[frame style=”modern” image_path=”” link_to_page=”” target=”” description=”” size=”two_col_small”]


Financial analysts and investors currently read news that dramatically affects the pricing and valuation of securities. Bad news often means a down day on Wall Street. Currently researchers must manually manage a great deal of complex information in order to measure and analyze these correlations for their reports, creating tremendous room for error and major delays in reaction time. Therefore, rigorously quantifying the relationship of financial news to financial securities, in real time could have significant impact on overall financial performance and returns.

HybridSTAT is an exclusive collaborator of Patterns and Predictions and represents its products in the European region.

[h2] Centiment Offers [/h2]

  • A mature statistics engine – Based on Patterns and PredictionsÔÇÖ industry┬áleading Bayesian statistics engine for advanced liguistics classification.
  • Cutting edge econometrics – Through a unique relationship with ┬áDartmouth/Harvard research ongoing in the area of semantic overlays ┬ábased upon observations and market psychology.
  • Multiple data source loading – Via the Attivio AIEÔäó technology, our system┬áloads data of nearly any document type.
  • Portfolio Stress Testing – Utilizing the S+ runtime, our system allows a large┬ávariety of out-of-the box portfolio testing, as well as the ability to extend┬áand customize analysis.
  • Industry recognized trading output – Fix ProtocolÔäó ┬á(

[h2] Supports [/h2]

  • Black box trading (as standalone trading system).
  • Overlay signal approaches (as a secondary signal to integrate with an existing trading method).
  • Any financial data format: Dow Jones, Reuters, Bloomberg, open source news.
  • Latest Computational Linguistics techniques supported: IR (TF/IDF),┬áBayesian, SVM.
  • Multiple Performance Metrics: Returns, Excess Returns, VAR, MAD, Sharpe Ratio, Jarque-Bera, Sortino, Downside Deviation, Omega.

[h2]Full Customization[/h2]
Integrate with existing solutions, or test new trading strategies.

Patterns and Predictions does certify that in all tests the semantic methods have met or exceeded our baseline signals (for instance S&P SPx), and that we will not consider a system deployed until the aggregate performance is nominal for our sample data (i.e. alpha > baseline signals).

The clientÔÇÖs data is the clientÔÇÖs intellectual property and ÔÇÿedgeÔÇÖ. Therefore, our professional services integration will offer the highest level of privacy with regards to client data.