Introduction to Factor Scores

At sharpely, we have a number of proprietary stock scores that can help users evaluate stocks on a particular factor or a combination of multiple factors.

The idea behind creating these stock scores is inspired by tons of academic research on factor investing.

Factor-based investing started with Arbitrage Pricing Theory in 1976 and gained popularity after a seminal paper by Fama and French in 1993 which introduced a 3-factor model for asset pricing. They argued that apart from the market, there are 2 other risk factors (value and size) that explain the returns of an asset. In other words, these factors generate a risk premium over the long term for investors who are exposed to them.

Since then, researchers have uncovered other factors that generate risk premiums like momentum and quality. The factors mentioned above are called systematic factors as they carry risk (just like the market) and hence compensate investors for bearing that risk (risk premium).

For a factor to be relevant, it should have worked in the past and should continue to work in the future as well – the factor must have an underlying risk or behavioural explanation for the existence of a positive risk premium.

To that extent, researchers have found that Value, Momentum, and Quality are 3 systematic factors that have worked in the past and will continue to work in the future as well. In this article, we will not go into the details of historical performance and the underlying risk/behavioural explanations for these factors. We will cover that in our series on Systematic Equity Investing.

These 3 factors and various combinations of them form the core of our proprietary stock scores.

Score-aggregate-score methodology

Stock scoring is based on a standard score-aggregate-score methodology. Let’s understand this with an example:

Suppose we want to create a factor score using the composite of 4 metrics – A, B, C, and D. Here are the steps we follow:

Step 1: Using our standard scoring methodology, create a standardized score for each factor such that scores are between 0 and 100 and a higher score is always better.

Step 2: Aggregate these scores by taking a weighted sum of the individual scores of A, B, C, and D.

Aggregate Score = W (A) x S (A)+ W (B) x S (B)+ W (C) x S (C)+ W (D) x S (D)

Where W (A) is the weight of metric A and S (A) is the score of metric A.

Step 3: Rescore (0 to 100) the stocks based on the aggregate score to arrive at the final score.

On a side note, this score-aggregate-score methodology is also used for creating Factor Models.

We will first discuss single factor scores (Value, Momentum and Quality) and then discuss multi factor scores (combination of Value, Momentum and Quality).

For more detail on our exact scoring methodology refer to this article. Also, on a side note, users can access score for any metric using Scorecards.

To avoid dilution of these stock scores, we limit the universe to all actively traded stocks on NSE that have a market capitalization of more than Rs 100 crore.