Hacker News Reader

Comment by nlitened | Hacker News Reader

One huge example is an assumption that portfolio returns are normally distributed. That’s not a minor nitpick, this invalidates every formula that goes after.

View on HN

OP does address this near the end of the post.

Well, you did say you didn't finish it. So would it surprise you to learn that topic is addressed directly?

Thats a fair assumption. But I did skim to the part that attempted to address non-normality—it doesn’t. Under non-normality, not a single formula in this post holds (“standard deviation” does not exist).

Hi, I wrote the post. I discussed the problem of non-normality in the post. Moreover, if non-normality is the problem, then pretty much all of modern finance is invalidated. Instead of saying its all bullshit, a better approach is to realize the assumptions of the model and use some discretion in trading based on its output.

> Moreover, if non-normality is the problem, then pretty much all of modern finance is invalidated.

I’d say, finance from the 70s is invalidated. And it has been invalidated countless times with all the major mutual funds going bust.

I’d love to encourage you to read more on this topic. Taleb’s “The Black Swan” is a good start, his other books are also good.

Black Scholes, VaR, factor models, CAPM, modern portfolio theory, etc are all based on the normal distribution and are all still used in industry today. Every quant fund in the world is using models that assume a log normal distribution of returns. Moreover, Taleb certainly used models based on log normal distributions at his hedge funds.

Reply on news.ycombinator.com