As a stock market becomes more mature, the amount of rules vs humans in investing goes up dramatically. In India, the last two years have caused our markets to mature quicker than the entire decade before that. This means, passives are now mainstream. Usually, while passives track an index, active funds try and beat the index by taking stock/asset level calls.
On the active side, most funds are run on human judgment. Active quant mutual funds are still a small category (less than 1% of the entire Assets Under Management of mutual funds). These are recent funds that still need to prove a track record.
Generally, ‘quant’ in India mostly means trading tools like high-frequency trading (HFT) or technical analysis.
At an institutional level, quant mutual funds use a mix of fundamental filters and technical analysis to build portfolios and sometimes add human decision-making. At higher ticket sizes, there are very few portfolio management services (PMS) or Alternative Investment Funds (AIF) that use quant to invest.
This is different in the US where one in three hedge funds claim to use some sort of quant to invest. This is because the quant model has proven to be more dynamic. If the right models are built, they should learn and change with market conditions. Quant allows a fund to be a different kind of investor in different markets. Therefore, the two primary pillars on which any good quant shop should rest are systemized investing and dynamic rules.
Key questions
Any machine is as good as its maker. Thus, it is important to ask the right questions to evaluate whether the quant model is built on strong pillars.
Data quality: Is it clean, complete, and accurate? You should ask or read the documents to understand what the source of the data is, how the fund cleans-up for missing, inaccurate, non-standard data, what is the process the fund follows to verify accuracy and how often is the data and process updated.
Quality of the model built: Just like human investors, quant models can be of differing quality. Ask questions about the logic, is it dynamic or static, is it robustly tested, what risk factors are built into the model. For example, if the model is supposed to have limited downside in bear markets; to check this, ask for data of performance in March 2020 or 2008.
Quality of back tests: Since models have shorter live history, you should ask how the back tests are performed, the companies/ time period considered and the techniques used. For example, are the tests blind, i.e. when the model is testing in 2019, it must have no information post-2019 in the system.
Quality of team: understanding whether the team is stable or there is high churn is important. Also, questions like ‘What is the founding team’s background in machine learning? Have they outsourced the tech? What happens to the tech if the team leaves? ‘are important.
Qualitative issues: How does the machine solve for corporate governance (i.e., catching frauds, operator-driven stocks, etc), management quality, etc. Perhaps the answer is that they rely only on numbers and no qualitative factors at all – understand their reasoning, what happens in extreme events, and their experience of dealing with frauds in companies they have bought.
Portfolio fit: Do the stocks of the fund fit with your portfolio and risk profile. For eg, if it buys only large caps, are you better off in an exchange traded fund?
Time horizon: Ask yourself, what your investment horizon is. Like with any equity strategy, you should have a horizon of 3-5 years for the quant to work in different market cycles.
There could be other challenges with quant investing. For example, in 2010, algos caused the US markets to drop 9% in minutes in a “flash crash”. Twenty years ago, LTCM, a quant fund started by Nobel prize winners, was wound up after its algos broke down during the Asian financial crisis. These stories are more high profile because computers caused these problems.
Nevertheless, quant investing will help in covering the human blind spots of bias and emotion. Over the next few years, we will see more rules-based products finding a place in investor portfolios.
Kanika Agarrwal is the co-founder at Upside AI.
