main.py

import numpy as np
import pandas as pd
import matplotlib.pyplot as plt
import requests
import datetime
import schedule
import time
from sklearn.metrics import r2_score

# Define Global Variables
EXCHANGE_URL = "https://www.deribit.com/api/v2/public"
INSTRUMENTS_END_POINT = "/get_instruments"
ORDER_BOOK_END_POINT = "/get_order_book_by_instrument_id"
CURRENCY = "BTC"
KIND = "option"
CONTRACT_MONTH = "2022-06-24"
RATE_LIMIT = 5
STRIKE_PRICE_EXCLUDE_ABOVE = 60000

# Define Helper Function(s)
def unix_to_date_str(unix_timestamp):
    return datetime.datetime.utcfromtimestamp(unix_timestamp/1000).strftime('%Y-%m-%d')

# Decorator for request error encounter
def retry(func):
    def repeat_every_five_sec(*args, **kwargs):
        while True:
            try:
                return func(*args, **kwargs)
            except:
                print("Error. Try again in 5 secs.")
                time.sleep(5)
                continue
            break
    return repeat_every_five_sec

@retry
def send_request(baseURL, endpoint, payloads, timeout):
    return requests.get(f"{baseURL}{endpoint}",params=payloads,timeout=timeout)

# Main Function to be schedule
def run():
    # Simply Get all options contract First
    payloads = {"currency": CURRENCY, "kind": KIND }
    res = send_request(EXCHANGE_URL, INSTRUMENTS_END_POINT, payloads, 10)
    option_list = res.json()['result']
    time.sleep(1)

    #Step 1: Filter by Contract Month
    relevant_options = [option for option in option_list if unix_to_date_str(option['expiration_timestamp']) == CONTRACT_MONTH]

    #Step 2: For each option contract, take their intrument unique id and get the mark IV from order book data and store it
    results = []
    count = 0
    for option in relevant_options:
        if count % RATE_LIMIT == 0:
            print(f"Querying Option Chain Data... {(count/len(relevant_options)*100):.2f}% Complete")
            time.sleep(1)
        payloads = {"instrument_id": option['instrument_id']}
        res = send_request(EXCHANGE_URL, ORDER_BOOK_END_POINT, payloads, 10)

        strike_price = option['strike']
        option_type = option['option_type']
        mark_iv = res.json()['result']['mark_iv']
        
        results.append([option_type, strike_price, mark_iv])
        count += 1

    # Step 3: Take those calls and puts and sorted by strike price and fit a quadratic curve on the mark IV array
    calls = [i for i in results if i[0] == 'call']
    calls = sorted(calls,key=lambda l:l[1], reverse=False)
    calls_x = [call[1] for call in calls if call[1] <= STRIKE_PRICE_EXCLUDE_ABOVE] #<=60000 just for better r-square
    calls_y = [call[2] for call in calls if call[1] <= STRIKE_PRICE_EXCLUDE_ABOVE]
    call_model = np.poly1d(np.polyfit(calls_x,calls_y, 2))

    puts = [i for i in results if i[0] == 'put']
    puts = sorted(puts,key=lambda l:l[1], reverse=False)
    puts_x = [put[1] for put in puts if put[1] <= STRIKE_PRICE_EXCLUDE_ABOVE]
    puts_y = [put[2] for put in puts if put[1] <= STRIKE_PRICE_EXCLUDE_ABOVE]
    put_model = np.poly1d(np.polyfit(puts_x,puts_y, 2))


    # Deliver the output in console 
    print("\n")
    print("\n")
    print("--------------------CALL volatility curve Equation-----------------------")
    print(f"Equation: {call_model}")
    print(f"R-Square: {r2_score(calls_y, call_model(calls_x))}")
    print("\n")
    print("--------------------PUT volatility curve Equation-----------------------")
    print(f"Equation: {put_model}")
    print(f"R-Square: {r2_score(puts_y, put_model(puts_x))}")
    print("\n")
    print("Check out the visualization in Walk Through.ipynb!!!")
    print("Now wait for another 5 mins zzz...")


if __name__ == '__main__':
    # Print the equation every 5 min
    schedule.every(5).minutes.do(run)
    print("Task: Print the quadratic equation generated by mark IV every 5 min")
    run()
    while True:
        schedule.run_pending()
        time.sleep(1)