Friday, March 11, 2011

A Twitter-Based Hedge Fund

I saw one of the most unusual segments on CNBC a few days ago involving a hedge fund announcing it will use Twitter-based inputs for trading.

If I recall correctly, a psychology professor and his grad student played with data and found an alleged leading indicator of Tweets, classified by sentiment, to the Dow's performance 3-4 days later.

Then a British guy with money from some other, distinctly non-financial venture, the exact nature of which I've forgotten, but think it had something to do with hair care, together with his colleague or brother, is investing some millions of pounds in a hedge fund- and promptly arranged to use the Twitter-based approach. The investor was presented in a brief clip extolling the importance of jumping on such a hot, 'leading edge' concept as Tweet-based trading.

This is almost too funny for words, isn't it? If it were three weeks from now, I'd think it was an April Fool's Day joke.

There are so many things about this story that don't make sense that it makes you wonder how it even made it onto CNBC a few afternoons ago.

First, it's clearly a high-frequency trading strategy. That means it's expensive. We learned nothing about how much and how rigorous was the backtesting the creators did on the strategy.

Of course, being psychologists, selling or licensing their technique to novice hedge fund owners, one wonders if any of them- concept originators or hedge fund owners- actually have any idea how professionals actually test these things?

Then there's the rather non-trivial question of just how it works. Believe it or not, most sophisticated investors don't want to invest with novice fund managers who operate a total black box investing strategy. Think, well, Bernie Madoff.

Especially one which relies, we are told, on the daily attitudes of the tweets of samples of several million users. Which are then, we're told, classified into one or more of six sentiment buckets, the disposition of which apparently creates a signal to buy or sell the Dow in anticipation of its level 3-4 days hence.

The actual example given in the interview on CNBC was a Tweet by a person stating that they were going to their gym to work off some excess weight. The psychologist sagely intoned that this was a negative market indicator.

Then there's the ease with which this could probably be cloned, with the market inefficiencies beaten out of it within a year- or less.

Or perhaps a competitor will clone it, then go one step further and poison the general Twitter feeds, while retaining their own pristine stream, the better to confuse others trying to trade on such sentiment.

Having been introduced to adapting what a novice equity strategist thinks is an investment-ready approach to actual operational status by experienced fund managers, I speak from experience when I say that I am sceptical of the Twitter-based fund concept. There just don't seem to be any experienced adults in sight to vet the idea.

But I guess time will tell whether we'll hear more about this one.

No comments: