Legacy architecture and an overwhelming amount of unstructured data make it difficult for finance professionals to perform forecasting, increase returns, and manage risk. When sound investment decisions rely on machine learning models with human-level precision and recall, Primer Engines instantly deliver performant, fast, and scalable AI. Engines are trained and tuned to work on domain-specific data for finance. They empower financial service companies to quickly extract, surface and connect the hard-to-find, but crucial information needed to advance their business.
A fintech company tested multiple leading NLP solutions — Primer outscored them all across all key performance metrics. Read the blog post
Rapidly uncover trading opportunities hidden in 10-Ks, analyst notes, publicly available research, and news sources
Traders pore through massive amounts of industry and historical data every day to identify trading opportunities and get that extra edge that will maximize their returns. With Primer Engines, traders can connect engines together to build custom data processing pipelines and find hidden connections across proprietary and public resources. Here are a few sample Engine pipelines:
Surface shifts in industry analyst reporting
When researching a particular industry, traders look for insights that will signal trading opportunities. By building their own Primer Engine pipeline, traders can find a gold mine of information about a public company like General Motors.
Named Entity Recognition (NER): Identify all analysts in a set of documents who have reported on General Motors.
Quote Attribution: Extract and attribute quotes from all the analysts about the company.
Sentiment Analysis: See how analyst sentiment about General Motors is changing over time and relate this to specific topics of interest.
Find the signal in the noise
Traders have too many sell-side reports to read and need to quickly make sense of what’s happening in their portfolio, yet all these reports seem to say the same thing. With Primer Engines, they can build their own pipeline to help them figure out what is signal and what is noise.
Key Phrase Extraction: Identify the most important phrases in analyst reports.
Topic Modeling: Identify the key topics that analysts are writing about for the company/sector they care about.
Difference Engine: Use this to uncover only the novel ideas and arguments in each sell-side report. Remove all text that is broadly repetitive.
Summarization: Set compression rates for summaries to automatically generate human-quality reports with key takeaways to share internally.
Identify regulatory impact on a company’s business operations
Traders who are interested in a particular company need to understand how it is perceived in major global markets and how regulatory changes may impact the business. For example, a Primer Engine pipeline with multiple engines building blocks can help traders analyze the business climate in China for a company like Tesla.
Named Entity Recognition (NER): Take a set of documents and identify all the people contained within this set.
Relation Extraction: Show the relationship between people and the Chinese ruling party.
Quote Attribution: Match quotes with the individuals of interest. Filter quotes via mentions of Tesla or Tesla products.
Sentiment Analysis: Uncover trends in positive or negative sentiment towards Tesla over the past two years.
Identify a set of companies to match your investment thesis
Traders want to find a set of companies that have an increased competitive risk exposure to China. Now with Primer, they can connect Primer Engines to create an output of a list of companies that are facing increased competitive risks from China. From there, traders can decide to go long or short on this bet.
Classifiers: Extract all risks from every 10-K report published over the past two years. Then identify all risks that are classified as “competition risks” or risks that the business would suffer from competitive pressures.
Difference: See if these risks are ranked higher this year than they were last year.
Location Extraction: From the companies that show an increase in competitive risks, identify all those that mention locations in China.
Identify signs of a deal or partnership that is not yet public
Traders look for insights into a company’s direction that indicate potential investment opportunities. They can assemble a Primer Engine pipeline like this one to get alerted of early indicators and take fast action.
Named Entity Recognition (NER): Identify public documents associated with a particular executive of interest.
Location Extraction: Find out where the executive has travelled to in the last 18 months.
Quote Attribution: Extract specific quotes made by the executive about a specific location.
NER on Social: Pull statements made by the executive on social media and extract all of the locations of interest.
Topic Modeling: Surface all of the specific topics the executive cares about to see if a new topic is emerging over time.