Equity prices drift away from fundamentals due to technical flow. Instead of using moving average to smoothen price movements, we apply deep learning techniques to extract stocks expected to exhibit momentum in next 1 month.

Why Deep Learning ?

• Traditionally, statistics has been used for markets, but the statistical methods implicitly assume normal distribution whereas markets are skewed and hence dont follow this pattern.

• In contrast deep learning models dont make such assumptions and hence can fit better the way markets behave.

Is it a black box ?

• Yes, all deep learning models are black boxes where the exact factors are never revealed. However, as this is self learning in nature, its expected to be dynamic

Can you please explain the process ?

• We take long term equity prices and use deep neural network to filter prices for drifts by training a deep neural network on large volume of price data.

• This is then use to estimate momentum as clean prices instead of noisy and raw prices.

• We believe is useful as momentum is usually mean reverting - meaning stocks that exhibit positive momentum usually eventually revert back, hence spikes are smoothened to remove what we think are false positives.

DISCLAIMER: This portfolio is not available for US persons, including US citizens and residents. The portfolio represented herein is available only to Indian residents. For US persons, please refer to the LotusDew website: www.lotusdew.co