def fitness_exp(strategy, params):
strategy_weights = weight.weights(datasets_dict['Sectors'], strategy,
params['NEUTRALIZATION'],
params['LONG LEVERAGE'],
params['SHORT LEVERAGE'])
# print("check")
#Simple Checks
empty = np.full_like(strategy_weights, np.nan)
check = ((strategy_weights == empty) |
(np.isnan(strategy_weights) & np.isnan(empty))).all()
if check:
strategy_weights = np.full_like(strategy_weights, 0)
non_nan_frac = np.count_nonzero(~np.isnan(strategy_weights)) / np.size(
strategy_weights)
if non_nan_frac <= 0.6:
strategy_weights = np.full_like(strategy_weights, 0)
clean_values_from_weights = wb.get_valid_index(strategy_weights)
cleaned_index_weights = (clean_index.values)[clean_values_from_weights]
cleaned_strategy_weights = strategy_weights[clean_values_from_weights]
strategy_log_returns, strategy_daily_returns = wb.bets_to_pnl(
params['STARTING VALUE'], strategy_weights,
clean_values_from_weights, base_data, params['LONG LEVERAGE'],
params['SHORT LEVERAGE'], params['ADV THRESHOLD PERCENTAGE'],
params['COST THRESHOLD BPS'], params['COMMISSION BPS'])
def equity_curve(amount, ret):
ret = hf.shift_array(ret, 1, 0)
return amount * np.cumprod(1 + ret)
def MDD(returns):
def returns_to_dollars(amount, ret):
return amount * np.cumprod(1 + ret)
doll_series = pd.Series(returns_to_dollars(100, returns))
Roll_Max = doll_series.cummax()
Daily_Drawdown = doll_series / Roll_Max - 1.0
Max_Daily_Drawdown = Daily_Drawdown.cummin()
return Max_Daily_Drawdown.min()
if np.sum(strategy_daily_returns) == 0:
return 0
else:
yrs = int(len(cleaned_index_weights)) / 252
curves = dict()
curves['Strategy'] = equity_curve(100, strategy_daily_returns)
cagr_strategy = ((
(curves['Strategy'][-1] / curves['Strategy'][0])**(1 / yrs)) -
1)
mdd_strat = MDD(strategy_log_returns)
fitness = (cagr_strategy) / abs(mdd_strat)
return fitness
def fitness_exp(strategy, params):
strategy_weights = weight.weights(datasets_dict['Sectors'], strategy,
params['NEUTRALIZATION'],
params['LONG LEVERAGE'],
params['SHORT LEVERAGE'])
# print("check")
#Simple Checks
empty = np.full_like(strategy_weights, np.nan)
check = ((strategy_weights == empty) |
(np.isnan(strategy_weights) & np.isnan(empty))).all()
if check:
strategy_weights = np.full_like(strategy_weights, 0)
non_nan_frac = np.count_nonzero(~np.isnan(strategy_weights)) / np.size(
strategy_weights)
if non_nan_frac <= 0.6:
strategy_weights = np.full_like(strategy_weights, 0)
clean_values_from_weights = wb.get_valid_index(strategy_weights)
cleaned_index_weights = (clean_index.values)[clean_values_from_weights]
cleaned_strategy_weights = strategy_weights[clean_values_from_weights]
strategy_log_returns, strategy_daily_returns = wb.bets_to_pnl(
params['STARTING VALUE'], strategy_weights,
clean_values_from_weights, base_data, params['LONG LEVERAGE'],
params['SHORT LEVERAGE'], params['ADV THRESHOLD PERCENTAGE'],
params['COST THRESHOLD BPS'], params['COMMISSION BPS'])
if np.sum(clean_values_from_weights) != 0:
rf_returns = datasets_dict['rf_rate'][clean_values_from_weights]
else:
rf_returns = datasets_dict['rf_rate']
# cleaned_index = cleaned_index_weights
if np.sum(strategy_daily_returns) == 0:
return 0
else:
excess_returns = strategy_log_returns - rf_returns
left_threshold = np.percentile(excess_returns, 5)
right_threshold = np.percentile(excess_returns, 95)
CVAR_left = -1 * (np.nanmean(
excess_returns[excess_returns <= left_threshold]))
CVAR_right = (np.nanmean(
excess_returns[excess_returns >= right_threshold]))
fitness = CVAR_right / CVAR_left
return fitness
def fitness_exp(strategy, params):
strategy_weights = weight.weights(datasets_dict['Sectors'], strategy,
params['NEUTRALIZATION'],
params['LONG LEVERAGE'],
params['SHORT LEVERAGE'])
# print("check")
#Simple Checks
empty = np.full_like(strategy_weights, np.nan)
check = ((strategy_weights == empty) |
(np.isnan(strategy_weights) & np.isnan(empty))).all()
if check:
strategy_weights = np.full_like(strategy_weights, 0)
non_nan_frac = np.count_nonzero(~np.isnan(strategy_weights)) / np.size(
strategy_weights)
if non_nan_frac <= 0.6:
strategy_weights = np.full_like(strategy_weights, 0)
clean_values_from_weights = wb.get_valid_index(strategy_weights)
cleaned_index_weights = (clean_index.values)[clean_values_from_weights]
cleaned_strategy_weights = strategy_weights[clean_values_from_weights]
strategy_log_returns, strategy_daily_returns = wb.bets_to_pnl(
params['STARTING VALUE'], strategy_weights,
clean_values_from_weights, base_data, params['LONG LEVERAGE'],
params['SHORT LEVERAGE'], params['ADV THRESHOLD PERCENTAGE'],
params['COST THRESHOLD BPS'], params['COMMISSION BPS'])
if np.sum(clean_values_from_weights) != 0:
rf_returns = datasets_dict['rf_rate'][clean_values_from_weights]
else:
rf_returns = datasets_dict['rf_rate']
# cleaned_index = cleaned_index_weights
if np.sum(strategy_daily_returns) == 0:
return 0
else:
excess_returns = strategy_daily_returns - rf_returns
returns_less_thresh = excess_returns - (((100)**(1 / 252)) - 1)
numer = sum(returns_less_thresh[returns_less_thresh > 0.0])
denom = -1.0 * sum(returns_less_thresh[returns_less_thresh < 0.0])
fitness = numer / denom
return fitness
def fitness_exp(strategy, params):
strategy_weights = weight.weights(datasets_dict['Sectors'], strategy,
params['NEUTRALIZATION'],
params['LONG LEVERAGE'],
params['SHORT LEVERAGE'])
# print("check")
#Simple Checks
empty = np.full_like(strategy_weights, np.nan)
check = ((strategy_weights == empty) |
(np.isnan(strategy_weights) & np.isnan(empty))).all()
if check:
strategy_weights = np.full_like(strategy_weights, 0)
non_nan_frac = np.count_nonzero(~np.isnan(strategy_weights)) / np.size(
strategy_weights)
if non_nan_frac <= 0.6:
strategy_weights = np.full_like(strategy_weights, 0)
clean_values_from_weights = wb.get_valid_index(strategy_weights)
cleaned_index_weights = (clean_index.values)[clean_values_from_weights]
cleaned_strategy_weights = strategy_weights[clean_values_from_weights]
strategy_log_returns, strategy_daily_returns = wb.bets_to_pnl(
params['STARTING VALUE'], strategy_weights,
clean_values_from_weights, base_data, params['LONG LEVERAGE'],
params['SHORT LEVERAGE'], params['ADV THRESHOLD PERCENTAGE'],
params['COST THRESHOLD BPS'], params['COMMISSION BPS'])
if np.sum(clean_values_from_weights) != 0:
benchmark_returns = datasets_dict['benchmark'][
clean_values_from_weights]
else:
benchmark_returns = datasets_dict['benchmark'][
clean_values_from_weights]
# cleaned_index = cleaned_index_weights
if np.sum(strategy_daily_returns) == 0:
return 0
else:
excess_returns_benchmark = strategy_log_returns - benchmark_returns
fitness = mth.sqrt(252) * np.mean(
excess_returns_benchmark) / np.std(excess_returns_benchmark)
return fitness
