dkav9

def remove_similar_sentences(sentences, model, similarity_threshold):
  """
  model: the nlp model to use from the sentence transformer library
  sentences: a list of sentences from a website
  Returns a list of new sentences which are not too similar to each other
  """
  new_sentences = sentences.copy()
  # Compute embeddings
  embeddings = model.encode(new_sentences, device='cpu', show_progress_bar=False)
  # Compute cosine-similarities for each sentence with each other sentence

dkav9

def get_documents(urls):
  """
  Takes a list of urls, gets unique sentences from each website,
  and returns all of those sentences in a single list
  """  
  documents = []
  for url in tqdm(urls):  
    sentences = get_sentences(url)
    new_sentences = remove_similar_sentences(sentences, model, similarity_threshold)
    documents.append(new_sentences)

dkav9

from backtesting import Strategy
from backtesting.lib import crossover
from backtesting import Backtest
from yfinance import Ticker
import pandas as pd
from talib import EMA, SMA, WMA

dkav9

class EmaCross(Strategy):
    # Define the two EMA lags as *class variables*
    # for later optimization
    n1 = 5
    n2 = 10
    
    def init(self):
        # Precompute two moving averages
        self.ema1 = self.I(EMA, self.data.Close, self.n1)
        self.ema2 = self.I(EMA, self.data.Close, self.n2)

dkav9

def get_data(pair):
  ticker = Ticker(pair)
  df = ticker.history(period='3y', rounding=True)[['Open','High','Low','Close']]
  return df

dkav9

df = get_data('ETH-USD')
df = df/1000
strategies = [SmaCross, EmaCross, WmaCross]
all_runs = []
for strategy in strategies:
  bt = Backtest(df, strategy, cash=100, commission=.002)
  stats = bt.optimize(n1=range(5, 90, 5),
                      n2=range(10, 120, 5),
                      maximize='Return (Ann.) [%]',
                      constraint=lambda param: param.n1 < param.n2)

dkav9

from ccxt import coinbasepro

def get_client():
    client = coinbasepro({
        'apiKey': '',
        'secret': '',
        'password': '',
        'enableRateLimit': True,
    })
    return client

dkav9

def buy_max(pair, client=client):
    # Use max possible usd size
    price = client.fetchTicker(pair)['last']
    size = client.fetchBalance()['USD']['free'] / price
    client.createMarketBuyOrder(pair, amount=size)
    print(f"bought ${size} of {pair}")

def sell_max(pair, client=client):
    symbol = pair.split("-")[0]
    size = client.fetchBalance()[symbol]['free']

dkav9

def run(request):
    alert = request.get_json()
    # # If alert request is a buy, we buy
    symbol = alert['ticker']
    pair = symbol + "-USD"
    side = alert['side']
    usd_price = client.fetch_ticker(pair)['last']
    position = client.fetch_balance()[symbol]['free']
    size = usd_price * position
    if side == 'buy' and size < 1:

dkav9

def get_ohlc(ticker, period='60d', interval='5m'):
  cols = ['Open','High','Low','Close']
  ohlc = Ticker(ticker).history(period=period, interval=interval)[cols][:-1]
  return ohlc

df = get_ohlc('ETH-USD')