Numbers rule the universe, according to Pythagoras. Whether the world in general believes this or not, many in the investment world believe so. The proliferation of quant investing globally -- a theme which is gaining traction in India as well -- is a testimony to that. Quantitative investing refers to investment strategies in which buying and selling of shares/investments is done primarily on the basis of a process-driven analysis of numerical data.

One of the biggest advantages that quant investing brings to the table is elimination or reduction of cognitive human bias in investing decisions. Volatility in markets can trigger human emotions of greed and fear and influence decision-making.When left unchecked, this will most probably yield painful results.

This apart, even during non-volatile phases in markets, in a digital era where a person is constantly bombarded with information and views on stocks across mediums, the propensity to be subconsciously influenced by this to making wrong or unnecessary investment decisions is high. Investing on the basis of a quantitative investing process is an ideal tool to deal with these human behavioural constraints.

Orginally, a domain exclusive to sophisticated institutional investors, growth in technology/internet and information dissemination have today made it a strategy within the realm of retail investors as well.

Building your model

If you are a seasoned investor, or even a beginner with a discerning eye for numbers, you can attempt to build your own model for quantitative investing. Building a good model involves the following five steps.

Metrics

Metrics represent the key factors that drive the stock performance. While it may vary according to investing strategies, some of the basic factors that drive the stock performance are valuation, growth opportunities, profitability, quality of earnings, management quality and financial strength. To this one can also add technical metrics as an additional factor if required.

Investors need to note that no single metric can give any valuable inference unless it is seen in relation to other factors. For example, valuation of a stock does not indicate much unless it is seen against factors like growth opportunities or financial strength – a stock may appear cheap, but this may be because its business is on a decline and it is expected to make losses in future or because it is highly leveraged and may face liquidity issues in future.

Sub-metrics

While valuation is one of the important metrics for all stocks, different valuation sub-etrics are important for different sectors. Important valuation metrics for the IT sector might be pice-to-arnings, enterprise value/sales, enterprise value/free cash flow; for the banking sector might it may be price/book and PE; for REITs and InVits it will be dividend (distribution) yield.

The reason to choose a combination of sub-metrics is to address skewness that a single sub-metric might display. For example, a company may score poorly on PE ratio, but might score well on EV/EBITDA. This implies that while profits are currently impacted by higher interest and depreciation (fixed costs), this could be offset by higher growth in revenues and benefit the bottom line due to higher operating leverage.

Ranking

This is very crucial as the rank score would play a key part in arriving at the final overall score of the stock. There are many ways to rank stocks. For example, the simplest option would be to rank stocks in ascending or descending order based on how it fares vs peers in each metric or sub-metric.Ranking using statistical techniques like Z-score is another. Ranking stocks into deciles or quintiles might be ideal in some cases.For example, in the illustrative table accompanying this story, the tier 1 Indian IT stocks have been ranked on the basis of quintiles. Under this ranking, irrespective of the number of stocks in the universe, the top 20 percent would get a score of 2 and the bottom 20 per cent would get a score of -2. Those in the middle will get a score of zero. Symmetrically ranking stocks will place them relative to the average company in the universe.

Assigning weights

This involves 2 parts – assigning percentage weights to the sub-metrics and assigning percentage weights to the metrics. The total weights in both cases must add to 100 per cent. Assigning weights will have to be done based on understanding of the sector and determining which of these add more value in picking the right stocks. For example, if you are building a quant model for growth stocks, at the level of the metrics, valuation might be less important than growth of the company. In such a case a higher weight to growth needs to be assigned. It would be the converse in ranking value stocks.

Back testing

Whether one has built a robust model or not can be validated only by testing it. Quant investing scores over other forms of fundamental investing as it gives an opportunity to back test the model.

A good back testing process will provide feedback on how the model – metrics, ranking methodology, weights etc. – need to be re-worked or tweaked to make the model more robust. For example, by changing weights for metrics one will get different selection of stocks each time and can then analyse which selection/assignment of weights yielded better returns over a period of time. Some open source software tools might be helpful for this process. Investors who don’t have proficiency in such tools can concurrently test their model/s for a year or so and then get into real investing based on the model that generates good and consistent performance.

Hygiene factors

Oscar Wilde once said: ‘Success is a science, if you have the conditions, you get the results’. Similarly, getting good results for a quant model requires having certain conditions in place.

One, a basic condition for success is that the portfolio must not be concentrated. This is for the simple reason that quant based investing in most cases does not involve in depth qualitative research of stocks and hence, risk must be diversified.. A portfolio of at least 25-30 stocks with no single stock exposure higher than 5 per cent would be ideal.

Two, while building the model and determining metrics/sub-metrics, it must be done after categorising the larger universe of stocks into clusters of companies that are similar in terms of business. It would not make sense, for example, to rank IT and pharma stocks together and then pick the stocks that rank on top.

Three, avoid frequent interventions in stock portfolio based on daily news flows. Usually, quant funds focussed on long -term investing, follow a process of doing a periodical refresh of the quant model on a monthly or quarterly basis to rebalance portfolio and do not intervene to change positions in the intervening period, except in the case of black swan events.

Four, ensuring data quality. There are many online data sources, some free and some paid that provide data required for quant models. https://tinyurl.com/hblscreener provides some such sources.

Subject to such conditions being fulfilled, one may get the desired results.

Readymade quant options for your portfolio

For those who don’t have the time, inclination or skill to build their own models, some mutual funds in India and US-based quant ETFs are readily available options.

MFs in India

Quant investing in India is an emerging field and there are five quant investing based mutual funds in India now -- Nippon India Quant, DSP Quant, Tata Quant, ICICI Pru Quant and Quant Quantamental. Being niche thematic funds, they manage less than ₹ 1,000 crore put together. Except Nippon's fund, most of these products have been launched after 2019 and so their track-record is limited. The three quant funds with at least a one-year return have under-performed the average returns of thematic category funds as well as the Sensex. However, it may be too early to judge them.

The quant funds follow rule-based investing, with most of them having a starting universe of S&P BSE 200 stocks. The focus is on limited human intervention to avoid any biases from creeping in. Different quant funds often follow different strategies. The DSP fund, for instance, eliminates highly leveraged companies and chooses stocks based on high return on equity, and earnings growth consistency and potential. Tata Quant uses artificial intelligence and machine-learning to pick stocks, while Nippon India Quant follows a quality-cum-momentum strategy.

US ETFs

The US quant ETF industry is now almost close to 1.5 trillion in AUM as per a recent report by Bloomberg. Some of the largest ETFs in the space are the Vanguard Value ETF and iShares MSCI EAFE Value ETF. Investors can check out websites such as https://tinyurl.com/quantbl for a more comprehensive list of quant ETFs to invest in. Before investing one must clearly understand the strategies adopted and the risks. Different strategies will have different time horizons and that also needs to be factored in. Being thematic funds, it is best investors take only a 10 per cent exposure to quant strategies.

COMMENT NOW