How do you best incorporate the weekends in the calculation of the Black Scholes historical volatility? (Of course historical volatility serves as approximation, if the market price of the options is not available).

In the book "Options, Futures and other Derivatives" by John C. Hull, it is described that over the weekend (from Friday till Monday) the variance is only 1.5 times higher than during the workweek (Monday-Friday), by using as example of a future (orange-juice futures) that is dependent on information that changes just as much during the weekend as through the week namely the weather; which shows that volatility is for a big part caused by trading and not by new information coming into the market. But still it seems to me that you would need some way to adapt to the weekend.

The volatility of the daily log-normal returns: ln(P1/P0)


The simplest approach is to use two different variables $T_1$ and $T_2$ instead of the single variable $T$ that denotes Time To Maturity in the classic Black Scholes Merton formula.

$T_1$, the time to maturity for interest rate computation purposes, is the calendar time in years between now and maturity. For example the term $-Ke^{-rT}N(d_2)$ in the formula would become $-Ke^{-rT_1}N(d_2)$. The exercise is $T_1$ years away and therefore the discounting takes this into account.

$T_2$, the effective volatility time until maturity is calculated by adding up time differently depending whether the market is open or not. For regular trading days between now and maturity a "1 " is scored, but for weekends or holiday a lower number (say 0.5 or 0.6) is taken, reflecting that volatility is expected to be lower on those days. The total is then converted to yearly basis, and $T_2$ is used in the BSM formula wherever it multiplies $\sigma^2$ or wherever $\sqrt{T}$ multiplies $\sigma$.

(I am not sure which book first presented this approach, it may be Cox & Rubinstein's Option Markets).

  • 1
    $\begingroup$ Splendid answer! Always great to learn new things. When you say "convert to yearly basis" , you divide the sum of all "scored" days by 360? Or 365? $\endgroup$ – Jan Stuller Jun 30 '20 at 19:46
  • $\begingroup$ Usually they use 252 trading days. $\endgroup$ – david laport Jul 1 '20 at 10:11
  • $\begingroup$ A day later, I realize I have misunderstood the question. You asked about historical volatility, I answered about implied. Another answer is needed in that case... But the idea is similar: non trading days are considered "fractional days" while trading days are considered "whole days". $\endgroup$ – noob2 Jul 1 '20 at 15:50

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.