Available Now

Order now and be among the first to learn from Alternative Investing expert Bob Rice. Begin building your alternatives portfolio today! Order from Amazon.com, Barnes & Noble or 800-CEO-Reads

This episode aired on BloombergTV on Apr 24, 2013

Alpha Model

An Alpha Model is one of the core components of quantitative trading, or so-called “black box” models that account for such a high quantity of trading volume today. Alpha models are the algorithms that model and predict the behavior of securities prices.

Q. You were the referee at the “Battle of the Quants” this morning, where the debate was whether humans or machines are better investors. There were some heavyweights at the event, including Bob High, the Chief Scientist for the IBM Watson– must have been an interesting argument. And one of the terms being debated was “alpha model”… so, what’s that?

A. It’s the key element of a quantitative trading model, the algorithm that predicts how a the price of a security or set of securities will behave in the future, extrapolated from patterns that the author of the algo (often another algo) has found in the past. That’s the heart of the famous “black box” automatic trading systems.

Q. But its not the only part, right?

A. No. Here’s a simple diagram of how these things work, taken from “Inside the Black Box”. Two other really crucial elements are the risk model, which pays particular attention to how costly mistakes might be; and the transaction cost model, which says: hey, great idea, but it will cost more to execute on the trade than we’ll make. All three of those feed into the portfolio construction model, which feeds into the execution model. And there’s your system.

Q. But we all know that the output is only as good as the data going in…

A. Yes, and that’s a huge point of debate here. Yesterday was a simple example… an unfiltered piece of data– that the President was injured– caused a big selloff. So how much does that data need to be filtered and checked? And if you spend the time to check it, can your trading idea still make money? Its one of the most central questions the quants have to face.

Q. So… who won the argument? The humans or the quants?

A. This year, the competition featured a new team, the “quantimentalists”, firms that try to combine machine speed and efficiency with human judgment. And, even though it might have been a bit of a cop-out, the crowed voted for them overwhelmingly.