The screener can help discretionary investors narrow down candidates from a larger investable universe. We select companies with a high number of news events and high sentiment polarization, a task that can be achieved using our Sum Excess Sentiment Indicator, SESI .
The main conclusions are:
- News Sentiment significantly outperforms price momentum as a stock screener, for which probability distributions clearly demonstrate a positive shift. This is more than a 4x increase in average performance during the 8+ year backtesting period.
- By creating targeted portfolios, we achieve high annualized returns (up to 80%) with attractive Information Ratios (IRs) of up to 4.0, but also high per-trade-returns (up to 33bps).
- When including a discretionary overlay, manually dropping 2 out of 22 daily trades, annualized returns and Information Ratios are boosted by up to 200%, depending on our skill at removing bad performers.
Alternative data is the “new black” of the investment industry. Today, rarely anyone would question the potential of using new and unique data sources to seek alpha opportunities. However, new opportunities often come at a cost.
To leverage alternative data, quantitative trading firms are progressively driven to boost their investments in research and IT infrastructure. Once obtained, this data is evaluated with the objective of converting it into alpha strategies. Most of the alpha strategies using alternative data work in the framework of weak predictor aggregation. That is, when combining individual predictions, the individual accuracy does not need to be very high in order to obtain a stable and strong-performing strategy.
Two key points needed for these quantitative strategies to work include: (1) signal breadth and (2) signal frequency. The former implies trading large universes, involving hundreds or even thousands of equities; the higher the number is, the better the strategy works. The latter defines the frequency at which the strategy can be rebalanced using new insightful information; the more times this occurs, the better it works. As a result, investors will typically desire to consume more and more data, and expand into new markets and trading horizons.
RavenPack Analytics provides a source of alpha that directly falls into the above described scenario: a fast feed of information, involving thousands of entities, which has been shown to predict market movements [2-6]. Accordingly, the resulting strategies, with no additional considerations, may still require more effort, both in research and in trading: in research, because they may be coupled with other alternative data sources or traditional market factors, optimized under the desired portfolio constraints, etc.; in trading, because they are generally destined to be part of a large portfolio framework that combines many other alpha signals.
Instead of trying to generate as many signals as possible, as you would expect in a quantitative strategy, in this study we aim to reduce the signal volume and build sentiment strategies that can be more attractive to discretionary traders. Often, traditional investors don´t have the necessary resources to test and manage large universes or they prefer to focus on fewer equities, where they can apply more traditional human-based trading methodologies.
Our objective is to offer an easier way for discretionary traders to incorporate alternative data into their portfolios, pointing them in the direction of new alpha sources, using a customized filtering layer that reduces the heavy resources involved in working with alternative data.
To this end, we propose a way to select trading signals that can be used to build portfolios or equity baskets of a desired size. This selection is based on choosing stocks with a high number
Click here to download the PDF and continue reading the "A News Sentiment Stock Screener for Discretionary Traders" White PaperRequest White Paper