def momentum_vote(self, prices_to_date):
rolling_momentum = indicators.rolling_momentum(prices_to_date[-(self.mom_window + 2):], self.mom_window).ix[-1]
if rolling_momentum > 0.32:
return 1
if rolling_momentum < 0.32:
return -1
return 0
def testPolicy(self, symbol = "IBM", \
sd=dt.datetime(2009,1,1), \
ed=dt.datetime(2010,1,1), \
sv = 10000):
# here we build a fake set of trades
# your code should return the same sort of data
start = sd - dt.timedelta(days=60)
dates = pd.date_range(start, ed)
data = ut.get_data([symbol], dates)
data.fillna(method="ffill", inplace=True)
data.fillna(method="bfill", inplace=True)
prices = indicators.normalized_prices(data[symbol])
trades = pd.DataFrame(index=prices[sd:].index, columns=[symbol])
self.rsi = indicators.rolling_rsi(prices, 5)
self.bollinger_bands = indicators.bollinger_bands(prices, 15)
self.momentum = indicators.rolling_momentum(prices, 2)
action = 2
holding = 0
for i in range(len(trades)):
prices_to_date = prices[:trades.index[i]]
state = self.create_state(trades.index[i], prices_to_date, action)
action = self.learner.querysetstate(state)
if action is 0:
trades.ix[i] = -1000 - holding
holding = -1000
if action is 1:
trades.ix[i] = 1000 - holding
holding = 1000
if action is 2:
trades.ix[i] = 0
return trades
def testPolicy(self, symbol, sd, ed, sv):
dates = pd.date_range(sd, ed)
data = util.get_data([symbol], dates)
data.fillna(method="ffill", inplace=True)
data.fillna(method="bfill", inplace=True)
prices = indicators.normalized_prices(data[symbol])
trades = pd.DataFrame(index=prices.index, columns=[symbol])
trades[:] = 0
momentum = indicators.rolling_momentum(prices, 2)
buy_days = []
sell_days = []
# when momentum is positive, buy on the last day its positive and vice versas
for i in range(1, len(momentum) -1):
if momentum[i] >=0 and momentum[i+1] < 0:
sell_days.append(i)
elif momentum[i] <=0 and momentum[i+1] > 0:
buy_days.append(i)
if buy_days[0] < sell_days[0]:
trades.ix[0] = -1000
else:
trades.ix[0] = 1000
for i in buy_days:
trades.ix[i] = 2000
for i in sell_days:
trades.ix[i] = -2000
return trades
def addEvidence(self, symbol = "IBM", \
sd=dt.datetime(2008,1,1), \
ed=dt.datetime(2009,1,1), \
sv = 10000):
self.cash = sv
start = sd - dt.timedelta(days=60)
dates = pd.date_range(start, ed)
data = ut.get_data([symbol], dates)
data.fillna(method="ffill", inplace=True)
data.fillna(method="bfill", inplace=True)
prices = indicators.normalized_prices(data[symbol])
self.bollinger_bands = None
self.momentum = None
self.rsi = indicators.rolling_rsi(prices, 5)
self.bollinger_bands = indicators.bollinger_bands(prices, 15)
self.momentum = indicators.rolling_momentum(prices, 2)
self.mom_mean = self.momentum.mean()
self.mom_std = self.momentum.std()
last_action = 2
current_state = self.create_state(sd, prices[:sd], last_action)
action = self.learner.querysetstate(current_state)
trades = pd.DataFrame(index=prices[sd:].index, columns=[symbol])
keepgoing = True
iteration = 0
while keepgoing:
for i in trades.index:
reward = self.get_reward(action, last_action, prices[:i])
last_action = action
current_state = self.create_state(i, prices[:i], last_action)
action = self.learner.query(current_state, reward)
cr = self.cash / sv - 1
if cr == self.cr or iteration == 20:
keepgoing = False
self.cr = cr
iteration += 1