Alacra targets long-only traders with algo news offering
Alacra, a trading technology provider, has started testing its PulsePro web-based news application with US traders to develop a machine-readable news offering for long-only buy-side firms.
PulsePro, a web-based news gathering service, offers event-specific filters including street pulse, which includes comments by sell-side, ratings agency and industry analysts; chief pulse, which includes remarks by ‘C-level’ corporate officers;
deal pulse, looking at news and commentary on potential and announced mergers and acquisitions; and weak pulse, corporate distress signals such as debt restructurings, reorganisations, layoffs and bankruptcies.
Alacra has started testing PulsePro with a handful of US firms with a view to extending the service to work with algorithms used by long-only asset managers.
“We started assigning sentiment scores to news events for a part of our offering, mainly out of curiosity, but soon realised that despite the high-latency, low-frequency nature of our data, we could use the information in PulsePro to act as signals for trades that have a longer time horizon,” Alacra CEO Steven Goldstein told theTRADEnews.com.
PulsePro is typically used as a tool to extract news using a four-step process –source curation, event detection, semantic analysis and human editorial oversight – to gather specific news for asset managers, investment bankers and professional services executives.
Machine-readable trading applications from vendors such as Bloomberg and Thomson Reuters are mainly geared towards quantitative and high-speed traders looking to feed automated trading applications.
However, Goldstein observes that even long-only traders do not have the time to assimilate all the news stories that may be relevant to a certain trade.
“Some attempts at using sentiment have been unsuccessful because the software doing the analysis looks at large quantities of information,” said Goldstein. “For example, a news story may contain a lot of false positives or information that can’t be assigned a sentiment score. We have built software that can extract only a few sentences from a news story, meaning the events we tag have a higher degree of accuracy.”
After current testing on S&P 100 stocks has been completed, Goldstein added that Alacra plans to start testing in Europe, most likely in the UK with FTSE 100 stocks.