Python With Binance Crypto Price Data

Hi everyone. In this programming post, I cover the topic of using Python with Binance crypto price data. This is from a personal project where I play around with Python and Binance price data to produce candlestick price charts. The main reference that I use is this Youtube video from Nicholas Renotte.

The link to my jupyterNotebook project is here.


Pixabay Image Source

Topics


  • Setup of Binance API
  • Some Functions
  • Obtaining Historical Data of Bitcoin/Tether (BTC/USDT)
  • Preprocessing Data
  • Candlestick Price Plots On Binance Crypto Data
  • Creating A Function For Producing Candlestick Price Charts Given Inputs

 

Setup of Binance API


Setting up the connection from Python (jupyterNotebook) to the Binance API is not too difficult. Go to your Binance account and obtain an API key there. This is done through the Profile picture icon then API Management.

API_mngt.PNG

 

For installation, you need the python-binance package installed. You can use pip install python-binance. The following code uses the API key and the secret key obtained from Binance and inputs them into Client.


from binance import Client, ThreadedWebsocketManager, ThreadedDepthCacheManager
import pandas as pd

client = Client(api_key, secret_key)


Pixabay Image Source

 

Some Functions


With the Client, you can obtain data from Binance. I test out a few functions.


# Obtain tickers:

tickers = client.get_all_tickers()

# First ten tickers
tickers[0:10]

[{'symbol': 'ETHBTC', 'price': '0.06240000'},
 {'symbol': 'LTCBTC', 'price': '0.00374400'},
 {'symbol': 'BNBBTC', 'price': '0.00912300'},
 {'symbol': 'NEOBTC', 'price': '0.00090000'},
 {'symbol': 'QTUMETH', 'price': '0.00277900'},
 {'symbol': 'EOSETH', 'price': '0.00175500'},
 {'symbol': 'SNTETH', 'price': '0.00003343'},
 {'symbol': 'BNTETH', 'price': '0.00142400'},
 {'symbol': 'BCCBTC', 'price': '0.07908100'},
 {'symbol': 'GASBTC', 'price': '0.00018060'}]

 

With list comprehension, the crypto pairs alone can be obtained:

# Obtain tickers pairs with list comprehension:

ticker_pairs = [x['symbol'] for x in tickers]

ticker_pairs[0:10]

['ETHBTC',
 'LTCBTC',
 'BNBBTC',
 'NEOBTC',
 'QTUMETH',
 'EOSETH',
 'SNTETH',
 'BNTETH',
 'BCCBTC',
 'GASBTC']

 

A list comprehension with an if statement can be used for obtaining crypto pairs where one of the cryptos is Tether (USDT):

# Obtain ticker pairs in USDT (Tether) as Tether is common in pairs for transactions:

usdt_pairs = [x['symbol'] for x in tickers if 'USDT' in x['symbol']]

usdt_pairs[0:10]

['BTCUSDT',
 'ETHUSDT',
 'BNBUSDT',
 'BCCUSDT',
 'NEOUSDT',
 'LTCUSDT',
 'QTUMUSDT',
 'ADAUSDT',
 'XRPUSDT',
 'EOSUSDT']

 


Pixabay Image Source

Obtaining Historical Data of Bitcoin/Tether (BTC/USDT)


As an example, I obtain histoical price data of the BTC/USDT crypto pair.

# Crypto trading pair:

base_cur = 'BTC'
quote_cur = 'USDT'

user_pair = base_cur + quote_cur


# Obtain 1 day timeframe data for BTCUSDT trading pair:
# Documentation: https://python-binance.readthedocs.io/en/latest/binance.html#binance.client.Client.get_historical_klines

klines = client.get_historical_klines(user_pair , Client.KLINE_INTERVAL_1DAY, "1 Dec, 2017", "July 22, 2021")

# Sample output
# It is a list of lists.
# Open time, Open, High, Low, Close, Volume, Close Time, Quote Asset Volume, Num Trades, Taker buy base asset vol,
# Taker buy quote asset vol, Ignore.

klines[0:3]

[[1512086400000,
  '9837.00000000',
  '10898.00000000',
  '9380.00000000',
  '10782.99000000',
  '6134.92363300',
  1512172799999,
  '62260697.58291551',
  32375,
  '3269.86858800',
  '33193672.16592109',
  '35258.70264511'],
 [1512172800000,
  '10775.04000000',
  '11190.00000000',
  '10620.00000000',
  '10890.01000000',
  '4765.43975700',
  1512259199999,
  '52046689.84006951',
  29694,
  '2390.19427500',
  '26111051.97904352',
  '35491.76836387'],
 [1512259200000,
  '10902.69000000',
  '11825.00000000',
  '10500.00000000',
  '11165.41000000',
  '5346.63652400',
  1512345599999,
  '60350708.29332783',
  39335,
  '2574.24618700',
  '29090180.16290429',
  '36967.53447487']]

 

The function for obtaining historical price data is client.get_historical_klines(<crypto_pair> , <timeframe>, <start_date>, <end_date>).

With the timeframe, it is not as simple as 1day, 1min or 1 month. The KLINE_INTERVAL part is needed. Here is the documentation link
which has more interval examples.


Pixabay Image Source

 

Preprocessing Data


The historical price data for BTCUSDT is then put into a pandas dataframe. Columns are renamed too.

# Convert klines historical data into Pandas dataframe:

btc_usdt_df = pd.DataFrame(klines)

# Name colums:

col_names = ['Open Time', 'Open', 'High', 'Low', 'Close', 'Volume', 'Close Time', 'Quote Asset Volume',                  'Num Trades', 'Taker buy base asset vol', 'Taker buy quote asset vol', 'Ignore']

btc_usdt_df.columns = col_names

 

Times are converted from UNIX format into datetime and the rest of the columns are converted into numeric columns.

# UNIX to datetime (divide by 1000):

btc_usdt_df['Open Time'] = pd.to_datetime(btc_usdt_df['Open Time']/1000, unit = 's')
btc_usdt_df['Close Time'] = pd.to_datetime(btc_usdt_df['Close Time']/1000, unit = 's')

numeric_cols = ['Open', 'High', 'Low', 'Close', 'Volume', 'Quote Asset Volume', 'Num Trades', 
             'Taker buy base asset vol', 'Taker buy quote asset vol']

btc_usdt_df[numeric_cols] = btc_usdt_df[numeric_cols].apply(pd.to_numeric, axis = 1)

# Check data types:

btc_usdt_df.dtypes

Open Time                    datetime64[ns]
Open                                float64
High                                float64
Low                                 float64
Close                               float64
Volume                              float64
Close Time                   datetime64[ns]
Quote Asset Volume                  float64
Num Trades                          float64
Taker buy base asset vol            float64
Taker buy quote asset vol           float64
Ignore                               object
dtype: object

 


Pixabay Image Source

Candlestick Price Plots On Binance Crypto Data


The key piece for creating candlestick price charts for Open, High, Low, Close, Volume data is mplfinance. Creating the candlestick plot is not too difficult. You just need mpf.plot().

import mplfinance as mpf

 

# Candlestick plot of BTCUSDT

mpf.plot(btc_usdt_df.set_index('Close Time').tail(200),
         type = 'candle', style = 'charles',
         volume = True,
         title = user_pair + " Price ",
         ylabel = '\n ' + base_cur + ' in ' + quote_cur)

 

candlestick01_BTCUSDT.PNG

 

Creating A Function For Producing Candlestick Price Charts Given Inputs


I wanted to go a little bit beyond the Youtube video from Nicholas Renotte. In this, I have made a function for producing candlestick price charts given user input.
This function combines the steps from earlier. Historical price data is obtained depending on user input. The data is then converted into a Pandas dataframe along with preprocessing. Once the data is ready, a candlestick price plot is produced.

# Create a function that does the above steps:

def candlestick_prices_plot(start_date, end_date, base_cur, quote_cur, timeframe):
    trade_pair = base_cur + quote_cur
    
    if timeframe == '1day':
        data = client.get_historical_klines(trade_pair , Client.KLINE_INTERVAL_1DAY, start_date, end_date)
    elif timeframe == '1week':
        data = client.get_historical_klines(trade_pair , Client.KLINE_INTERVAL_1WEEK, start_date, end_date)
    elif timeframe == '1month':
        data = client.get_historical_klines(trade_pair , Client.KLINE_INTERVAL_1MONTH, start_date, end_date)
    else:
        print("Enter one of 1day, 1week or 1 month")
    
    # Into pandas dataframe:
    
    df = pd.DataFrame(data)
    col_names = ['Open Time', 'Open', 'High', 'Low', 'Close', 'Volume', 'Close Time', 'Quote Asset Volume', 'Num Trades', 
             'Taker buy base asset vol', 'Taker buy quote asset vol', 'Ignore']
    df.columns = col_names
    
    # Preprocess:
    # UNIX to datetime (divide by 1000):
    df['Open Time'] = pd.to_datetime(df['Open Time']/1000, unit = 's')
    df['Close Time'] = pd.to_datetime(df['Close Time']/1000, unit = 's')
    
    numeric_cols = ['Open', 'High', 'Low', 'Close', 'Volume', 'Quote Asset Volume', 'Num Trades', 
             'Taker buy base asset vol', 'Taker buy quote asset vol']
    
    df[numeric_cols] = df[numeric_cols].apply(pd.to_numeric, axis = 1)
    
    # Candlestick plot of BTCUSDT

    mpf.plot(df.set_index('Close Time').tail(200),
         type = 'candle', style = 'charles',
         volume = True,
         title = trade_pair + " Price from \n" + start_date + ' to ' + end_date,
         ylabel = '\n ' + base_cur + ' in ' + quote_cur


)

 

Here is one function call with the output (as a screenshot).

# Function Call 1:

candlestick_prices_plot(start_date = "1 Jan, 2018", end_date = "July 22, 2021", 
                        base_cur = 'ETH', quote_cur = 'USDT', timeframe = '1day')

 

candlestick02_ETHUSDT.PNG

Thank you for reading.



0
0
0.000
4 comments
avatar

The blog is great sir👍

0
0
0.000
avatar

Thank you very much. As I am taking a break from posting math topics, I am getting back into the programming/ data analytics side of things.

I have more coming as I found some nice resources for obtaining crypto data.

0
0
0.000
avatar

I also want to venture myself into data science and have started self preparing through various tutorials, so I do understand the code. Your function candlestick_prices_plot is pretty impressive and I learned something.
Will wait for your blogs sir............

0
0
0.000
avatar

I use Binance but I never tried with their API function because I think it's too technical for me and if I do something wrong then I might lose my funds.
After going through your posts I can say my decision is right because I tried to understand a lot but was still not able to get it more. I think that it's not my cup of tea. good for people who are tech experts.

0
0
0.000