L6

cryptoPredict.py

@@ -0,0 +1,127 @@
+import requests
+import time
+import datetime
+import numpy as np
+import statsmodels.api as sm
+from sklearn.linear_model import LinearRegression
+import random 
+import matplotlib.pyplot as plt
+from itertools import chain
+
+def growthOrLoss(probability):
+    rand = random.uniform(0,1)
+    if rand <= probability:
+        return 1
+    else:
+        return -1
+
+def simulate(iterations, x, reg, dt):
+    sim_result = []
+    for i in range(0, iterations):
+        sim_day = []
+        y = reg.predict(x)
+        y = y.tolist()
+        x = x.tolist()
+        x = x[0]
+        sim_day.append(1)
+        sim_day.append(abs(y[0][0]))
+        sim_day.append(abs(y[0][1]))
+        sim_day.append(growthOrLoss(y[0][2])*abs(y[0][0])*x[3]+x[3])
+        sim_day.append(growthOrLoss(y[0][3])*abs(y[0][1])*x[4]+x[4])
+        sim_day.append(sim_day[2]/sim_day[3])
+        x = sim_day
+        x = np.matrix(x)
+        sim_day.append(dt)
+        dt += 86400
+        sim_result.append(sim_day)
+    return sim_result
+
+def toMean(multi_simulation, iterations):
+    meanSimulation = []
+    for j in range (0,iterations):
+        sum_sim = 0
+        for i in range (0, 100):
+            sum_sim = sum_sim + multi_simulation[i][j:j+1,[3,6]].tolist()[0][0]
+        meanSimulation.append(sum_sim/100)
+    return meanSimulation
+
+def run_simulation(currency, dateFrom,dateTo):
+    df = time.mktime(datetime.datetime.strptime(dateFrom, "%d/%m/%Y").timetuple())
+    dt = time.mktime(datetime.datetime.strptime(dateTo, "%d/%m/%Y").timetuple())
+    if currency == "BTC":
+        parameters = {"pair": "XBTUSD", "interval": 1440, "since": df}
+    elif currency == "ETH" or "DAI":
+        parameters = {"pair": currency+"USDT", "interval": 1440, "since": df}
+    else:
+        exit(0)
+    response = requests.get("https://api.kraken.com/0/public/OHLC", params = parameters)
+    data = response.json()
+    if currency == "BTC":
+        result = data['result']['XXBTZUSD']
+    else:
+        result = data['result'][currency+'USDT']
+    dataset = []
+    x_axis = []
+    y_axis = []
+    i = 0
+    for record in result:
+        if record[0] > dt:
+            break
+        data_day = []
+        x_axis.append(record[0])
+        y_axis.append(float(record[4]))
+        data_day.append(1)
+        data_day.append(abs((float(record[4]) - float(record[1]))/float(record[1]))) 
+        if i > 0:
+            data_day.append(abs((dataset[i-1][6] - float(record[6]))/dataset[i-1][6]))
+        else:
+            data_day.append(0)
+        if i > 0 and ((float(record[4]) - float(record[1]))/float(record[1])) > 0:
+            data_day.append(1)
+        else:
+            data_day.append(0)
+        if i > 0 and ((dataset[i-1][6] - float(record[6]))/dataset[i-1][6]) > 0:
+            data_day.append(1)
+        else:
+            data_day.append(0)
+        data_day.append(float(record[4]))
+        data_day.append(float(record[6]))
+        data_day.append(float(record[4])/float(record[6]))
+        dataset.append(data_day)
+        i = i+1
+
+    D = np.matrix(dataset)
+    D = np.squeeze(np.asarray(D))
+    y = D[1:,[1,2,3,4]]
+    X = D[:-1,[0,1,2,5,6,7]]
+    regressor = LinearRegression()
+    model = regressor.fit(X, y)
+
+    X_pred = D[-1:,[0,1,2,5,6,7]]
+    iterations = int((dt-df)/86400)
+    single_simulation = np.matrix(simulate(iterations, X_pred, model, dt))
+
+    multi_simulation = []
+    for i in range (0,100):
+        multi_simulation.append(np.matrix(simulate(iterations, X_pred, model, dt)))
+    meanSimulation = toMean(multi_simulation, iterations)
+
+    second_x_axis = single_simulation[:,-1:]
+    second_x_axis = second_x_axis.tolist()
+    single_simulation = single_simulation[:,[3,4]]
+    single_simulation_values = single_simulation[:,[0]].tolist()
+    single_simulation_volumes = single_simulation[:,[1]].tolist()
+    second_x_axis = list(chain.from_iterable(second_x_axis))
+    single_simulation_values = list(chain.from_iterable(single_simulation_values))
+    single_simulation_volumes = list(chain.from_iterable(single_simulation_volumes))
+    plt.plot(x_axis, y_axis, label = "Real")
+    plt.plot(second_x_axis, single_simulation_values, label = "Single simulation")
+    plt.plot(second_x_axis, meanSimulation, label = "Mean simulation")
+    plt.legend(loc="lower center")
+
+    plt.show()
+
+if __name__ == "__main__":
+    #BTC ETH OR DAI 
+    run_simulation("BTC","01/02/2020","01/06/2020")
+