Event study with Bitcoin and stock prices: How should I approach non-trading days?

[Evan Kim]

Hi everyone,

I am trying to run a basic event study using daily Bitcoin returns and the S&P500 index (also the USD-EUR exchange rate and 3 - month T bill rate). As many of you are aware, the stock market closes on the weekends/holidays and thus has no data. Cryptocurrency markets trade 24/7 and do not close, so I have Bitcoin prices for weekends/holidays where I do NOT have S&P500 data.

From my understanding, I can either omit non-trading days from my analysis altogether (and use a business calendar to account for this) or aggregate my data to a weekly returns level/average weekly returns level to avoid missing values. I am hesitant to omit non-trading days as I will lose the variation in Bitcoin prices during that time, and aggregating to a weekly level will cause me to lose variation as well. Is it possible for me to create dummy variables during those non-trading days? Since non-trading days have returns of 0%, would it be "valid" to fill in 0's for non-trading day returns?

Thank you for your help in advance!

[Carlo Lazzaro]

Evan:
I would go with the business calendar option (see -help datetime business calendars-), provided that it is accepted in your research field (with which I'm not familiar with).