class Trading:
def __init__(self):
self.Training_Done = 0
self.Trading_Required = 0
self.System_Date = datetime(2011, 1, 1, 00, 00)
self.popSize = 10
self.i = 0
self.cash_bal = 10000000
self.asset_bal = 0
self.TradingInfo = pd.DataFrame() #to hold back testing results
self.BackTestingInfo = pd.DataFrame() #to hold back testing results
self.train_dict = None
self.test_dict = None
self.FCPO = None
self.j = 0
self.initMarketRates()
self.ga = GA()
print("in init")
def initMarketRates(self):
print("In Initrates")
self.FCPO = pd.read_csv('FCPO_Aggregated_PerDay_Data.csv')
self.FCPO['Date'] = pd.to_datetime(self.FCPO['Date'])
#FCPO['Date'] = pd.to_datetime(FCPO['Date'], format = "%d/%m/%Y")
#FCPO.set_index('Date', inplace=True)
self.FCPO['Date'] = pd.to_datetime(self.FCPO['Date'],
format="%d/%m/%Y")
self.FCPO.set_index('Date', inplace=True)
self.FCPO = self.FCPO.sort_index()
self.FCPO['isAvailable'] = 0
self.FCPO['isAvailable'].iloc[self.i] = 1
self.FCPO['Sentiment'] = 'Neutral'
self.Update_Sentiment()
# Partition the data to years
self.train = self.FCPO['2011-01-01':'2013-12-31']
self.train_count = self.train.shape[0]
self.test = self.FCPO['2014-01-01':]
self.test_count = self.test.shape[0]
#form a dictionary pair of dataframes for training and testing separately
#training set - divided into 6 equal parts aapprox.
self.train_dict = {
n: self.train.iloc[n:n + round(self.train_count / 6), :]
for n in range(0, len(self.train), round(self.train_count / 6))
}
#testing set - divided into parts of 5 days
self.test_dict = {
n: self.test.iloc[n:n + 5, :]
for n in range(0, len(self.test), 5)
}
print(self.train_dict.keys())
#train_dict.keys()
#len(train_dict)
#for key in train_dict.keys
def UpdateSysDate(self):
print("in update date")
self.System_Date += timedelta(days=1)
idx_date = self.FCPO.index
if self.i < len(idx_date) - 1:
self.i = self.i + 1
while (self.System_Date.date() < idx_date[self.i].date()):
self.System_Date += timedelta(days=1)
#increment the system date till we reach the date for which rates data is available
def Update_Sentiment(self):
while self.j < len(self.FCPO) - 1:
wiki = TextBlob(self.FCPO['News'].iloc[self.j])
if (wiki.sentiment.polarity > -0.4
and wiki.sentiment.polarity < 0.3):
self.FCPO['Sentiment'].iloc[self.j] = "Neutral"
elif (wiki.sentiment.polarity >= 0.3):
self.FCPO['Sentiment'].iloc[self.j] = "Positive"
else:
self.FCPO['Sentiment'].iloc[self.j] = "Negative"
def UpdateRateAvailability(self):
self.FCPO['isAvailable'].iloc[self.i] = 1
def BackTesting(self):
result_bt = self.ga.CheckTrade(self.cash_bal, self.tradingRules,
self.asset_bal)
self.UpdateSysDate()
self.UpdateRateAvailability()
print("In BackTesting")
self.BackTestingInfo = pd.concat([self.BackTestingInfo, result_bt])
self.cash_bal = result_bt['CashBal'][len(result_bt) - 1]
self.asset_bal = result_bt['Asset'][len(result_bt) - 1]
#Execute this to call GA and Fuzzy logic
def Training(self, dfTrain):
self.tradingRules = self.ga.selectPop(
self.popSize, dfTrain, self.cash_bal,
self.asset_bal) #pass the dataframe for training
print("in training")
self.Training_Done = 1
#Flag-off that one round of training is completed and be prepared for back-testing
if self.Training_Done == 1:
self.trading_required = 1
def Trading(self):
self.result_trade = self.ga.CheckTrade(self.cash_bal, self.rates,
self.asset_bal)
self.UpdateSysDate()
self.UpdateRateAvailability()
self.TradingInfo = pd.concat([self.TradingInfo, self.result_trade])
self.cash_bal = self.result_trade['CashBal'][len(self.result_trade) -
1]
self.asset_bal = self.result_trade['Asset'][len(self.result_trade) - 1]
def BuildTradingSystem(self):
print("In BuildTradingSystem")
for key in self.train_dict.keys():
self.Training(self.train_dict[key])
self.BackTesting()
self.BackTestingInfo.to_csv("TrainingReport.csv")
#this is for actual trading + Training @ end of every day from 2014- 2016
def EarnMoney(self):
print("In Trading - Earn Money")
for key in self.test_dict.keys():
for self.rates in self.test_dict[key]:
self.Trading()
self.UpdateSysDate()
self.UpdateRateAvailability()
self.Training(self.test_dict[key])
self.TradingInfo.to_csv("TradingReport.csv")
print("End of Automated Trading")
