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.
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.
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)
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']
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.
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
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)
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')
Thank you for reading.