Hedge Fund data: Friend or foe?

Data as a friend

I am a big believer in data. From helping us choose just the right camera or dinner spot to tracking our daily activity in an effort to stay fit, data is enabling our society to make more objective, educated decisions. Too much data or data overload can be confusing and problematic, but when filtered wisely and used nobly it can fight crime and save lives. (Don’t believe me, just ask the guys at Palantir.) Data is objective by its very nature and if handled right, it can tell a beautiful story, a (sometimes surprising) reflection of the truth about any particular subject. Companies like Google and Amazon have figured out clever ways to harness the power of data and their ingenuity has propelled them to stratospheric heights. We at Novus like data so much in fact, that over seven years ago we decided to start a company based on aggregating and analyzing vast amounts of data on investment managers. One of our main value propositions to our long list of institutional clients is making sense of their own investment data as well as public data from all over the world. But more on that later.

Data as the enemy

Recently, I found out that not everyone shares our passion for data (Imagine that!?) in fact some refer to data as a “dangerous idea”. What’s more, this is happening in our very industry of alternative investments and by one of the most prominent and well regarded groups in game. Now that we have a blog, I could not resist opining on the matter but have to emphasize that ideas expressed here are my own and do not represent an official view of Novus.

One of the industry’s largest consulting firms, known for their work with pensions, serves 300 clients overseeing assets over $800 Billion. Imagine the treasure trove of manager data they must possess. But last month their CIO gave a presentation at a conference dubbed “The Summit of Dangerous Ideas” where he proposed that “Data is the danger”. And not in the security threat / identity theft sense but in the sense of investors analyzing data to select great managers. What is ironic here is that the example used to illustrate this point was Michael Lewis’s excellent book, Moneyball. For me, Moneyball signified the triumph of data over ignorance, innovation over conventional thinking. It was the toppling of an established regime of gut-feel judgment calls in favor of progressive, data-driven decisions that made the Oakland A’s incredible feat a reality. Data was the savior, not the danger. When our clients refer to Novus as the Moneyball for hedge funds, I smile and nod because that is exactly what we aspire to be. But in his presentation, the CIO argues that focusing on metrics of investment managers can be dangerous and…  I partially agree with him.

Different types of data

Source: https://www.bl.uk/

Focusing on the wrong metrics as sole drivers of investment decisions is a very bad idea. The CIO is absolutely right to point out there is no correlation between past and future alpha. The mistake here is in the definition of data. Bear with me for a math moment. The “alpha” he is referring to is calculated as the intercept of a regression line of monthly returns on some benchmark (as it is by most investors). Monthly returns, twelve data points a year, are a tiny sliver of the massive amounts of data available on investment managers. To say that looking at returns does not work hence we should not look at data is faulty logic since monthly returns are less than .0001% of all manager data and are not representative of all the other data sets. Returns are just that – they are past results, not predictors of future results. Basing investment decisions on analysis of past returns is like navigating the oceans with Ptolemy’s map in the era of GPS. Much more granular data is required to understand how you got to your results. Doing that can help uncover the principles of a manager’s process that can be counted on to repeat in the future. Hence, we agree with the CIO on the danger of using returns but not with the conclusion that data is the danger. Data is the savior, we just need to differentiate between various types of data and weight them appropriately.

Weight your data

The world is awash in data and the pace at which it’s created and thrown at us is increasing. When we walk down the street there is countless amounts of data bombarding our sensory system – we feel the sun and the wind, see an entire spectrum of colors, we subconsciously process the amount of effort to it takes to step forward and balance our bodies, and we hear various signals. Our brains process this data for us, but one of the key abilities and advantages humans have had over computers (and computers are catching up fast) is that we efficiently weight this data. So the gum wrapper flapping at your feet gets a small weight, but the car horn honking as a taxi speeds in your direction gets a ton of weight, flashing an alarm in your mind to take action in the interest of self-preservation.

The key, again, is how we weight data. It’s a good thing your brain focuses on that oncoming taxi. The same is true for making risk/reward decisions. If you are on a racetrack, do you bet on a horse because it’s pretty? Or maybe because you like the name? Perhaps it won the last race? It races better in current conditions? Because its average speed is faster than that of its competitors? Or because the odds of the bet produce an attractive expected value? Clearly, in sports betting as in investing in managers, we don’t always have ideal data. Sometimes the right data is not available, not aggregated, or analyzable. But generally, the course of scientific history has been plotted by individuals making the best with the data they had while trying to improve the data available to them.

Which data is useful?

Useful data helps us make better forward looking decisions. In the world of manager evaluation, more granular, larger data sets help inform on skill, portfolio composition and process far better than returns. This granular data is admittedly harder to cope with. As a simple comparison of the pure size of the data sets: Monthly returns produce twelve data points annually. Information contained on manager risk reports produce thousands in a given year. Trade-level data produce millions or billions of data points each month. Still, the rewards are well worth it. Our research points to heavily weighting daily trading data: positions tagged by industry and sector and other identifiers all tied to historical P&L for each strategy of each manager. That’s really granular data. And it’s powerful, and it’s predictive and it’s beautiful with the right tools on hand.

In the same presentation, the CIO further argues that investors should spend time understanding the culture of a manager and make an effort to understand the manager’s philosophy and thesis, beliefs on the market. That sounds a bit like scouting for ball players based on looks. I’d like to understand if they can make money, and consistently so. Our clients overweight data like manager holdings, exposures and their demonstrated investment skills rather than their skills of persuasion or their ability to describe their great culture. Those two skill sets – investing and salesmanship, do not appear correlated.

Just how powerful can this granular data be? Our manager clients are using the analytics daily to improve their process and future returns similar to the way Google optimizes their search results. Our allocators use it to get a sense of their risk across managers as well as understand the investment skills of their managers. These are skills like stock selection in the healthcare sector, trading acumen, sizing and the ability to short for alpha, to name a few. None of it is possible without good hard data. I propose that data is our friend not foe, and we should embrace it, learn from it and use it to improve. Handle it with care and weight it properly but ignore it at your own risk. Fearing it and calling data dangerous we run the risk of missing out on the huge opportunity that data has to offer us. After all, we can choose to ignore data but what if our competitors don’t?