Rick Di Mascio, founder of Inalytics explains why it is important to take a deep data dive into the investment research process.
Going back to 1998, what were the drivers behind starting Inalytics? You would now call it a start-up.
When I was running money and teams of portfolio managers, we did the usual performance and attribution analysis. However, they only recorded what happened but not why. There weren’t any objective measures of skill which meant I had no idea what a manager’s strengths and weaknesses were. This was not a great way to run anything, let alone one of the largest plans in the UK, the British Coal Scheme. I thought there must be other people in similar positions who wanted to have independent empirical measures of skill to test what is real versus what is being said. And so here we are over 20 years later,
As for being start-up, in many ways it is not surprising that I am entrepreneurial. I come from a long line of Italian restaurant owners and ice cream makers and knew that one day I would get back to my roots and work for myself.
One of your recent studies underscored the importance of research skills. What about portfolio construction?
There is a lot of talk about the benefits of portfolio construction, but the reality is research is the source of excess returns. We analysed over 750 equity mandates managed on behalf of pension schemes and institutional investors globally and found that 88% of portfolios generated positive alpha from research. Managers with demonstrable research skills generated 383 basis points alpha per annum, which was far greater than any other sources of alpha. By contrast, the analysis found that other key elements of investment processes, like portfolio construction and trading, generated little or no positive alpha.
What it comes down is the success of the investment process. It is not just about stock picking. This may be the main area in which managers can demonstrate skill, but allocators or asset owners also need to spend more time trying to understand the managers’ research process while conducting due diligence. If the research process, for whatever reason, stops adding value, then it is quite clear that the sizing decisions and portfolio construction are not going to make up the difference.
The study also mentions that there is an elite group of stock pickers. Why is that the case? And does that mean there will always be a small group who outperform?
Yes, our analysis showed that there is an elite or small group, and that fund management is a skill based activity. I always use a sports analogy in that there is a cohort of people who are the best at their game. Look at tennis, you would not pay £200 to watch a mediocre tennis player but you might for someone who is a Wimbledon champion.
Although there is a move to passive investing, there will always be a need for people who have real investment skills and can add value in an active management approach.
What are the reasons for some of the problems and shortcomings?
Last year, I co-authored a paper in the Journal of Finance – Selling Fast, Buying Slow – and many of these explanations can be found in behavioural finance literature. Take confirmation bias. This is where the PMs would look for stories to justify that good decisions were down to skill, and bad decisions to bad luck. Then there would be the disposition effect as they constantly sold the winners and held onto the losers. This is also known as “chopping the flowers and watering the weeds”.
More subtly, in hindsight, when I worked in fund management firms, there was also clear evidence of the endowment effect because it was very difficult to get PMs to sell existing positions and switch to equivalent stocks with better prospects and better value.
Where does the new product DECIS help the process?
The aim is to help asset owners to see the difference between luck and skill when selecting managers. It has been built over 20 years and analyses $7 trillion of trades in a database. It uses data science to quantify the impact of the decisions that asset managers make when running the concentrated, bottom-up portfolios that have become more prominent today. It analyses the four core investment decisions and processes – or alpha drivers – that we think have the potential to generate alpha in equity portfolios: stock picking, sizing positions, trading activity and holding periods.
This enables asset owners to see exactly where asset managers add or detract value, facilitating more efficient manager searches and better-informed due diligence and monitoring exercises. The model can also be used as an evaluation and coaching tool by fund managers, providing them with granular insight into the decisions they need to improve to enhance performance.
The model is different from the traditional attribution models of the past, which were introduced in the 1980s. They covered portfolios that typically had around 160 holdings and managers would allocate to a sector or region and then select the ‘best’ stocks within it,” Today, portfolios are much more concentrated, high conviction with around 40 stocks, and we thought there needs to be new models.
However, we not only analyse the specific skills that drive returns today, we also quantify their precise contribution to overall performance, showing asset owners exactly how good managers are at making the decisions that matter in modern investment portfolios.
What are the differences and similarities between your asset owner and manager clients?
I will start off with the common thread in that they both want to use data science to improve decision making. We identify the strengths and weaknesses of the investment process, but the information is used differently. For example, the asset owner or allocator, wants to get behind the track records and spin to see what’s actually driving the performance. In the words of one client, “you help us ask the questions we wouldn’t have been able to ask otherwise, and sometimes the ones that the managers wouldn’t want us to ask.”
Managers, on the other hand, focus on improving the investment process. A CIO, for example, may use the information to re-enforce their perceived strengths and to identify and rectify the problems. Up until several years ago that wasn’t really widely understood, but now it’s becoming much more accepted that you can use data to improve a process as well as decision making.
