What Is Monte Carlo Simulation? (Taleb's Decision-Making Application)

Monte Carlo simulation is a computational method that generates large numbers of random samples from a probability distribution to estimate the range of possible outcomes of a process. Named for the famous casino, it uses randomness to answer questions about randomness.

In Taleb's framework, Monte Carlo simulation is primarily a thinking tool — a way of forcing the mind to confront the distribution of possible outcomes rather than the single realized outcome.

The Core Application

Consider 10,000 traders, each making trades with a 52% edge on each individual bet. Simulate their portfolios over 20 years. What does the distribution of outcomes look like?

Most traders end up near where you'd expect from a 52% edge. But some — through accumulated favorable variance — end up with portfolios 50x to 100x the median. A few end up with portfolios 1,000x the median.

Now look only at the top performers after 20 years. Their track records look extraordinary. If you didn't know the simulation was producing random draws from a distribution, you might conclude they had special skill. You might write books about their methods. They might write books about their methods. But the skill was identical in all 10,000 traders — only the variance differed.

This is why Taleb calls Monte Carlo simulation "the best way to de-luck yourself." It shows the distribution, including the favorable end that looks like skill.

Why It Changes How You Evaluate Track Records

Pre-Monté Carlo thinking sees a fund manager with a 12-year excellent track record and concludes skill. Post-Monte Carlo thinking asks: how many fund managers started 12 years ago? If 5,000 started, what fraction of them would have a record like this by chance? If the answer is 3-5%, and there are 150-250 managers with similar records, the "outlier" might just be drawing from the right tail.

This isn't to say skill doesn't exist. Some performers in the distribution have genuine edge that the simulation can't identify. Monte Carlo's contribution is establishing the null hypothesis — what results look like if it's all luck — before evaluating any particular result against that baseline.

For the full framework, read Fooled by Randomness: How Luck Masquerades as Skill.