Insider Views

Derive Actionable Insights – Improve Your Investment Returns Using NLP

YUKKA Lab focuses on the sentiment analysis of financial news to combat the growing problem of information overload for knowledge workers. We understand that the increase in information in general as well as the rise in fake news can have negative impacts on business and investment decisions.

Never miss insider information, top jobs and mandates. Sign-up here.

Our proprietary in-house built algorithms analyze over 500’000 news articles from 100’000 different, trusted sources in real-time through what we call Augmented Human Intelligence. As news articles are examined through our Natural Language Processing (NLP) algorithms, advanced machine learning techniques extract and allocate sentiment scores to the entities that are affected by the specific information.

As we will discuss in this article, not only can these insights facilitate a more efficient overview of market movements, but sentiment can also act as a proven early warning risk indicator. It provides awareness and a nuanced view on the development of corporate and macro events.

Custom-designed framework

On the first point, using our custom-designed signal development framework we have identified several key areas within the investment cycle where sentiment can significantly augment existing components. These include methods such as style factor/smart beta investing, event-based signal construction, trade execution and dynamic hedge indicators. Given the recent turmoil seen in the markets due to the COVID-19 outbreak, it only seems fitting to dig a bit deeper in the last application.

Our dynamic hedge indicators are built by aggregating sentiment scores on single stocks within each industry and comparing them with their long-term averages and ranges. These are then combined to an index-level score on a broad market index like the STOXX 600 Europe, indicating a desired exposure to the market. Since the sentiment is an indicator of market participants’ outlook on the underlying constituents, a significantly negative indicator value can be expected to serve as an early warning on potential upcoming drawdowns. Fund managers and traders can, therefore, use our indicator as another pillar to adjust their market exposure.

A model to clearly reduce exposure such as trade war tension or outbreaks

The resulting indicator together with its outperformance chart is shown in the above graph. First, we want to stress that the model was fit with data up to 2017, meaning the period from 2018 onwards is out of sample. Over the next years, our indicator manages to avoid a significant drawdown in the second half of 2018, coinciding with a peak in US-China trade war tensions, and then most importantly it reduces the exposure very early on in the COVID-19 outbreak and resulting market pullback. Equally important, we see that the system increases exposure again in a very timely manner, therefore participating in a large part of the rebound. In numbers, by using our hedge indicator the maximum drawdown was reduced by around 60% and a total outperformance of 30% was created over the 5-year period.

Our continuous hedge indicator provides a template that can be used to tailor the signal to individual use cases. For instance, not everyone is interested in adjusting their market exposure on a daily basis, so one of our extensions modifies the indicator to a binary signal; either the investor remains fully exposed to the market, or when sentiment hits a critical value a full hedge is put in place. Of course, the extensions are endless: instead of making it binary, we can introduce different levels. Going to 0% exposure is not really feasible, the lower bound is set at some intermediate level, or leverage can be introduced by increasing the upper limit above 100%.

YUKKA News Lab offers further diversification

Diving into the second application of sentiment, our Software as a Service platform, the YUKKA News Lab, further diversifies the offering by creating a set of cutting-edge data analytics built on the very same engine that powers the financial signals. The platform exceeds at providing a methodical approach to monitoring the macro-trends affecting the world and micro-trends specific to a client’s universe and point of view, in real-time. One of 2020’s overarching topics has been the COVID-19 pandemic, with far-reaching impact on industries and countries globally. To help our clients be better prepared for the pandemic, we used our News Lab platform to aggregate all the news content around COVID, from the first flu-like symptoms in the automotive hub of Wuhan through the ongoing pandemic. Our analytics around the sentiment of the aggregated data painted a clear picture of the spread of the virus, from a global macroeconomic level down to every single country, sector and enterprise affected.

Customized reporting

Leveraging this dataset, we were able to create customized reports for clients that highlighted top winner/losers on a sector and enterprise-level per country, even identify repeating cyclical patterns around consumer behavior (e.g., sanitary equipment and food supplies). As a result, not only does this data provide a clear macro-picture of the pandemic, but also actionable insights for the sales and client advisory arms. In Switzerland, on request from a top Swiss Bank, we used the News Lab to distill this dataset into an actionable lead list of 500+ SME Enterprises in Switzerland affected by the pandemic, categorized by types of effect such large sales volume changes, production bottlenecks, supply chain disruptions and workforce problems.

In conclusion, whether you are looking for a qualitative market analysis using sentiment or a more quantitative approach, YUKKA Lab can assist you in augmenting your workflow and gain a further edge on your competition.

About the author


Through our Augmented Language Intelligence, we combine state of the art NLP, Machine Learning & AI to provide experts with a systematic overview of the relevant sentiment trends in the news for their job role. They can see the topics that fuel those trend shifts and drill down to the relevant articles and at a glance see where the company and where the topic is mentioned. This way experts have the overview over all relevant trends without having to read a single article.

Our technology empowers experts to act on trigger events detected in the news and thereby to augment their performance. Making them not only more efficient but also adding new additional value to their processes and products. Our indicators for sentiment and mentions of companies, industries, regions and events serve as an additional pilar for better decision making in investment and risk management.

Our proprietary Augmented Language Intelligence combines cutting edge natural language processing, machine learning and artificial intelligence to analyse over 500'000 articles per day from more then 100'000 global sources in near real-time. As a data agnostic analytics pipeline we can integrate any content set. For different use cases we feature different source panels that range from Pro Feeds, Licensed Print to Online Content. Our solutions empowers different experts such as asset managers, advisors, risk managers, market researchers and B2B sales managers to be on top of the latest news and always one step ahead of the curve!

Reach out to us to learn how we can augment your business and processes!