def simulate():
'''
strategy_params=dict(top=60,btm=40,rsi=14,average=111,duration=5))
+--------+-----+-------+------+---------+---------+---------+---------+
| _score | cnt | w_pct | pr | pnl | mtm_pnl | max_equ | max_dd |
+--------+-----+-------+------+---------+---------+---------+---------+
| 5.70 | 97 | 0.61 | 1.25 | 7163.57 | 7163.57 | 7608.81 | 1129.33 |
+--------+-----+-------+------+---------+---------+---------+---------+
equity curve = 570x.png
strategy_params=dict(top=60,btm=40,rsi=12,average=180,duration=7))
+--------+-----+-------+------+----------+----------+----------+---------+
| _score | cnt | w_pct | pr | pnl | mtm_pnl | max_equ | max_dd |
+--------+-----+-------+------+----------+----------+----------+---------+
| 7.56 | 106 | 0.67 | 1.22 | 11139.79 | 11139.79 | 11158.79 | 1330.16 |
+--------+-----+-------+------+----------+----------+----------+---------+
equity curve = 756x.png
'''
m = RSIStrategy(name='rsi2_spy',
strategy_params=dict(top=60,
btm=40,
rsi=12,
average=180,
duration=7))
s = Simulator()
s.add_strategy(m)
s.run(DataFeedList(['SPY.csv'], data_type='D'))
s.write('rsi2_SPY_%s' % datetime.now().strftime("%Y%m%d"))
def simulate_AAPL_SPY():
m = RSIStrategy(name='rsi2_aapl_spy',
strategy_params=dict(top=60,
btm=40,
rsi=12,
average=180,
duration=7))
s = Simulator()
s.add_strategy(m)
s.run(DataFeedList(['AAPL.csv', 'SPY.csv'], data_type='D'))
s.write('rsi2_AAPL_SPY_%s' % datetime.now().strftime("%Y%m%d"))
def simulate_SPY_AAPL_composite():
'''
test to make sure that a composite file (AAPL and SPY combined)
is the same as the individuals (AAPL.csv and SPY.csv separately)
'''
m = RSIStrategy(name='rsi2_spy_v_aapl',
strategy_params=dict(top=60,
btm=40,
rsi=12,
average=180,
duration=7))
s = Simulator()
s.add_strategy(m)
s.run(DataFeedList(['SPY_AAPL.csv'], data_type='D'))
s.write('rsi2_SPYvAAPL_%s' % datetime.now().strftime("%Y%m%d"))
def run(self):
run_id = 0
while not self.converged():
params_set = self.generate_set()
for strategy_params in params_set:
log.info('run_id = %d: %s' %
(run_id, pprint.pformat(strategy_params)))
name = '%s_%s' % (self.strategy_class.__name__,
self._tempname())
s = Simulator()
s.reset_on_EOD = self.reset_on_EOD
## if you want have unique portfolio names per optimizer run
## s.portfolio.name = "".join(['portfolio_',name])
s.verbose = self.verbose
s.add_strategy(
self.strategy_class(name,
strategy_setup=self.strategy_setup,
strategy_params=strategy_params))
self.data_feed.reset()
summary = s.run(self.data_feed)
self.score(summary, run_id)
run_id += 1
self.dump(display=self.display_dump)
from Simulator import Simulator
from RetraceStrategy import RetraceStrategy
from DataFeed import DataFeedList
if __name__ == '__main__':
r = RetraceStrategy('tiger',
strategy_params=dict(average=150,
duration=20,
momentum=40))
d = DataFeedList(['tiger_data.csv'], data_type='D')
s = Simulator()
s.add_strategy(r)
s.run(d)
s.write('tiger_150_20_40')
s.show()
def run(self, count=50):
log.info('strategy: %s' % self.strategy_class.__name__)
log.info('strategy_params: %s' % self.strategy_params)
self.best = None
self.shelf = {}
for i in xrange(count):
s = Simulator()
s.reset_on_EOD = self.reset_on_EOD
name = '%s_%s' % (self.strategy_class.__name__, self._tempname())
s.add_strategy(
self.strategy_class(name,
strategy_setup=self.strategy_setup,
strategy_params=self.strategy_params))
self.data_feed.reset()
log.info('\n%d/%d\nStart Strategy= %s' % (i + 1, count, name))
summary = s.run(self.data_feed)
log.info('End Strategy= %s' % name)
key = (summary['cnt'], summary['pnl'], summary['mtm_pnl'],
summary['max_dd'])
self.record[key] += 1
if not self.best or (self.record[key] >= self.best[1]):
self.best = (key, self.record[key], name)
self.shelf[key] = s.portfolio.storage
group = []
columns = ['score', 'cnt', 'trades', 'pnl', 'mtm_pnl', 'max_dd']
p = PrettyTable(columns)
for k, v in self.record.iteritems():
u = dict(cnt=k[0], mtm_pnl=k[2], max_dd=k[3])
score = fitness_function(u)
u.update({'score': score, 'trades': k[0], 'pnl': k[1]})
group.append(u)
p.add_row([score, v, k[0], k[1], k[2], k[3]])
## tack on summary information
df = pandas.DataFrame(group)
means = []
devs = []
for c in columns:
if c in ['score', 'pnl', 'mtm_pnl', 'max_dd']:
p.float_format[c] = '0.2'
p.align[c] = 'r'
m = df[c].mean()
means.append(m)
## calc % of dispersion relative to avg pnl
devs.append(df[c].std() / m)
elif c in ['trades']:
means.append(' ')
devs.append(' ')
else:
means.append('AVG')
devs.append('DSP')
p.add_row(means)
p.add_row(devs)
log.info('\n%s' % p)
log.info('Best Simulation: Strategy= %s' % self.best[2])
## return dataframe - just in case you want further analysis
self.df_out = df
from Simulator import Simulator
from RetraceStrategy import RetraceStrategy
from DataFeed import DataFeedList
from MStrategy2 import MStrategy2
s = Simulator()
'''
s.add_strategy(RetraceStrategy('R2',strategy_params = dict(average=141,momentum=21,duration=15)))
#data_feed = DataFeedList(['daily.SPY.csv','daily.SPY.csv'],data_type='D')
data_feed = DataFeedList(['daily.SPY.csv','daily.SPY.csv'],data_type='D')
s.run(data_feed)
s.write('R2')
## testing DataFeedBar serial files
strat = MStrategy2('M2',strategy_params = dict(length=20,duration=15))
#strat.capture_data = True
s.add_strategy(strat)
s.reset_on_EOD = False
data_feed = DataFeedList(['20081210.SPY.1min.csv','20081211.SPY.1min.csv','20081212.SPY.1min.csv'],data_type='B')
s.run(data_feed)
#strat.dump_data().to_csv('M2_data_dump.csv')
s.write('M2')
'''
## testing DataFeedIntraday serial files
strat = MStrategy2('M2', strategy_params=dict(length=30, duration=50))
#strat.capture_data = True
s.add_strategy(strat)
s.reset_on_EOD = False
#data_feed = DataFeedList(['20081210.SPY.30s.csv','20081211.SPY.30s.csv'],data_type='I')
#data_feed = DataFeedList(['combo.SPY.30s.csv'],data_type='I')
