JustinGuese

apiVersion: apps/v1
kind: Deployment
metadata:
  labels:
    service: storagenode
  name: storagenode
  namespace: storj
spec:
  replicas: 1
  selector:

JustinGuese

def getDecision(dfrow):
    if dfrow['trend_aroon_ind'] <= 10.0:
        if dfrow['volatility_dcp'] <= 0.69:
            if dfrow['volatility_dcp'] <= 0.46:
                return 0
            else:  # if dfrow['volatility_dcp'] > 0.46
                if dfrow['trend_aroon_down'] <= 34.0:
                    return 1
                else:  # if dfrow['trend_aroon_down'] > 34.0
                    return 0

JustinGuese

from sklearn.tree import _tree

def tree_to_code(tree, feature_names):
    tree_ = tree.tree_
    feature_name = [
        feature_names[i] if i != _tree.TREE_UNDEFINED else "undefined!"
        for i in tree_.feature
    ]
    feature_names = [f.replace(" ", "_")[:-5] for f in feature_names]
    print("def getDecision(dfrow):")

JustinGuese

import graphviz 
dot_data = export_graphviz(clf, out_file=None,
                        feature_names=X.columns,  
                        class_names=["sell", "buy"],
                        filled=True, rounded=True,
                        special_characters=True) 
graph = graphviz.Source(dot_data) 
graph

JustinGuese

from tqdm import tqdm

bestDepth = -1
bestLeafes = -1
bestLookback = -1
bestDepthWin = -9999
collection = []
# explicitly force a max of 20 to prevent overfitting
for depth in tqdm([100,70,50,40,30,20,15]):
    for leafes in [100,70,50,40,30,20,15]:

JustinGuese

bigDf = []
for ticker in ["CWEG.L", "IWDA.AS", "EEM", "AAPL", "MSFT", "GOOG", "TSLA", 'AMD', 'AMZN', 'DG', "ETH-USD", "BTC-USD"]:
    print("geddin stock: " + ticker)
    tmp = getData(ticker = ticker, start = date(2010,1,1))
    tmp = getTrend(tmp)
    bigDf.append(tmp)
    
bigDf = pd.concat(bigDf)
bigDf.shape

JustinGuese

from sklearn import metrics, DecisionTreeClassifier

X, Y = df.drop(["signal"], axis=1), df["signal"]

clf = DecisionTreeClassifier()
clf.fit(X, Y)
preds = clf.predict(X)
print(metrics.classification_report(Y, preds))
print(metrics.confusion_matrix(Y, preds))

JustinGuese

from random import random
class FakeModel:
    def predict(self, X):
        # random decision for demo purposes
        preds = []
        for i in range(len(X)):
            if random() > .5:
                preds.append(1)
            else:
                preds.append(-1)

JustinGuese

def oneRun(model):
    global X
    bestLookbackWin = -9999
    bestLookbackPortfolio = []
    money = startMoney
    nrStocks = 0
    portfolio = []
    for i in range(len(X)):
        prednow = model.predict(X.iloc[i])[0]
        print("prednow is, " , prednow)

JustinGuese

startMoney = 10000
COMMISSION = 0.00025 # interactive brokers commission
howmany = startMoney / msft.iloc[0]["Adj Close"]
win = howmany * msft.iloc[-1]["Adj Close"] - startMoney

days = (msft.index[-1] - msft.index[0]).days

print("with just holding you would have made %.2f$" % win)
winPerMonth = win / (days / 30)
winPctPerYear = winPerMonth * 12 / startMoney * 100