def findPattern(commodity, crypto, start, end, interval):
dfweekly = arrano.correlations(commodity, crypto, start, end, interval)[1]
dfData = arrano.correlations(commodity, crypto, start, end, interval)[0]
count = 0
avg_skips = []
index_high = []
for i in range(0,dfweekly.shape[0]-1):
if dfweekly.iloc[i]['weekframe'] >= 0.5:
while dfweekly.iloc[i+count+1]['weekframe'] <= 0.5 and i+count+1<dfweekly.shape[0]-1: #Loop runs till the final entry only
count = count +1
avg_skips.append(count)
index_high.append(i)
count = 0
trippledays = []
weekframe = []
for i in index_high:
trippledays.append(dfData.iloc[i*3]['trippledays'])
weekframe.append(dfweekly.iloc[i]['weekframe'])
df = pd.DataFrame().assign(tripple = trippledays, week = weekframe)
array =[]
for i in range(0, df.shape[0]):
if df.iloc[i]['tripple'] == 1:
array.append(df.iloc[i]['week'])
#print(df)
#print (sum(array)/len(array))
#print(sum(avg_skips)/len(avg_skips))
#print(weekframe)
return(sum(avg_skips)/len(avg_skips))
def rawData():
commodity = request.form['comname']
crypto = request.form['cryname']
start = request.form['startday']
end = request.form['endday']
interval = request.form['interval']
cor_type = request.form['cor_type']
arrano.correlations(commodity, crypto, start, end, interval, cor_type).savefig('static/price.png')
arrano.returns_graph(commodity, crypto, start, end, interval, cor_type).savefig('static/returns.png')
return render_template("kviewData.html", column_names=df_commodity.columns.values, column_names_1=df_crypto.columns.values, row_data=list(df_commodity.values.tolist()), row_data_1=list(df_crypto.values.tolist()), zip = zip)
def inputting(commodity, crypto, start, end, interval):
dfCorr = arrano.correlations(commodity, crypto, start, end, interval)
dfChanges = helper(commodity, crypto, start, end, interval)
dfCorr = dfCorr.drop(dfCorr.index[-1:])
dfavgPrice_change = avgPrices(commodity, crypto, start, end, interval)
dfavgPrice_change = dfavgPrice_change.drop(dfavgPrice_change.index[-1:])
dfHighLow = getBestTrades(commodity, crypto, start, end, interval)
dfHighLow = dfHighLow.drop(dfHighLow.index[-1:])
data = pd.concat([dfCorr, dfChanges, dfavgPrice_change, dfHighLow], axis = 1)
print(data)
indexes = []
data = data[(data['weekly']>=0.5) | (data['weekly']<=-0.5)]
data = data[(data['biweekly']>=0.5) | (data['biweekly']<=-0.5)]
positive_to_positive = []
positive_to_negative = []
negative_to_positive = []
negative_to_negative = []
for index, row in data.iterrows():
if 0.5<row['weekly'] and 0.5<row['biweekly']:
positive_to_positive.append(index)
elif row['weekly']<-0.5 and row['biweekly']<-0.5:
positive_to_negative.append(index)
elif row['weekly']<-0.5 and row['biweekly']>0.5:
negative_to_positive.append(index)
else:
negative_to_negative.append(index)
#print(data.iloc[i]['weekly'])
print(negative_to_negative)
print(negative_to_positive)
print(positive_to_negative)
print(positive_to_positive)
def correlations_analysis(commodity, crypto, start, end, interval):
corr = arrano.correlations(commodity, crypto, start, end, interval)
positive_relation_consecutive = []
negative_relation_consecutive = []
relations_change_inverse = []
no_imp_change = []
for i in range(0,len(corr)-1):
if corr[i]>0.5:
c = 1
while corr[i+c]>0.5:
c = c +1
positive_relation_consecutive.append(c)
c = 1
elif corr[i]<-0.5:
c = 1
while corr[i+c]<-0.5:
c = c+1
negative_relation_consecutive.append(c)
c = 1
print(sum(positive_relation_consecutive)/len(corr))
print(negative_relation_consecutive)
print(no_imp_change)
print(len(corr))
def final(commodity, crypto, start, end, interval):
skips = arrano_new.findPattern(commodity, crypto, start, end, interval) #Returns the average skips before prices highly correlated.
skips = round(skips)
dfData = arrano.correlations(commodity, crypto, start, end, interval)[0] # Gives 9 day correlation and 3 day correlations
df1 = arrano.crypto_p(crypto, start, end) # Rteurns a dataframe of OHLC values of crypto
df2 = arrano.yahoo_fina(commodity, start, end, interval) # Rteurns a dataframe of OHLC values of crypto
df3 = avgCrypto(commodity, crypto, start, end, interval) # Rteurns a dataframe of average prices of crypto for a 9 day range
df4 = trader.getBestTrades(commodity, crypto, start, end, interval) # Returns a data frame of High and Low prices of crypto for a 9 day range.
cryArrayT = []
comArrayT = []
index_date = []
"""
Running a loop on the dataset using the average skips to find trading weeks.
"""
for i in range(9, df1.shape[0] - df1.shape[0]%(9*skips), skips*9): #Running loop till values divisible by 9 and skips.
cryArrayT.append((df1.iloc[i +2]['closePriceUsd']-df1.iloc[i]['closePriceUsd'])/df1.iloc[i+2]['closePriceUsd'])
comArrayT.append((df2.iloc[i +2]['close']-df2.iloc[i]['close'])/df2.iloc[i +2]['close'])
df = pd.DataFrame().assign(cryT = cryArrayT, comT = comArrayT, tripplesCorr = dfData[['trippledays']])
semiPredict = []
"""
Finding the change percentage by multiplying correlation value and change in price in the 3 day time frame.
"""
for i in range(0, df.shape[0]):
if df.iloc[i]['tripplesCorr'] > 0 and df.iloc[i]['cryT']>0:
trade = df.iloc[i]['cryT']*df.iloc[i]['tripplesCorr']
semiPredict.append(trade)
index_date.append(i)
print(len(semiPredict))
final = []
c = 0
"""
Using the index of trading week, find the previous week average and predict deviated prices based on the calculated changes.
"""
for i in index_date:
trades = (df3.iloc[(i-1)*skips]['crypto']*semiPredict[c]) + df3.iloc[(i-1)*skips]['crypto']
final.append(trades)
c = c+1
lowCry = []
highCry = []
"""
Getting trading week high and low prices to understand whether prediction lies in range.
"""
for i in index_date:
lowCry.append(df4.iloc[i*skips]['lowCrypto'])
highCry.append(df4.iloc[i*skips]['highCrypto'])
dataFrame = pd.DataFrame().assign(tradesFinal = final, lowWeek = lowCry, highWeek = highCry)
fig, ax = plt.subplots()
ax.set_xlabel('Date')
ax.set_ylabel('Prices')
ax.plot(dataFrame['lowWeek'], label = 'low')
ax.plot(dataFrame['highWeek'], label = 'high')
ax.plot(dataFrame['tradesFinal'], label = 'predicted')
plt.legend(loc="upper left")
plt.show()
def dualdayCorr(commodity, crypto, start, end, interval):
df1 = avgPrices(commodity, crypto, start, end, interval)
df2 = arrano.correlations(commodity, crypto, start, end, interval)
data = pd.concat([df1, df2], axis = 1, sort = False)
print(data)