generate_data.py

import struct
import time
import random

FILE_NAME = "market_data.bin"
NUM_RECORDS = 40_000_000 
SYMBOL = b"NVDA    "

print(f"Generating {NUM_RECORDS} records...")

with open(FILE_NAME, "wb") as f:
    start_time = time.time()
    
    # We use struct.pack to turn Python numbers into raw C-style bytes
    # 'Q' = uint64, '8s' = 8-byte string, 'd' = double
    packer = struct.Struct("<Q8sd") 
    
    for i in range(NUM_RECORDS):
        timestamp = 1700000000000000000 + i
        price = 150.0 + random.uniform(-1, 1)
        
        f.write(packer.pack(timestamp, SYMBOL, price))
        
        if i % 5_000_000 == 0:
            print(f"Progress: {i}/{NUM_RECORDS}...")

    end_time = time.time()
    print(f"Finished! Created {FILE_NAME} in {end_time - start_time:.2f}s")