Probabilistic Programming in Quantitative Finance

http://blog.quantopian.com/probabilistic-programming-quant-finance/

by botbooty • 13 years ago

17 4 13 years ago

blog.quantopian.com

For those confused by the title, Probabilistic Programing = Bayesian Statistics.

digler999• 13 years ago

I was confused by the article not delivering on the title. 1) you can't measure it so "model it" , and throw out fancy acronyms like MCMG and hand-waving.

Just in case anyone's wondering, there are other avenues (read: real science) where BA and other analysis are applied toward a definite result. An actuarial table will accurately predict what an insurance premium should be. These "quant" people are pseudo-scientists, alchemists.

Just ask this question: what results are they delivering ? 2 days ago there was, on HN, a 2010 article about how "investment bankers" don't actually do anything. What does a quant do ? What value do they deliver to anyone ?

Guess what ? I can "model" the statistical distribution of mold growing in my fridge or "monty carlo" simulate the path of rodents in my garage and I've arrived at nothing more substantial in the end. It's a disgrace what these people are doing. Bright minds wasting their talent.

yetanotherphd2• 13 years ago

I agree it is a waste of talent, but what quants are supposed to be doing is predicting the market (their ultimate goal is do construct a profitable trading strategy, but this is roughly equivalent to predicting the market).

So it is quite straightforward to test them, in principle. In practice, data snooping makes it hard to really test quant trading strategies, and as a purely practical matter trades have small effects on the market which makes it hard to test a trading strategy without actually doing it.

Mikeb85• 13 years ago

These "quant" people are pseudo-scientists, alchemists.

To a certain degree, you're right. Their job is essentially that of a fortune teller - to predict the future.

Markets are so complex that there is no single 'model' that can predict them. Furthermore, there are always winners and losers, people who lag the market and those who beat it. Coming up with a 'model' to beat the market is essentially the same as coming up with a model to win at poker. You can use statistics and math to inform yourself, but at the end of the day there's a certain amount of the unknown you have to deal with...

Probabilistic Programming in Quantitative Finance

Related Stories