In the fast-paced world of equity markets, outperforming requires access to sophisticated tools and data. MarketPsych Analytics are table stakes for quants, analysts and data scientists seeking to enhance their alpha-generation. This rich dataset analyses news and social media in real time and publishes sentiment and thematic scores for every asset. The data shows value across a broad range of investment horizons and is consumed by a firms spanning intraday traders to mid-frequency to pension funds looking for multi-year outperformance.

Standalone Signal and Additional Factor

Data applications within the quant space include machine learning-based equity strategies, improving event trading around earnings, capturing turning points in momentum, and arbitraging across mispriced assets.

The thematic sentiments published on equities can also be directly used in standalone portfolio construction. For instance, companies with higher public trust in their management teams tend to outperform, and rotating a portfolio monthly of the most trusted companies results in sustained outperformance over the last 20 years. Other thematics also have similar properties, and it was research performed internally and externally on these thematic sentiments which led to the development of MarketPsych’s equity forecasting model.

Short & Long-term Investment Horizons

Academic Studies

Research shows that corporate perceptions expressed in global investment media correlate with stock returns over multiple horizons. With over 100 academic papers supporting this dataset, the relationship between attention, sentiment, returns and volatility can be thoroughly explored with MarketPsych data.

The image above was generated by first averaging the media sentiment of each stock over the prior 30-days. The stocks were then ranked by their sentiment scores and binned into color-coded decile (10%) portfolios. The performance of each decile was then tracked from the first trading day of the next month over the following 90 days. The results from all periods from 2006 through 2020 were averaged together. The average annual spread between the top and bottom deciles is approximately 2.5% annually for U.S. stocks and 4% globally. As described further below and in our whitepapers, with tuning and additional features, the annual spread both stabilizes and widens to greater than 10%.

StarMine MarketPsych Media Sentiment Model

Co-branded with StarMine, the MMS model uses MarketPsych Analytics as its sole input. It produces a 1-100 rank for regional baskets of equities which forecasts their 30-day return relative to the basket. This models has run live and unchanged since January 1, 2020.

A strategy which takes a long-short portfolio of the decile extremes has maintained a Sharpe ratio above 1 since release with a market-neutral annualized return of ~10%. As a factor, the model is uncorrelated with other factors and complements other StarMine models.

Figure: Cumulative performance of the MarketPsych MMS Signal incorporated into a monthly rotational standalone portfolio for US equities. The top- and bottom-ranking equities are held in simulated portfolios for 1 month and before rebalancing.