The Black Scholes formula gives the formula for European calls, for stock with no dividends, $$ c = S N(d_1) - K e^{- r (T-t)}N(d_2) $$ with $S$ the price of stock at time $t$, $T$ is the maturity date, $K$ is the strike price, $r$ is the risk free interest rate.

I have learned that change in stock price is normally distributed is a 'good' assumption. I think that this is playing some role in this Black Scholes formula, seeing that there is a normal distribution popping out in the formula.

Could someone please clarify if this is indeed the case or not? If so, what is the technical assumption equivalent of change in stock price being normally distributed? thank you

  • 2
    $\begingroup$ The B-S framework implies that $r_t:=\ln(S_t/S_0)\sim \mathcal{N}$ (so the log-change is normally distributed). This is found by solving the SDE $dS_t/S_t=\mu dt +\sigma dW_t$ using Ito on $\ln(S_t)$. This also implies that the volatility is constant through time. Turns out this is not a good assumption at all. For more details you should ask on the quant.se. $\endgroup$
    – Snoop
    Oct 23, 2022 at 17:27

1 Answer 1


As @Snoop points out, the assumption is not that changes in $S$ are normally distributed, which would allow the possibility of $S < 0$ but that changes are lognormally distributed, i.e., changes in $\ln S$ is normally distrbuted. This makes sense since the magnitude of a change $S$ depends on the currency you use but the ratio of a change in $S$ over $S$ itself does not.

The normal distribution is the most fundamental distribution on the real line. In particular, it satisfies the central limit theorem. This makes it easy to sum up normally distributed changes in $S$ with respect to a small change in time to get the distribution of $S$ over a larger time interval. You can then take a limit to get a stochastic process known as Brownian motion. So it is mathematically far easier to work with than other distributions. In particular, if you assume that $\ln S$ follows a Brownian motion with constant volatility, it leads to the Black-Scholes formula.

The problem with a normal distribution is that its density function decays far too fast and underestimates the probability of "rare" events. So there has been an effort to develop stochastic processes using so-called fat-tailed distributions. This has so far been met with limited success.

  • $\begingroup$ I'm a bit confused; "changes in $S$ is lognormally distributed" and "change in $\ln S$ are normally distributed" are the same thing? $\endgroup$
    – Johnny T.
    Oct 24, 2022 at 14:53
  • 1
    $\begingroup$ If $X$ is a normally distributed random variable, then $e^X$ is a lognormally distributed random variable. $\endgroup$
    – Deane
    Oct 26, 2022 at 1:45

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .