def __call__(self):
'''
>>> from ppf.date_time import *
>>> import math
>>> from ppf.math.interpolation import loglinear
>>> times = [0.0, 0.5, 1.0, 1.5, 2.0, 3.0, 4.0, 5.0, 6.0, 7.0, 8.0, 9.0, 10.0, 11.0]
>>> factors = [math.exp(-0.05*t) for t in times]
>>> import ppf.market
>>> c = ppf.market.curve(times, factors, loglinear)
>>> from numpy import zeros
>>> expiries = [0.0, 0.5, 1.0, 1.5, 2.0, 3.0, 4.0, 5.0, 6.0, 7.0, 8.0, 9.0, 10.0, 11.0]
>>> tenors = [0, 90]
>>> values = zeros((14, 2))
>>> for i in range(14): values[i, 0] = 0.01
>>> for i in range(14): values[i, 1] = 0.01
>>> surf = ppf.market.surface(expiries, tenors, values)
>>> env = ppf.market.environment(date(2007, Mar, 29))
>>> key = "zc.disc.usd"
>>> env.add_curve(key, c)
>>> key = "ve.term.usd.hw"
>>> env.add_surface(key, surf)
>>> key = "cv.mr.usd.hw"
>>> env.add_constant(key, 0.01)
>>> from ppf.core.pay_receive import *
>>> from ppf.core.generate_flows import *
>>> from ppf.core.generate_observables import *
>>> from ppf.core.generate_exercise_table import *
>>> from ppf.core.exercise_type import *
>>> from ppf.core.leg import *
>>> from ppf.core.trade import *
>>> libor_observables = generate_libor_observables(
... start = date(2007, Jun, 29)
... , end = date(2010, Jun, 29)
... , roll_period = 6
... , roll_duration = ppf.date_time.months
... , reset_period = 6
... , reset_duration = ppf.date_time.months
... , reset_currency = "USD"
... , reset_basis = basis_act_360
... , reset_shift_method = shift_convention.modified_following)
>>> coupon_observables = generate_fixed_coupon_observables(
... start = date(2007, Jun, 29)
... , end = date(2010, Jun, 29)
... , roll_period = 6
... , reset_currency = "USD"
... , coupon_shift_method = shift_convention.modified_following
... , coupon_rate = 0.045)
>>> #semi-annual flows
>>> pay_flows = generate_flows(
... start = date(2007, Jun, 29)
... , end = date(2010, Jun, 29)
... , duration = ppf.date_time.months
... , period = 6
... , notional = 0
... , pay_shift_method = shift_convention.modified_following
... , pay_currency = "USD"
... , basis = "30/360"
... , observables = coupon_observables)
>>> from ppf.core.generate_adjuvant_table import *
>>> from numpy import array
>>> rcv_flows = generate_flows(
... start = date(2007, Jun, 29)
... , end = date(2010, Jun, 29)
... , duration = ppf.date_time.months
... , period = 6
... , pay_shift_method = shift_convention.modified_following
... , pay_currency = "USD"
... , basis = "A/360"
... , observables = libor_observables)
>>> rcv_adjuvant_table = generate_adjuvant_table(
... keys = ["spread0"]
... , tenors = [48]
... , values = array([[0.0]])
... , shift_method = shift_convention.modified_following
... , start_date = date(2007, Jun, 29))
>>> from ppf.pricer import *
>>> pay_leg = leg(pay_flows, PAY, None, payoffs.fixed_leg_payoff())
>>> receive_leg = leg(rcv_flows, RECEIVE, rcv_adjuvant_table, payoffs.float_leg_payoff())
>>> #1y nc
>>> ex_sched = generate_exercise_table(
... start = date(2007, Jun, 29)
... , end = date(2009, Jun, 29)
... , period = 1
... , duration = ppf.date_time.years
... , shift_method = shift_convention.modified_following)
>>> structure = trade((pay_leg, receive_leg))#, (ex_sched, exercise_type.callable))
>>> from ppf.model import *
>>> factory = hull_white_lattice_model_factory()
>>> model = factory(structure, env)
>>> pricer = lattice_pricer(structure, model, env)
>>> print pricer()
1376834.46742
'''
# create controller
ctr = controller(self.__trade, self.__model, self.__env)
times = self.__timeline.times()
from_ = self.__env.relative_date(times[-1]) / 365.0
# initialise symbols
ctr.insert_symbol("underlying", from_)
ctr.insert_symbol("berm", from_)
cnt = 0
for l in self.__trade.legs():
symbol = "leg" + str(cnt)
ctr.insert_symbol(symbol, from_)
cnt += 1
# reverse iterate through the timeline
for i in range(len(times) - 1, -1, -1):
time = times[i]
to_ = 0
if i <> 0:
to_ = self.__env.relative_date(times[i - 1]) / 365.0
events = self.__timeline.events(time)
for event in events:
# set event on controller
ctr.set_event(event)
# evaluate
if is_pay_event(event):
#print 'evaluating leg ', event.leg_id(), ' at ', from_
# evaluate payoff
cpn = ctr(from_)
# rollback symbol
symbol = "leg" + str(event.leg_id())
leg_pv = ctr.retrieve_symbol(symbol)
leg_pv += cpn
self.__symbol_listener_(from_, symbol, leg_pv)
ctr.update_symbol(symbol, leg_pv, from_)
else:
#print 'evaluating berm at ', from_
# evaluate underlying
underlying = ctr.retrieve_symbol("underlying")
underlying *= 0 # not pretty
cnt = 0
for l in self.__trade.legs():
underlying += ctr.retrieve_symbol("leg" + str(cnt))
cnt += 1
self.__symbol_listener_(
from_, "underlying",
self.__trade.exercise_type() * underlying)
# rollback berm
berm = ctr.retrieve_symbol("berm")
self.__symbol_listener_(from_, "berm", berm)
berm = ctr.rollback_max(
from_, to_, berm,
self.__trade.exercise_type() * underlying)
# rollback underlying
underlying = ctr.rollback(from_, to_, underlying)
# update symbols
ctr.update_symbol("underlying", underlying, to_)
ctr.update_symbol("berm", berm, to_)
# rollback any symbols in symbol table not already rolled back
symbols = ctr.retrieve_symbols_to_rollback(to_)
for symbol in symbols:
from_ = ctr.retrieve_symbol_update_time(symbol)
value = ctr.retrieve_symbol(symbol)
value = ctr.rollback(from_, to_, value)
ctr.update_symbol(symbol, value, to_)
from_ = to_
# calculate pv
underlying = ctr.retrieve_symbol("underlying")
underlying *= 0
cnt = 0
for l in self.__trade.legs():
underlying += ctr.retrieve_symbol("leg" + str(cnt))
cnt += 1
ctr.update_symbol("underlying", underlying, to_)
pv = 0
if self.__trade.has_exercise_schedule():
if self.__trade.exercise_type() == exercise_type.callable:
pv = ctr.retrieve_symbol("berm").mean()
else:
pv = ctr.retrieve_symbol(
"underlying").mean() + ctr.retrieve_symbol("berm").mean()
else:
pv = ctr.retrieve_symbol("underlying").mean()
return pv
def __call__(self):
'''
>>> from ppf.date_time import *
>>> import math
>>> from ppf.math.interpolation import loglinear
>>> times = [0.0, 0.5, 1.0, 1.5, 2.0, 3.0, 4.0, 5.0, 6.0, 7.0, 8.0, 9.0, 10.0, 11.0]
>>> factors = [math.exp(-0.05*t) for t in times]
>>> import ppf.market
>>> c = ppf.market.curve(times, factors, loglinear)
>>> from numpy import zeros
>>> expiries = [0.0, 0.5, 1.0, 1.5, 2.0, 3.0, 4.0, 5.0, 6.0, 7.0, 8.0, 9.0, 10.0, 11.0]
>>> tenors = [0, 90]
>>> values = zeros((14, 2))
>>> for i in range(14): values[i, 0] = 0.01
>>> for i in range(14): values[i, 1] = 0.01
>>> surf = ppf.market.surface(expiries, tenors, values)
>>> env = ppf.market.environment(date(2007, Mar, 29))
>>> key = "zc.disc.usd"
>>> env.add_curve(key, c)
>>> key = "ve.term.usd.hw"
>>> env.add_surface(key, surf)
>>> key = "cv.mr.usd.hw"
>>> env.add_constant(key, 0.01)
>>> from ppf.core.pay_receive import *
>>> from ppf.core.generate_flows import *
>>> from ppf.core.generate_observables import *
>>> from ppf.core.generate_exercise_table import *
>>> from ppf.core.exercise_type import *
>>> from ppf.core.leg import *
>>> from ppf.core.trade import *
>>> libor_observables = generate_libor_observables(
... start = date(2007, Jun, 29)
... , end = date(2010, Jun, 29)
... , roll_period = 6
... , roll_duration = ppf.date_time.months
... , reset_period = 6
... , reset_duration = ppf.date_time.months
... , reset_currency = "USD"
... , reset_basis = basis_act_360
... , reset_shift_method = shift_convention.modified_following)
>>> coupon_observables = generate_fixed_coupon_observables(
... start = date(2007, Jun, 29)
... , end = date(2010, Jun, 29)
... , roll_period = 6
... , reset_currency = "USD"
... , coupon_shift_method = shift_convention.modified_following
... , coupon_rate = 0.045)
>>> #semi-annual flows
>>> pay_flows = generate_flows(
... start = date(2007, Jun, 29)
... , end = date(2010, Jun, 29)
... , duration = ppf.date_time.months
... , period = 6
... , notional = 0
... , pay_shift_method = shift_convention.modified_following
... , pay_currency = "USD"
... , basis = "30/360"
... , observables = coupon_observables)
>>> from ppf.core.generate_adjuvant_table import *
>>> from numpy import array
>>> rcv_flows = generate_flows(
... start = date(2007, Jun, 29)
... , end = date(2010, Jun, 29)
... , duration = ppf.date_time.months
... , period = 6
... , pay_shift_method = shift_convention.modified_following
... , pay_currency = "USD"
... , basis = "A/360"
... , observables = libor_observables)
>>> rcv_adjuvant_table = generate_adjuvant_table(
... keys = ["spread0"]
... , tenors = [48]
... , values = array([[0.0]])
... , shift_method = shift_convention.modified_following
... , start_date = date(2007, Jun, 29))
>>> from ppf.pricer import *
>>> pay_leg = leg(pay_flows, PAY, None, payoffs.fixed_leg_payoff())
>>> receive_leg = leg(rcv_flows, RECEIVE, rcv_adjuvant_table, payoffs.float_leg_payoff())
>>> #1y nc
>>> ex_sched = generate_exercise_table(
... start = date(2007, Jun, 29)
... , end = date(2009, Jun, 29)
... , period = 1
... , duration = ppf.date_time.years
... , shift_method = shift_convention.modified_following)
>>> structure = trade((pay_leg, receive_leg))#, (ex_sched, exercise_type.callable))
>>> from ppf.model import *
>>> factory = hull_white_lattice_model_factory()
>>> model = factory(structure, env)
>>> pricer = lattice_pricer(structure, model, env)
>>> print pricer()
1376834.46742
'''
# create controller
ctr = controller(self.__trade, self.__model, self.__env)
times = self.__timeline.times()
from_ = self.__env.relative_date(times[-1])/365.0
# initialise symbols
ctr.insert_symbol("underlying", from_)
ctr.insert_symbol("berm", from_)
cnt = 0
for l in self.__trade.legs():
symbol = "leg"+str(cnt)
ctr.insert_symbol(symbol, from_)
cnt += 1
# reverse iterate through the timeline
for i in range(len(times)-1,-1,-1):
time = times[i]
to_ = 0
if i <> 0:
to_ = self.__env.relative_date(times[i-1])/365.0
events = self.__timeline.events(time)
for event in events:
# set event on controller
ctr.set_event(event)
# evaluate
if is_pay_event(event):
#print 'evaluating leg ', event.leg_id(), ' at ', from_
# evaluate payoff
cpn = ctr(from_)
# rollback symbol
symbol = "leg"+str(event.leg_id())
leg_pv = ctr.retrieve_symbol(symbol)
leg_pv += cpn
self.__symbol_listener_(from_, symbol, leg_pv)
ctr.update_symbol(symbol, leg_pv, from_)
else:
#print 'evaluating berm at ', from_
# evaluate underlying
underlying = ctr.retrieve_symbol("underlying")
underlying *= 0 # not pretty
cnt = 0
for l in self.__trade.legs():
underlying += ctr.retrieve_symbol("leg"+str(cnt))
cnt += 1
self.__symbol_listener_(from_, "underlying", self.__trade.exercise_type()*underlying)
# rollback berm
berm = ctr.retrieve_symbol("berm")
self.__symbol_listener_(from_, "berm", berm)
berm = ctr.rollback_max(from_, to_, berm, self.__trade.exercise_type()*underlying)
# rollback underlying
underlying = ctr.rollback(from_, to_, underlying)
# update symbols
ctr.update_symbol("underlying", underlying, to_)
ctr.update_symbol("berm", berm, to_)
# rollback any symbols in symbol table not already rolled back
symbols = ctr.retrieve_symbols_to_rollback(to_)
for symbol in symbols:
from_ = ctr.retrieve_symbol_update_time(symbol)
value = ctr.retrieve_symbol(symbol)
value = ctr.rollback(from_, to_, value)
ctr.update_symbol(symbol, value, to_)
from_ = to_
# calculate pv
underlying = ctr.retrieve_symbol("underlying")
underlying *= 0
cnt = 0
for l in self.__trade.legs():
underlying += ctr.retrieve_symbol("leg"+str(cnt))
cnt += 1
ctr.update_symbol("underlying", underlying, to_)
pv = 0
if self.__trade.has_exercise_schedule():
if self.__trade.exercise_type() == exercise_type.callable:
pv = ctr.retrieve_symbol("berm").mean()
else:
pv = ctr.retrieve_symbol("underlying").mean()+ctr.retrieve_symbol("berm").mean()
else:
pv = ctr.retrieve_symbol("underlying").mean()
return pv
def __exercise_boundary_regression(self, symbol_value_pairs_to_add):
num_expl_vars = self.__regression_model.exercise().num_explanatory_variables()
num_sims = self.__regression_model.state().num_sims()
num_exercises = self.__timeline.number_of_exercises()
# create controller
ctr = controller(self.__trade, self.__regression_model, self.__env, 1.0)
times = self.__timeline.times()
from_ = 0.0
# initialise symbols
ctr.insert_symbol("underlying", from_)
# add extra symbols
if symbol_value_pairs_to_add:
for symbol_value_pair in symbol_value_pairs_to_add:
symbol, value = symbol_value_pair
ctr.insert_symbol(symbol, value)
cnt = 0
for l in self.__trade.legs():
symbol = "leg"+str(cnt)
ctr.insert_symbol(symbol, from_)
cnt += 1
# forward iterate through the timeline
vs = numpy.zeros([num_exercises, num_sims, num_expl_vars])
ies = numpy.zeros([num_exercises, num_sims])
ns = numpy.zeros([num_exercises, num_sims]) # for fees - ignored for present
normalisation = 1.0/self.__trade.legs()[0].flows()[0].notional()
ex_cnt = 0
for time in times:
to_ = self.__env.relative_date(time)/365.0
# evolve
ctr.evolve(from_, to_)
events = self.__timeline.events(time)
# evaluate explanatory variables and immediate exercise values
for event in events:
# set event on controller
ctr.set_event(event)
# evaluate
if is_exercise_event(event):
# evaluate underlying
underlying = ies[ex_cnt, :]
cnt = 0
for l in self.__trade.legs():
underlying += ctr.retrieve_symbol("leg"+str(cnt))
cnt += 1
underlying *= self.__trade.exercise_type()
underlying *= normalisation
# evaluate explanatory variables and numeraire
ns[ex_cnt, :] = ctr.numeraire(to_)
vs[ex_cnt, :] = ctr.explanatory_variables(to_)
ex_cnt += 1
# evaluate cash flows
for event in events:
# set event on controller
ctr.set_event(event)
# evaluate
if is_pay_event(event):
# evaluate payoff
cpn = ctr(to_)
symbol = "leg"+str(event.leg_id())
leg_pv = ctr.retrieve_symbol(symbol)
leg_pv += cpn
self.__symbol_listener_(to_, symbol, leg_pv)
ctr.update_symbol(symbol, leg_pv, to_)
from_ = to_
# final immediate exercise value
underlying = ctr.retrieve_symbol("underlying")
cnt = 0
for l in self.__trade.legs():
underlying += ctr.retrieve_symbol("leg"+str(cnt))
cnt += 1
underlying *= self.__trade.exercise_type()
underlying *= normalisation
# subtract immediate exercise values from final
for i in range(num_exercises):
ies[i, :]=underlying-ies[i, :]
# perform regression
self.__fitted_fos = pickup_value_regression(ies, ns, vs)
# create helper class for dealing with exercise indicator - note number of simulations
self.__exercise_helper = exercise_helper(self.__model.state().num_sims())
def __call__(self, symbol_value_pairs_to_add = None):
# do regression if required
if self.__regression_model:
self.__exercise_boundary_regression(symbol_value_pairs_to_add)
# create controller
ctr = controller(self.__trade, self.__model, self.__env, 1.0)
times = self.__timeline.times()
from_ = 0.0
# initialise symbols
ctr.insert_symbol("underlying", from_)
ctr.insert_symbol("berm", from_)
# add extra symbols
if symbol_value_pairs_to_add:
for symbol_value_pair in symbol_value_pairs_to_add:
symbol, value = symbol_value_pair
ctr.insert_symbol(symbol, value)
cnt = 0
for l in self.__trade.legs():
symbol = "leg"+str(cnt)
ctr.insert_symbol(symbol, from_)
cnt += 1
# forward iterate through the timeline
ex_cnt = 0
for time in times:
to_ = self.__env.relative_date(time)/365.0
# evolve
ctr.evolve(from_, to_)
events = self.__timeline.events(time)
# set initial immediate exercise value - sum of all flows before exercise date
for event in events:
if is_exercise_event(event):
# evaluate underlying
underlying = ctr.retrieve_symbol("underlying")
underlying *= 0
cnt = 0
for l in self.__trade.legs():
underlying += ctr.retrieve_symbol("leg"+str(cnt))
cnt += 1
if ex_cnt == 0:
self.__exercise_helper.set_last_cfs(self.__trade.exercise_type()*underlying)
self.__symbol_listener_(to_, "underlying", underlying)
# evaluate cash flows
for event in events:
# set event on controller
ctr.set_event(event)
# evaluate
if is_pay_event(event):
# evaluate payoff
cpn = ctr(to_)
symbol = "leg"+str(event.leg_id())
leg_pv = ctr.retrieve_symbol(symbol)
leg_pv += cpn
self.__symbol_listener_(to_, symbol, leg_pv)
ctr.update_symbol(symbol, leg_pv, to_)
# evaluate exercise using regression
for event in events:
# set event on controller
ctr.set_event(event)
# evaluate
if is_exercise_event(event):
# evaluate underlying
underlying = ctr.retrieve_symbol("underlying")
underlying *= 0
cnt = 0
for l in self.__trade.legs():
underlying += ctr.retrieve_symbol("leg"+str(cnt))
cnt += 1
# explanatory variables and numeraire
ns = ctr.numeraire(to_) # for fees but not used at present
vs = ctr.explanatory_variables(to_)
# evaluate regression
self.__exercise_helper.update_indicator(to_, vs, self.__fitted_fos[ex_cnt])
berm = ctr.retrieve_symbol("berm")
berm = self.__exercise_helper.max(to_, self.__trade.exercise_type()*underlying, berm)
# update symbols
ctr.update_symbol("underlying", underlying, to_)
ctr.update_symbol("berm", berm, to_)
ex_cnt = ex_cnt+1
from_ = to_
# calculate pv
underlying = ctr.retrieve_symbol("underlying")
underlying *= 0
cnt = 0
for l in self.__trade.legs():
underlying += ctr.retrieve_symbol("leg"+str(cnt))
cnt += 1
self.__symbol_listener_(to_, "underlying", underlying)
ctr.update_symbol("underlying", underlying, to_)
if self.__regression_model:
berm = ctr.retrieve_symbol("berm")
berm = self.__exercise_helper.max(to_, self.__trade.exercise_type()*underlying, berm)
ctr.update_symbol("berm", berm, to_)
pv = 0
if self.__trade.has_exercise_schedule():
if self.__trade.exercise_type() == exercise_type.callable:
pv = ctr.retrieve_symbol("berm").mean()
else:
pv = ctr.retrieve_symbol("underlying").mean()+ctr.retrieve_symbol("berm").mean()
else:
pv = ctr.retrieve_symbol("underlying").mean()
return pv
def __exercise_boundary_regression(self, symbol_value_pairs_to_add):
num_expl_vars = self.__regression_model.exercise(
).num_explanatory_variables()
num_sims = self.__regression_model.state().num_sims()
num_exercises = self.__timeline.number_of_exercises()
# create controller
ctr = controller(self.__trade, self.__regression_model, self.__env,
1.0)
times = self.__timeline.times()
from_ = 0.0
# initialise symbols
ctr.insert_symbol("underlying", from_)
# add extra symbols
if symbol_value_pairs_to_add:
for symbol_value_pair in symbol_value_pairs_to_add:
symbol, value = symbol_value_pair
ctr.insert_symbol(symbol, value)
cnt = 0
for l in self.__trade.legs():
symbol = "leg" + str(cnt)
ctr.insert_symbol(symbol, from_)
cnt += 1
# forward iterate through the timeline
vs = numpy.zeros([num_exercises, num_sims, num_expl_vars])
ies = numpy.zeros([num_exercises, num_sims])
ns = numpy.zeros([num_exercises,
num_sims]) # for fees - ignored for present
normalisation = 1.0 / self.__trade.legs()[0].flows()[0].notional()
ex_cnt = 0
for time in times:
to_ = self.__env.relative_date(time) / 365.0
# evolve
ctr.evolve(from_, to_)
events = self.__timeline.events(time)
# evaluate explanatory variables and immediate exercise values
for event in events:
# set event on controller
ctr.set_event(event)
# evaluate
if is_exercise_event(event):
# evaluate underlying
underlying = ies[ex_cnt, :]
cnt = 0
for l in self.__trade.legs():
underlying += ctr.retrieve_symbol("leg" + str(cnt))
cnt += 1
underlying *= self.__trade.exercise_type()
underlying *= normalisation
# evaluate explanatory variables and numeraire
ns[ex_cnt, :] = ctr.numeraire(to_)
vs[ex_cnt, :] = ctr.explanatory_variables(to_)
ex_cnt += 1
# evaluate cash flows
for event in events:
# set event on controller
ctr.set_event(event)
# evaluate
if is_pay_event(event):
# evaluate payoff
cpn = ctr(to_)
symbol = "leg" + str(event.leg_id())
leg_pv = ctr.retrieve_symbol(symbol)
leg_pv += cpn
self.__symbol_listener_(to_, symbol, leg_pv)
ctr.update_symbol(symbol, leg_pv, to_)
from_ = to_
# final immediate exercise value
underlying = ctr.retrieve_symbol("underlying")
cnt = 0
for l in self.__trade.legs():
underlying += ctr.retrieve_symbol("leg" + str(cnt))
cnt += 1
underlying *= self.__trade.exercise_type()
underlying *= normalisation
# subtract immediate exercise values from final
for i in range(num_exercises):
ies[i, :] = underlying - ies[i, :]
# perform regression
self.__fitted_fos = pickup_value_regression(ies, ns, vs)
# create helper class for dealing with exercise indicator - note number of simulations
self.__exercise_helper = exercise_helper(
self.__model.state().num_sims())
def __call__(self, symbol_value_pairs_to_add=None):
# do regression if required
if self.__regression_model:
self.__exercise_boundary_regression(symbol_value_pairs_to_add)
# create controller
ctr = controller(self.__trade, self.__model, self.__env, 1.0)
times = self.__timeline.times()
from_ = 0.0
# initialise symbols
ctr.insert_symbol("underlying", from_)
ctr.insert_symbol("berm", from_)
# add extra symbols
if symbol_value_pairs_to_add:
for symbol_value_pair in symbol_value_pairs_to_add:
symbol, value = symbol_value_pair
ctr.insert_symbol(symbol, value)
cnt = 0
for l in self.__trade.legs():
symbol = "leg" + str(cnt)
ctr.insert_symbol(symbol, from_)
cnt += 1
# forward iterate through the timeline
ex_cnt = 0
for time in times:
to_ = self.__env.relative_date(time) / 365.0
# evolve
ctr.evolve(from_, to_)
events = self.__timeline.events(time)
# set initial immediate exercise value - sum of all flows before exercise date
for event in events:
if is_exercise_event(event):
# evaluate underlying
underlying = ctr.retrieve_symbol("underlying")
underlying *= 0
cnt = 0
for l in self.__trade.legs():
underlying += ctr.retrieve_symbol("leg" + str(cnt))
cnt += 1
if ex_cnt == 0:
self.__exercise_helper.set_last_cfs(
self.__trade.exercise_type() * underlying)
self.__symbol_listener_(to_, "underlying", underlying)
# evaluate cash flows
for event in events:
# set event on controller
ctr.set_event(event)
# evaluate
if is_pay_event(event):
# evaluate payoff
cpn = ctr(to_)
symbol = "leg" + str(event.leg_id())
leg_pv = ctr.retrieve_symbol(symbol)
leg_pv += cpn
self.__symbol_listener_(to_, symbol, leg_pv)
ctr.update_symbol(symbol, leg_pv, to_)
# evaluate exercise using regression
for event in events:
# set event on controller
ctr.set_event(event)
# evaluate
if is_exercise_event(event):
# evaluate underlying
underlying = ctr.retrieve_symbol("underlying")
underlying *= 0
cnt = 0
for l in self.__trade.legs():
underlying += ctr.retrieve_symbol("leg" + str(cnt))
cnt += 1
# explanatory variables and numeraire
ns = ctr.numeraire(to_) # for fees but not used at present
vs = ctr.explanatory_variables(to_)
# evaluate regression
self.__exercise_helper.update_indicator(
to_, vs, self.__fitted_fos[ex_cnt])
berm = ctr.retrieve_symbol("berm")
berm = self.__exercise_helper.max(
to_,
self.__trade.exercise_type() * underlying, berm)
# update symbols
ctr.update_symbol("underlying", underlying, to_)
ctr.update_symbol("berm", berm, to_)
ex_cnt = ex_cnt + 1
from_ = to_
# calculate pv
underlying = ctr.retrieve_symbol("underlying")
underlying *= 0
cnt = 0
for l in self.__trade.legs():
underlying += ctr.retrieve_symbol("leg" + str(cnt))
cnt += 1
self.__symbol_listener_(to_, "underlying", underlying)
ctr.update_symbol("underlying", underlying, to_)
if self.__regression_model:
berm = ctr.retrieve_symbol("berm")
berm = self.__exercise_helper.max(
to_,
self.__trade.exercise_type() * underlying, berm)
ctr.update_symbol("berm", berm, to_)
pv = 0
if self.__trade.has_exercise_schedule():
if self.__trade.exercise_type() == exercise_type.callable:
pv = ctr.retrieve_symbol("berm").mean()
else:
pv = ctr.retrieve_symbol(
"underlying").mean() + ctr.retrieve_symbol("berm").mean()
else:
pv = ctr.retrieve_symbol("underlying").mean()
return pv
