I’ve been thinking about this for awhile and couldn’t figure it out myself.

Assume you have a trading strategy, which return is normally distributed. strategy return has a mean of 1 basis point and a standard deviation of 20 basis points. Now, how many times do you have to run this strategy so there’s a 99% chance you don’t lose money (your expected return is 0) ?

I understand that if returns are normally distributed , then from Z score I can calculate the probability of any return. However that doesn’t solve the above problem...

  • 1
    $\begingroup$ Are returns multiplicative or additive? Zero times is one answer ;) $\endgroup$
    – Bob Jansen
    Mar 30 at 8:09
  • $\begingroup$ Set up the problem in Excel and try it out yourself! You can use Monte Carlo simulation to answer your question. Generate a bunch time series of random draws using RAND() and NORM.DIST. Then in the final column, use COUNTIF() to see how many of your time series are above 0 at every time step. If you do 10000 time series, look for where your COUNTIF shows 9900. $\endgroup$ Mar 30 at 11:19
  • $\begingroup$ @BobJansen that’s true! $\endgroup$
    – OllyG
    Mar 30 at 11:34
  • $\begingroup$ It does make a difference whether returns are multiplicative or additive though. $\endgroup$
    – Bob Jansen
    Mar 30 at 12:01
  • $\begingroup$ By Cantelli's Inequality, if you run it 40,000 times (!) for additive returns, the mean is 40K bps, and the standard deviation is 4K bps, so the probability of a loss is smaller than $1/101$. This is covered in chapter 2 of Short Sharpe Course. $\endgroup$ Mar 30 at 16:31

Your Answer

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

Browse other questions tagged or ask your own question.