Monday, August 13, 2007

A reader's comment on quant funds' losses

A reader of mine (who wish to remain anonymous) pointed out that most of the losses seem to come from low-frequency trading models, while high frequency models continue to perform superbly. This also confirms my own experience. My enthusiasm for high frequency trading was expressed previously here and here.

8 comments:

OneEyedMan said...

Did you see the article on Goldman's Global Equity Opportunities fund?

The NY times is reporting that it was "worth a little less than $5 billion slightly more than a week ago. After the market’s wild roller coaster ride last week, which affected a wide swath of quantitative or computer-generated trading models, its assets fell to $3.6 billion"

They describe the fund as "pure quantitative strategies, meaning computers are programmed with millions of mini-strategies to buy and sell when certain things happen in the marketplace."

That's a prominent example of a high frequency fund that is in trouble.

OneEyedMan said...

http://www.nytimes.com/2007/08/13/business/13cnd-goldman.html?hp

Ernie Chan said...

Dear Benjamin,
Just because the market value of a fund can drop rapidly does not necessarily mean that it is a high-frequency fund. The definition of high-frequency is that the holding period of a stock cannot be more than a few hours. For a $5 billion fund, it is unlikely that they can trade this frequently and not cause excessive market impact.
Ernie

Anonymous said...

Could you please list the names of some high-frequency hedge funds? Where can I get some reliable information on their returns?

Thanks,

Quantonymous

Ernie Chan said...

Hi Quantonymous,
Almost every major multi-strategy quantitative hedge fund has a high-frequency portfolio. I know of a specific example, but unfortunately its name was provided to me in confidence. It is not likely that they will break out the returns for this specific strategy though, unless you are an investor already.
Ernie

OneEyedMan said...

This sounds like a hypothesis that is difficult to refute or establish.

Anonymous said...

Dear Ernie

I agree with your observation that high-freq models are not only weathering the storm, but are actually thriving right now, probably because they are able to 'harvest' the volatility.

The only problem is one of scalability: low-freq models are generally much more scalable than high-freq models for obvious reasons. Thus, for a good sized fund managing > $ 3billion, there is only so much you can allocate to a high-freq model. How do you get around this prob?

Ernie Chan said...

Dear anonymous,

You are right that high frequency strategies do not have a large capacity. If you need to manage a large amount of capital, one way is to have a diversity of strategies, high frequency or otherwise. Another way is to just accept the volatility in returns and adopt a low-frequency strategy, all the while taking a conservative approach to leveraging.

Ernie