm_returns = np.append(0.0, self.algorithm_returns)
self.algorithm_cumulative_returns[dt_loc] = cum_returns(
self.algorithm_returns)[-1]
algo_cumulative_returns_to_date = \
self.algorithm_cumulative_returns[:dt_loc + 1]
self.mean_returns_cont[dt_loc] = \
algo_cumulative_returns_to_date[dt_loc] / self.num_trading_days
self.mean_returns = self.mean_returns_cont[:dt_loc + 1]
self.annualized_mean_returns_cont[dt_loc] = \
self.mean_returns_cont[dt_loc] * 252
self.annualized_mean_returns = \
self.annualized_mean_returns_cont[:dt_loc + 1]
if self.create_first_day_stats:
if len(self.mean_returns) == 1:
self.mean_returns = np.append(0.0, self.mean_returns)
self.annualized_mean_returns = np.append(
0.0, self.annualized_mean_returns)
self.benchmark_returns_cont[dt_loc] = benchmark_returns
self.benchmark_returns = self.benchmark_returns_cont[:dt_loc + 1]
if self.create_first_day_stats:
if len(self.benchmark_returns) == 1:
self.benchmark_returns = np.append(0.0, self.benchmark_returns)
self.benchmark_cumulative_returns[dt_loc] = cum_returns(
self.benchmark_returns)[-1]
benchmark_cumulative_returns_to_date = \
self.benchmark_cumulative_returns[:dt_loc + 1]
self.mean_benchmark_returns_cont[dt_loc] = \
benchmark_cumulative_returns_to_date[dt_loc] / \
self.num_trading_days
self.mean_benchmark_returns = self.mean_benchmark_returns_cont[:dt_loc]
self.annualized_mean_benchmark_returns_cont[dt_loc] = \
self.mean_benchmark_returns_cont[dt_loc] * 252
self.annualized_mean_benchmark_returns = \
self.annualized_mean_benchmark_returns_cont[:dt_loc + 1]
self.algorithm_cumulative_leverages_cont[dt_loc] = leverage
self.algorithm_cumulative_leverages = \
self.algorithm_cumulative_leverages_cont[:dt_loc + 1]
if self.create_first_day_stats:
if len(self.algorithm_cumulative_leverages) == 1:
self.algorithm_cumulative_leverages = np.append(
0.0, self.algorithm_cumulative_leverages)
if not len(self.algorithm_returns) and len(self.benchmark_returns):
message = "Mismatch between benchmark_returns ({bm_count}) and \
algorithm_returns ({algo_count}) in range {start} : {end} on {dt}"
message = message.format(bm_count=len(self.benchmark_returns),
algo_count=len(self.algorithm_returns),
start=self.start_session,
end=self.end_session,
dt=dt)
raise Exception(message)
self.update_current_max()
self.benchmark_volatility[dt_loc] = annual_volatility(
self.benchmark_returns)
self.algorithm_volatility[dt_loc] = annual_volatility(
self.algorithm_returns)
# caching the treasury rates for the minutely case is a
# big speedup, because it avoids searching the treasury
# curves on every minute.
# In both minutely and daily, the daily curve is always used.
treasury_end = dt.replace(hour=0, minute=0)
if np.isnan(self.daily_treasury[treasury_end]):
treasury_period_return = choose_treasury(
self.treasury_curves,
self.start_session,
treasury_end,
self.trading_calendar,
)
self.daily_treasury[treasury_end] = treasury_period_return
self.treasury_period_return = self.daily_treasury[treasury_end]
self.excess_returns[dt_loc] = (
self.algorithm_cumulative_returns[dt_loc] -
self.treasury_period_return)
self.alpha[dt_loc], self.beta[dt_loc] = alpha_beta_aligned(
self.algorithm_returns,
self.benchmark_returns,
)
self.sharpe[dt_loc] = sharpe_ratio(self.algorithm_returns, )
self.downside_risk[dt_loc] = downside_risk(self.algorithm_returns)
self.sortino[dt_loc] = sortino_ratio(
self.algorithm_returns, _downside_risk=self.downside_risk[dt_loc])
self.max_drawdown = max_drawdown(self.algorithm_returns)
self.max_drawdowns[dt_loc] = self.max_drawdown
self.max_leverage = self.calculate_max_leverage()
self.max_leverages[dt_loc] = self.max_leverage
def calculate_statistics(self, df: DataFrame = None, output=True):
""""""
self.output("开始计算策略统计指标")
# Check DataFrame input exterior
if df is None:
df = self.daily_df
# Check for init DataFrame
if df is None:
# Set all statistics to 0 if no trade.
start_date = ""
end_date = ""
total_days = 0
profit_days = 0
loss_days = 0
end_balance = 0
max_drawdown = 0
max_ddpercent = 0
max_drawdown_duration = 0
max_drawdown_end = 0
total_net_pnl = 0
daily_net_pnl = 0
total_commission = 0
daily_commission = 0
total_slippage = 0
daily_slippage = 0
total_turnover = 0
daily_turnover = 0
total_trade_count = 0
daily_trade_count = 0
total_return = 0
annual_return = 0
daily_return = 0
return_std = 0
sharpe_ratio = 0
sortino_info = 0
win_ratio = 0
return_drawdown_ratio = 0
tail_ratio_info = 0
stability_return = 0
win_loss_pnl_ratio = 0
pnl_medio = 0
duration_medio = 0
calmar_ratio = 0
else:
# Calculate balance related time series data
df["balance"] = df["net_pnl"].cumsum() + self.capital
df["return"] = np.log(df["balance"] /
df["balance"].shift(1)).fillna(0)
df["highlevel"] = (df["balance"].rolling(min_periods=1,
window=len(df),
center=False).max())
df["drawdown"] = df["balance"] - df["highlevel"]
df["ddpercent"] = df["drawdown"] / df["highlevel"] * 100
# Calculate statistics value
start_date = df.index[0]
end_date = df.index[-1]
total_days = len(df)
profit_days = len(df[df["net_pnl"] > 0])
loss_days = len(df[df["net_pnl"] < 0])
end_balance = df["balance"].iloc[-1]
max_drawdown = df["drawdown"].min()
max_ddpercent = df["ddpercent"].min()
max_drawdown_end = df["drawdown"].idxmin()
if isinstance(max_drawdown_end, date):
max_drawdown_start = df["balance"][:max_drawdown_end].idxmax()
max_drawdown_duration = (max_drawdown_end -
max_drawdown_start).days
else:
max_drawdown_duration = 0
total_net_pnl = df["net_pnl"].sum()
daily_net_pnl = total_net_pnl / total_days
win = df[df["net_pnl"] > 0]
win_amount = win["net_pnl"].sum()
win_pnl_medio = win["net_pnl"].mean()
# win_duration_medio = win["duration"].mean().total_seconds()/3600
win_count = win["trade_count"].sum()
pnl_medio = df["net_pnl"].mean()
# duration_medio = df["duration"].mean().total_seconds()/3600
loss = df[df["net_pnl"] < 0]
loss_amount = loss["net_pnl"].sum()
loss_pnl_medio = loss["net_pnl"].mean()
# loss_duration_medio = loss["duration"].mean().total_seconds()/3600
total_commission = df["commission"].sum()
daily_commission = total_commission / total_days
total_slippage = df["slippage"].sum()
daily_slippage = total_slippage / total_days
total_turnover = df["turnover"].sum()
daily_turnover = total_turnover / total_days
total_trade_count = df["trade_count"].sum()
win_ratio = (win_count / total_trade_count) * 100
win_loss_pnl_ratio = -win_pnl_medio / loss_pnl_medio
daily_trade_count = total_trade_count / total_days
total_return = (end_balance / self.capital - 1) * 100
annual_return = total_return / total_days * 240
daily_return = df["return"].mean() * 100
return_std = df["return"].std() * 100
if return_std:
sharpe_ratio = daily_return / return_std * np.sqrt(240)
else:
sharpe_ratio = 0
return_drawdown_ratio = -total_return / max_ddpercent
#calmar_ratio:年化收益率与历史最大回撤率之间的比率
calmar_ratio = annual_return / abs(max_ddpercent)
#sortino_info
sortino_info = sortino_ratio(df['return'])
omega_info = omega_ratio(df['return'])
#年化波动率
annual_volatility_info = annual_volatility(df['return'])
#年化复合增长率
cagr_info = cagr(df['return'])
#年化下行风险率
annual_downside_risk = downside_risk(df['return'])
"""CVaR即条件风险价值，其含义为在投资组合的损失超过某个给定VaR值的条件下，该投资组合的平均损失值。"""
c_var = conditional_value_at_risk(df['return'])
"""风险价值（VaR）是对投资损失风险的一种度量。它估计在正常的市场条件下，在设定的时间段（例如一天）中，
一组投资可能（以给定的概率）损失多少。金融业中的公司和监管机构通常使用VaR来衡量弥补可能损失所需的资产数量"""
var_info = value_at_risk(df['return'])
#收益稳定率
stability_return = stability_of_timeseries(df['return'])
#尾部比率0.25 == 1/4,收益1，风险4
tail_ratio_info = tail_ratio(df['return'])
# Output
if output:
self.output("-" * 30)
self.output(f"首个交易日：\t{start_date}")
self.output(f"最后交易日：\t{end_date}")
self.output(f"总交易日：\t{total_days}")
self.output(f"盈利交易日：\t{profit_days}")
self.output(f"亏损交易日：\t{loss_days}")
self.output(f"起始资金：\t{self.capital:,.2f}")
self.output(f"结束资金：\t{end_balance:,.2f}")
self.output(f"总收益率：\t{total_return:,.2f}%")
self.output(f"年化收益：\t{annual_return:,.2f}%")
self.output(f"最大回撤: \t{max_drawdown:,.2f}")
self.output(f"百分比最大回撤: {max_ddpercent:,.2f}%")
self.output(f"最长回撤天数: \t{max_drawdown_duration}")
self.output(f"总盈亏：\t{total_net_pnl:,.2f}")
self.output(f"总手续费：\t{total_commission:,.2f}")
self.output(f"总滑点：\t{total_slippage:,.2f}")
self.output(f"总成交金额：\t{total_turnover:,.2f}")
self.output(f"总成交笔数：\t{total_trade_count}")
self.output(f"日均盈亏：\t{daily_net_pnl:,.2f}")
self.output(f"日均手续费：\t{daily_commission:,.2f}")
self.output(f"日均滑点：\t{daily_slippage:,.2f}")
self.output(f"日均成交金额：\t{daily_turnover:,.2f}")
self.output(f"日均成交笔数：\t{daily_trade_count}")
self.output(f"日均收益率：\t{daily_return:,.2f}%")
self.output(f"收益标准差：\t{return_std:,.2f}%")
self.output(f"胜率：\t{win_ratio:,.2f}")
self.output(f"盈亏比:\t\t{win_loss_pnl_ratio:,.2f}")
self.output(f"平均每笔盈亏:\t{pnl_medio:,.2f}")
self.output(f"calmar_ratio：\t{calmar_ratio:,.3f}")
# self.output(f"平均持仓小时:\t{duration_medio:,.2f}")
self.output(f"Sharpe Ratio：\t{sharpe_ratio:,.2f}")
self.output(f"sortino Ratio：\t{sortino_info:,.3f}")
self.output(f"收益回撤比：\t{return_drawdown_ratio:,.2f}")
statistics = {
"start_date": start_date,
"end_date": end_date,
"total_days": total_days,
"profit_days": profit_days,
"loss_days": loss_days,
"capital": self.capital,
"end_balance": end_balance,
"max_drawdown": max_drawdown,
"max_ddpercent": max_ddpercent,
"max_drawdown_end": max_drawdown_end,
"max_drawdown_duration": max_drawdown_duration,
"total_net_pnl": total_net_pnl,
"daily_net_pnl": daily_net_pnl,
"total_commission": total_commission,
"daily_commission": daily_commission,
"total_slippage": total_slippage,
"daily_slippage": daily_slippage,
"total_turnover": total_turnover,
"daily_turnover": daily_turnover,
"total_trade_count": total_trade_count,
"daily_trade_count": daily_trade_count,
"total_return": total_return,
"annual_return": annual_return,
"daily_return": daily_return,
"return_std": return_std,
"sharpe_ratio": sharpe_ratio,
'sortino_info': sortino_info,
"win_ratio": win_ratio,
"return_drawdown_ratio": return_drawdown_ratio,
"tail_ratio_info": tail_ratio_info,
"stability_return": stability_return,
"win_loss_pnl_ratio": win_loss_pnl_ratio,
"pnl_medio": pnl_medio,
"calmar_ratio": calmar_ratio
}
# Filter potential error infinite value
for key, value in statistics.items():
if value in (np.inf, -np.inf):
value = 0
statistics[key] = np.nan_to_num(value)
self.output("策略统计指标计算完成")
return statistics
model_price.to(device)
trading_threshold = 0.6
with torch.no_grad():
print('Threshold: ', trading_threshold)
profits = []
for i in tqdm(range(len(testdata))):
embedding = testdata[i]["embedding"].to(device).unsqueeze(0)
length_data = testdata[i]["length_data"].to(device).unsqueeze(0)
time_feats = testdata[i]["time_feature"].to(device).squeeze(
-1).unsqueeze(0)
cp_last = testdata_1[i]["cp_last"]
cp_target = testdata_1[i]["cp_target"]
outputs_price, _ = model_price(embedding, length_data, time_feats)
outputs_probs = torch.nn.functional.softmax(outputs_price,
dim=-1).squeeze(0)
if (torch.max(outputs_probs) > trading_threshold):
if torch.argmax(outputs_price) == 1:
profits.append(cp_target - cp_last)
else:
profits.append(cp_last - cp_target)
profits = np.array(profits)
sortino = sortino_ratio(profits)
downside = downside_risk(profits)
print("Downside Risk: ", downside)
print("Sortino Ratio: ", sortino)
