Get Crypto Data from Bybit

from pybit.unified_trading import HTTP
import pandas as pd
import numpy as np
import matplotlib.pyplot as plt
import datetime as dt
import time 
import json

with open('authcreds.json') as j:
        creds = json.load(j)
        
        

key = creds['KEY_NAME']['key']
secret = creds['KEY_NAME']['secret']
  
session = HTTP(api_key=key, api_secret=secret, testnet=False)

result = session.get_tickers(
                category="linear").get('result')['list']

tickers = [asset['symbol'] for asset in result if asset['symbol'].endswith('USDT')]
print(tickers)

['10000NFTUSDT', '1000BONKUSDT', '1000BTTUSDT', '1000FLOKIUSDT', '1000LUNCUSDT', '1000XECUSDT', '1INCHUSDT', 'AAVEUSDT', 'ACHUSDT', 'ADAUSDT', 'AGIXUSDT', 'AGLDUSDT', 'AKROUSDT', 'ALGOUSDT', 'ALICEUSDT', 'ALPHAUSDT', 'ANKRUSDT', 'ANTUSDT', 'APEUSDT', 'API3USDT', 'APTUSDT', 'ARBUSDT', 'ARPAUSDT', 'ARUSDT', 'ASTRUSDT', 'ATOMUSDT', 'AUDIOUSDT', 'AVAXUSDT', 'AXSUSDT', 'BAKEUSDT', 'BALUSDT', 'BANDUSDT', 'BATUSDT', 'BCHUSDT', 'BELUSDT', 'BICOUSDT', 'BITUSDT', 'BLURUSDT', 'BLZUSDT', 'BNBUSDT', 'BNXUSDT', 'BOBAUSDT', 'BSVUSDT', 'BSWUSDT', 'BTCUSDT', 'BUSDUSDT', 'C98USDT', 'CEEKUSDT', 'CELOUSDT', 'CELRUSDT', 'CFXUSDT', 'CHRUSDT', 'CHZUSDT', 'CKBUSDT', 'COCOSUSDT', 'COMPUSDT', 'COREUSDT', 'COTIUSDT', 'CREAMUSDT', 'CROUSDT', 'CRVUSDT', 'CTCUSDT', 'CTKUSDT', 'CTSIUSDT', 'CVCUSDT', 'CVXUSDT', 'DARUSDT', 'DASHUSDT', 'DENTUSDT', 'DGBUSDT', 'DODOUSDT', 'DOGEUSDT', 'DOTUSDT', 'DUSKUSDT', 'DYDXUSDT', 'EGLDUSDT', 'ENJUSDT', 'ENSUSDT', 'EOSUSDT', 'ETCUSDT', 'ETHUSDT', 'ETHWUSDT', 'FETUSDT', 'FILUSDT', 'FITFIUSDT', 'FLMUSDT', 'FLOWUSDT', 'FLRUSDT', 'FTMUSDT', 'FXSUSDT', 'GALAUSDT', 'GALUSDT', 'GFTUSDT', 'GLMRUSDT', 'GMTUSDT', 'GMXUSDT', 'GPTUSDT', 'GRTUSDT', 'GTCUSDT', 'HBARUSDT', 'HFTUSDT', 'HIGHUSDT', 'HNTUSDT', 'HOOKUSDT', 'HOTUSDT', 'ICPUSDT', 'ICXUSDT', 'IDUSDT', 'ILVUSDT', 'IMXUSDT', 'INJUSDT', 'IOSTUSDT', 'IOTAUSDT', 'IOTXUSDT', 'JASMYUSDT', 'JOEUSDT', 'JSTUSDT', 'KAVAUSDT', 'KDAUSDT', 'KLAYUSDT', 'KNCUSDT', 'KSMUSDT', 'LDOUSDT', 'LINAUSDT', 'LINKUSDT', 'LITUSDT', 'LOOKSUSDT', 'LPTUSDT', 'LQTYUSDT', 'LRCUSDT', 'LTCUSDT', 'LUNA2USDT', 'MAGICUSDT', 'MANAUSDT', 'MASKUSDT', 'MATICUSDT', 'MINAUSDT', 'MKRUSDT', 'MTLUSDT', 'NEARUSDT', 'NEOUSDT', 'NKNUSDT', 'OCEANUSDT', 'OGNUSDT', 'OMGUSDT', 'ONEUSDT', 'ONTUSDT', 'OPUSDT', 'PAXGUSDT', 'PEOPLEUSDT', 'QTUMUSDT', 'RDNTUSDT', 'REEFUSDT', 'RENUSDT', 'REQUSDT', 'RLCUSDT', 'RNDRUSDT', 'ROSEUSDT', 'RPLUSDT', 'RSRUSDT', 'RSS3USDT', 'RUNEUSDT', 'RVNUSDT', 'SANDUSDT', 'SCRTUSDT', 'SCUSDT', 'SFPUSDT', 'SHIB1000USDT', 'SKLUSDT', 'SLPUSDT', 'SNXUSDT', 'SOLUSDT', 'SPELLUSDT', 'SSVUSDT', 'STGUSDT', 'STMXUSDT', 'STORJUSDT', 'STXUSDT', 'SUNUSDT', 'SUSHIUSDT', 'SWEATUSDT', 'SXPUSDT', 'THETAUSDT', 'TLMUSDT', 'TOMOUSDT', 'TRBUSDT', 'TRUUSDT', 'TRXUSDT', 'TUSDT', 'TWTUSDT', 'UNFIUSDT', 'UNIUSDT', 'USDCUSDT', 'VETUSDT', 'WAVESUSDT', 'WOOUSDT', 'XCNUSDT', 'XEMUSDT', 'XLMUSDT', 'XMRUSDT', 'XNOUSDT', 'XRPUSDT', 'XTZUSDT', 'YFIUSDT', 'YGGUSDT', 'ZECUSDT', 'ZENUSDT', 'ZILUSDT', 'ZRXUSDT']

Parameter	Required	Type	Comments
category	true	string	Product type. spot, linear, inverse
symbol	true	string	Symbol name
interval	true	int/string	Kline interval. 1, 3, 5, 15, 30, 60, 120, 240, 360, 720, D, M, W
start	false	integer	The start timestamp (ms)
end	false	integer	The end timestamp (ms)

 

The table below shows the response parameters , the function format_data above extracts the information in the list from the API and formats them in to a pandas dataframe. 

 

Request Response

 

Parameter	Type	Comments
category	string	Product type
symbol	string	Symbol name
list	array	An string array of individual candle Sort in reverse by startTime list[0]: startTime - Start time of the candle (ms) list[1]: openPrice - Open price list[2]: highPrice - Highest price list[3]: lowPrice - Lowest price list[4]: closePrice - Close price. Is the last traded price when the candle is not closed list[5]: volume - Trade volume. Unit of contract: pieces of contract. Unit of spot: quantity of coins list[6]: turnover - Turnover. Unit of figure: quantity of quota coin

 

 

The python script below calls the get_kline method from Pybit. Note that we do not pass the start/end timestamp argument, which we will come back to later in the article. 

 

 

response = session.get_kline(category='linear', 
                             symbol='ETHUSDT', 
                             interval='D').get('result')


def format_data(response):
    '''
    

    Parameters
    ----------
    respone : dict
        response from calling get_klines() method from pybit.

    Returns
    -------
    dataframe of ohlc data with date as index

    '''
    data = response.get('list', None)
    
    if not data:
        return 
    
    data = pd.DataFrame(data,
                        columns =[
                            'timestamp',
                            'open',
                            'high',
                            'low',
                            'close',
                            'volume',
                            'turnover'
                            ],
                        )
    
    f = lambda x: dt.datetime.utcfromtimestamp(int(x)/1000)
    data.index = data.timestamp.apply(f)
    return data[::-1].apply(pd.to_numeric)

df = format_data(response)

print(df)
'''

               timestamp         open  ...        volume      turnover
timestamp                              ...                            
2022-10-12  1.665533e+12  1279.300049  ...  7.275909e+05  9.415818e+08
2022-10-13  1.665619e+12  1293.849976  ...  2.185120e+06  2.725300e+09
2022-10-14  1.665706e+12  1286.500000  ...  1.124476e+06  1.480705e+09
2022-10-15  1.665792e+12  1295.650024  ...  5.074429e+05  6.510826e+08
2022-10-16  1.665878e+12  1274.250000  ...  5.975026e+05  7.729099e+08
                 ...          ...  ...           ...           ...
2023-04-25  1.682381e+12  1841.000000  ...  8.319169e+05  1.524671e+09
2023-04-26  1.682467e+12  1865.099976  ...  1.701774e+06  3.202957e+09
2023-04-27  1.682554e+12  1865.209961  ...  1.035564e+06  1.968957e+09
2023-04-28  1.682640e+12  1907.699951  ...  5.036653e+05  9.560499e+08
2023-04-29  1.682726e+12  1890.010010  ...  2.651855e+05  5.038554e+08

[200 rows x 7 columns]      
'''

 

 

The problem with this approach is that we are constrained to only 200 rows of data. Therefore if we want data exceeding (in this case) 200 days we have to make multiple function calls. Let's say we wanted to get hourly data for DOGE since the start of 2023. We will have to make multiple API calls. 

 

def get_last_timestamp(df):
    return int(df.timestamp[-1:].values[0])

start = int(dt.datetime(2023, 1, 1).timestamp()* 1000)

interval = 60
symbol = 'DOGEUSDT'
df = pd.DataFrame()

while True:
    response = session.get_kline(category='linear', 
                                 symbol=symbol, 
                                 start=start,
                                 interval=interval).get('result')
    
    latest = format_data(response)
    
    if not isinstance(latest, pd.DataFrame):
        break
    
    start = get_last_timestamp(latest)
    
    time.sleep(0.1)
    
    df = pd.concat([df, latest])
    print(f'Collecting data starting {dt.datetime.fromtimestamp(start/1000)}')
    if len(latest) == 1: break


df.drop_duplicates(subset=['timestamp'], keep='last', inplace=True)

 

Cool now we have hourly data!

 

 

 

 

 

Get Orderbook Data

 

A cryptocurrency order book is similar to the order book for any other asset, but it's specific to cryptocurrencies like Bitcoin or Ethereum.

In a cryptocurrency order book, you'll see the current bids and asks for a particular cryptocurrency, along with the amount of cryptocurrency being offered and the price being asked for it.

On the buy side, people place bids to purchase a certain amount of cryptocurrency at a specific price. On the sell side, people place asks to sell a certain amount of cryptocurrency at a specific price.

The order book will typically show the highest bids and lowest asks at the top, with lower bids and higher asks listed further down. This provides traders with an idea of the current market depth and the supply and demand of the cryptocurrency.

 

 

 

Traders use the information in the order book to make informed decisions about when to buy or sell cryptocurrency. For example, if there are a lot of buy orders at a certain price level, it may indicate that demand for the cryptocurrency is increasing and that the price is likely to rise. Conversely, if there are a lot of sell orders at a certain price level, it may indicate that there is a lot of supply and that the price is likely to fall.

 

 

response = session.get_orderbook(
    category="linear",
    symbol="DOGEUSDT",
    limit=50).get('result')


def format_order_book(response):
    '''
    

    Parameters
    ----------
    response : dict
    
    Returns
    -------
    two list of lists containing bid/ask price with associated volume

    '''
    bids = response.get('b')
    asks = response.get('a')
    
    return bids, asks
    



bids, asks = format_order_book(response)          

for bid in bids:
    print(f'Bid price - {bid[0]} , quantity = {bid[1]}')


'''
Bid price - 0.081 , quantity = 1177521
Bid price - 0.08099 , quantity = 936765
Bid price - 0.08098 , quantity = 541063
Bid price - 0.08097 , quantity = 483457
Bid price - 0.08096 , quantity = 1211693
'''

 

 

Get Open Interest Data

 

Open interest is the total number of outstanding futures contracts that have not been settled or closed out by an offsetting trade. In other words, it represents the number of contracts that are currently open or active in the market.

 

When someone buys or sells a futures contract, the open interest increases by one. When someone closes out a position by selling or buying a contract to offset their original trade, the open interest decreases by one.

 

Open interest is an important indicator of market liquidity and sentiment, as it represents the number of market participants who are actively trading a particular futures contract. High open interest suggests that there is a lot of trading activity and interest in the contract, while low open interest may indicate that the market is less active or less popular. 

 

The input arguments are similar to the get_klines method we used previously. However, the interval seems to be different for some reason. Also note the loop through the dictionaries, this is due to the fact that Pandas doesn't seem to like the 7 trailing zeros , and sometimes returns NaN and sometimes doesn't! So we use a suboptimal solution as shown below. 

 

Parameter	Required	Type	Comments
category	true	string	Product type. linear,inverse
symbol	true	string	Symbol name
intervalTime	true	string	Interval. 5min,15min,30min,1h,4h,1d
startTime	false	integer	The start timestamp (ms)
endTime	false	integer	The end timestamp (ms)
limit	false	integer	Limit for data size

 

 

def format_oi(response):
    '''
    

    Parameters
    ----------
    respone : dict
        response from calling get_open_interest

    Returns
    -------
    dataframe of open interest and timestamp

    '''
    data = response.get('list', None)
    
    if not data:
        return 
    
    ois =[] 
    tss = []
    for row in data:
        oi = float(row.get('openInterest'))
        ts = int(row.get('timestamp'))
        ois.append(oi)
        tss.append(ts)
        
    data = pd.DataFrame()
    data['open_interest'] = ois
    data['timestamp'] = tss
    
    f = lambda x: dt.datetime.utcfromtimestamp(int(x)/1000)
    data.index = data.timestamp.apply(f)
    return data[::-1]



response = session.get_open_interest(
    category="linear",
    symbol="DOGEUSDT",
    intervalTime="1h",
    limit=200
).get('result')



oi = format_oi(response)

'''
                     openInterest     timestamp
timestamp                                      
2023-04-28 01:00:00   934277376.0  1.682644e+12
2023-04-28 02:00:00   935817856.0  1.682647e+12
2023-04-28 03:00:00   935150784.0  1.682651e+12
2023-04-28 04:00:00   934870976.0  1.682654e+12
2023-04-28 05:00:00   933942976.0  1.682658e+12

'''

 

If you want to get open interest data for longer than 200 periods, you can apply the same logic as we used for the get_klines method previously.