Why should we care about Alternative Data?

Excerpt

#AI  #altdata  #bigdata  #machinelearning  #quantamental

These terms have received a lot of attention from the financial press, and it seems every investing conference these days has a panel discussion about how great alternative data is.  But should we care?

Many of alt data’s early adopters – and ExtractAlpha’s first clients – were quant hedge funds.  This crowd learned the hard way that if they all traded on the same information, they not only were at a competitive disadvantage; they may even face an existential risk. 

In early August 2007, massive drawdowns rippled through many of the factors then used by equity quants, forcing redemptions, fund closures, and a lot of soul-searching. We discussed this Quant Quake in some detail on its ten-year anniversary and provided some takeaways from it.

One key takeaway, which applies equally well to discretionary managers as to quants, is that sources of investing edge should be diversified, and in particular diversified away from what one’s competitors are doing. We need to diversify to provide an investing edge, but also to avoid being exposed to the same factors as our competitors, since their losses could then cascade through to our losses, as happened in 2007. And of course we need to diversify to tell a better story to our own investors as to why we are different and more innovative than other investment managers.

Diversification can be methodological (how to act on information) but can also be materials-based (what information to use).  And alternative datasets are a great way to get at new perspectives on a company’s prospects.

But embracing alternative data is tricky in a few ways, which a some recent surveys have noted, and which we observe from our discussions with clients:

• There are too many alternative data providers out there – hundreds! – and it’s hard to know which ones to pay attention to
• The datasets are often large and complex
• Many managers don’t have the infrastructure to efficiently ingest and evaluate new datasets – especially fundamental managers
• The datasets can be expensive
• Turning the datasets into actionable intelligence is not always easy

And most importantly, after all that,

• The datasets often don’t deliver any value

Visit ExtractAlpha website to read the full article: https://extractalpha.com/2019/12/16/why-should-we-care-about-alternative-data