Is Technology a Job Killer in Asset Management?

Is Technology a Job Killer in Asset Management?

Halved The Coverage but Twice the Alpha

Another positive result of using technology is the narrowing of the focus to generate unique perspectives, which resulted in better returns. Three years ago, an average analyst or portfolio manager may have a coverage of a hundred names, noted O'Malley.

«But in contrast that by today, I get a portfolio manager with a coverage of fifty names by each member of their team and generates twice the alpha, than someone who does who has a hundred names under coverage in their universe,» he said. «What I don’t see as much are the AI fears, the machine learning, all the heuristics of assisted-tech taking over the industry.»

Big Data Overrated

«There are two terms that drive me crazy: Big data – that was the most overused marketing term in 2016 and Quantamental is the other in 2017,» said O'Malley. For example, if someone can get revenue data  purely for the iPhone (in isolation), he is pretty sure it can be monetised. However, many data sets are only «incrementally helpful» while others provide «very limited advantage for very short periods of time» without consistent predictive power, he noted.

DeAddio concurred, saying that terms like «Quantamental» and «Big Data» drive him nuts for multiple reasons. «Data is just data. It’s the predictive power and what you do with the data that really matters, so it should be called big prediction or something that is a little more thoughtful,» DeAddio explained.

Think Small Data 

In other cases, even thin alphas or small signals provide tremendous value, as long as one gets good signal quality around it from other sources, like those that a fundamental stock picker would typically find, said O'Malley.

«So data is integrated into the (investment) process but it’s a much lower value than people perceive it to be. The stuff that people talk most about tends to be the least value. I don’t know if u met someone who is doing big data on healthcare or big data on insurance - that’s the stuff that’s interesting,» he said.

Yield Still Low

While quantitative funds may use tools and machine learning techniques to assess the value of data before purchasing, asset managers have to accept that not every data set is useful. Sometimes, they could trial a thousand data sets but maybe land a yield of 25 percent, noted DeAddio.

«From a statistics perspective, not every data set has to work, sometimes you buy them and they don’t work out, so you don’t know the value of them until you put your researchers on them for 6 to 9 months and then see if you can actually find value in it,» said DeAddio. Hence, the use of data is a multi-step process, starting from the initial assessment, the negotiation, the buying of data, and then finally the research.

Cost of Doing Business

«And if you think you’re going to get every single one right, you’re fooling yourself. The world is increasing so quickly the data types you have, you’ll make some mistakes. It’s just the cost of doing business in the data world,» DeAddio noted.