Days randomized


Can you give more details about what you mean by “days have been randomized”:

“the ‘date’: related to a specific day, it is shared between several stocks (for practical reasons, days have been randomized”

  • Does it mean that (day 1) can be any day of the week ?

  • Does it mean that (day 2) is not the day after (day 1) ? (Tuesday after Monday)

  • Does it mean that if (day 1) is a Monday then (day 1 + 7) is not a Monday ?


We meant that a random date number was assigned to each date.

Thus, for example, we could have something like:

Day 1 = 2017-01-24
Day 2 = 2017-01-15
Day 3 = 2018-01-23


Hi Neantis,
I think that if days were ordered, you could forecast the volatility between 14pm and 16pm using the information of the next day (opening volatility value of the next day could help you to predict the closing value of previous day). This is not good because you would use future information to predict the past. This is impossible in reality. So I think it may be one of the reason why they randomised the dates.

So at the end:

  • day 1 can be any day of the week (more particularly any day of the whole dataset range)
  • day 2 is not likely to be the day after day 1 and the time between day 1 and day 2 is unknown to us since it has been randomised.
  • if day 1 is a monday, then day 1+7 can be any another day before or after day 1.

I hope my explanations are clear to you.


Thanks for your reply.