MACD_Algorithm.py

import os
from binance.client import Client
import pandas as pd
import numpy as np
import ta 


# Set the API key and secret
api_key = "key"
api_secret = "key"

# Set the Binance API URL and connect to the client
client = Client(api_key, api_secret)

# Set the directory path where your CSV files are stored
directory_month = '<<insert_directory_here>>'
directory_day = '<<insert_directory_here>>'

# Define the columns to use for scaling and training the model
cols = ['Open', 'High', 'Low', 'Volume', 'Quote asset volume', 'Number of trades', 'Taker buy base asset volume', 'Taker buy quote asset volume','Close']

# Create an empty DataFrame to store the concatenated data
concatenated_df = pd.DataFrame()

# Iterate through each CSV file in the directory and concatenate them
for filename in os.listdir(directory_month):
    if filename.endswith(".csv"):
        # Read the CSV file into a DataFrame
        filepath = os.path.join(directory_month, filename)
        df = pd.read_csv(filepath, header=None, names=['Open time', 'Open', 'High', 'Low', 'Close', 'Volume', 'Close time', 'Quote asset volume', 'Number of trades', 'Taker buy base asset volume', 'Taker buy quote asset volume', 'Ignore'])

        # Append the DataFrame to the concatenated DataFrame
        concatenated_df = pd.concat([concatenated_df, df])

for filename in os.listdir(directory_day):
    if filename.endswith(".csv"):
        # Read the CSV file into a DataFrame
        filepath = os.path.join(directory_day, filename)
        df = pd.read_csv(filepath, header=None, names=['Open time', 'Open', 'High', 'Low', 'Close', 'Volume', 'Close time', 'Quote asset volume', 'Number of trades', 'Taker buy base asset volume', 'Taker buy quote asset volume', 'Ignore'])

        # Append the DataFrame to the concatenated DataFrame
        concatenated_df = pd.concat([concatenated_df, df])

# Get the raw data using Binance API
raw_data = client.get_klines(symbol="SOLUSDT", interval="1d", limit=200)

# Convert raw data into pandas dataframe
live_data = pd.DataFrame(raw_data, columns=['timestamp', 'open', 'high', 'low', 'close', 'volume', 'close_time', 'quote_asset_volume', 'number_of_trades', 'taker_buy_base_asset_volume', 'taker_buy_quote_asset_volume', 'ignore'])

# Convert timestamp to datetime
live_data['timestamp'] = pd.to_datetime(live_data['timestamp'], unit='ms')

# Set timestamp as index
live_data = live_data.set_index('timestamp')

# Drop unnecessary columns from the live data
live_data.drop(['close_time', 'quote_asset_volume', 'number_of_trades', 'taker_buy_base_asset_volume', 'taker_buy_quote_asset_volume', 'ignore'], axis=1, inplace=True)

# Convert the data types of columns to float
live_data = live_data.astype(float)

# Concatenate the historical and live data
data = pd.concat([concatenated_df, live_data], axis=0)

#drop dublicate data and reset index
data.drop_duplicates(subset=['Close'], inplace=True)
data.reset_index(drop=True, inplace=True)

# Calculate MACD and signal lines using the 'Close' column of the concatenated data
macd = ta.trend.macd(data['Close'])
signal = ta.trend.macd_signal(data['Close'])

# Create a DataFrame of MACD and signal lines with the same index as the concatenated data
macd_df = pd.DataFrame({'MACD': macd, 'Signal': signal}, index=data.index)

# Determine whether to buy, sell or hold based on MACD and signal lines
macd_df['Signal'] = np.where(macd_df['MACD'] > macd_df['Signal'], 1, -1)

# Print the last 10 rows of the MACD and signal DataFrame and the trading signal
print(macd_df.tail(10))

if not macd_df.empty and macd_df['Signal'].iloc[-1] == 1:
    print('Buy')
elif not macd_df.empty and macd_df['Signal'].iloc[-1] == -1:
    print('Sell')
else:
    print('Hold')