top of page

Portfolio Prediction

Updated: Mar 13

Author: Tanmey Ajmera


In this modern era, the one thing which is absolutely essential is to have financial literacy. According to a report, only 27% of the total population of our country is somewhat financially literate. Although the covid-19 pandemic had a major effect on the number of people investing in the stock market, as a country, we are still lacking behind as only 5-6% of the population actually invests. Apart from this, people are now investing in various domains.

Since financial literacy in India is increasing day by day, people are investing more and more on their own, yet there are a lot of people who don't know which are the right assets to pick up for investing to be in a sustainable and return-giving scenario. And a very important aspect of investing is diversification; one should diversify their investments in order to make them better. Here comes portfolio diversification which is there to increase the returns, i.e., profit for the investor to the amount invested. Hence, the one-stop solution to maximize your returns.

What is a portfolio?

A portfolio is a combination of many different investments, such as stocks, commodities, etc., which are grouped together in one particular set of investments. Each ready-made portfolio is created by us and invests in a diversified manner of assets across a range of sectors - helping to balance the risk with the potential to invest in the right places. Predicting these portfolios with a low-risk factor could result in efficient investing.


Portfolios are the right way to invest as it is diversified, yet there are many problems faced by an average investor on a daily basis which are-

  • Lack of core financial knowledge

  • Not sufficient knowledge, tools, or resources to do a technical and fundamental analysis of stocks

  • Not enough return on investment mindset

  • Risk-taking ability is different for all people

Taking a look at all these hurdles, we decided to make a PORTFOLIO PREDICTOR based on the needs of an average investor.

A portfolio can be predicted based on the close prices of a particular stock which in turn can be predicted by the LSTM model that is used.

Machine learning Model

The machine learning model that is used in the portfolio predictor is the LSTM model, which is abbreviated for the Long Short-Term Memory model. The Long Short-Term Memory network, AKA LSTM Algorithm, is a type of machine learning neural network that gives the most efficient results on challenging prediction problems. Here, it is used for stock price prediction, and since the financial data is very sensitive and complex, there was no other machine learning model that could handle this type of data.

LSTM Architecture


Architecture of LSTM

  1. Forget Gate

  2. Input Gate

  3. Output Gate

Let’s just not dive into the model deeply. So, moving on, the LSTM model is used to predict the close price of a particular stock. For example:

This graph shows the prediction with the historical data invested in the model and came up with a close price of a particular stock.

Now, once these predicted prices of the stocks are taken into account, then we can predict the portfolios based on these predictions.

Portfolio Generation

For generating the portfolio, mainly two factors are taken into account, i.e., risk and return. So, for the return percentage, we only need the initial price at which the stock or investment is bought, and the final price will be the predicted price. And for the risk part, the beta value is taken into account. Both of these parameters are discussed in the following part of this blog.

A rate of return is the total profit or loss of an investment over a particular period of time, which in turn is expressed as a percentage of the investment's initial cost.


Here P0 is the price at which the stock is bought, and P1 is the predicted price which is predicted by the LSTM model.

Now, for the risk, part BETA value is there. But what is beta value?

Beta is used to determine a stock's volatility in lieu of the overall market. By definition, the market, such as the NIFTY 50 Index, has a beta of 1.0, and individual stocks are ranked based on how much they deviate or fluctuate from the market. A stock that has its chart going above then the market over time has a beta above 1.0. If a stock’s performance via its chart is less than the market, it is less than 1.0. High-beta stocks are supposed to be riskier but provide signs of high returns, whereas low-beta stocks are of a less risky nature but with lower returns. Analysts use it when they want to do a stock's risk assessment. However, beta determines the volatility of a particular stock and predicts a stock’s performance to some extent; it has its own limits for investors looking to determine fundamental risk factors.


Once these beta (for risk) and return percentages are calculated, we can sort or filter out the stocks which have low risk and high return percentages. Hence, the portfolio is generated.


Technical analysis is a very hard and tedious task for an average investor, and overcoming this problem, including the problem of diversified investment, has been solved by portfolio prediction. These portfolios will give maximum returns for the investors and partner them up with a long-term, low-risk, diversified investment without doing the hassle of technical analysis for every asset they need to invest in.

35 views0 comments
bottom of page