Carpe Ventures is a research startup which uses the potential of new NLP strategies to improve market projections. Specifically, we are using the ever-growing powers of NLP machine learning to find non-linear relations across the markets as a way of uncovering new, original alphas.
For the past two decades, trading algorithms and econometrics have taken over the global markets. In fact, JP Morgan estimates that as of 2018, 60+% of all trading is algorithmic. To add, the Eureka Hedge Index which tracks the performance of all ML-based hedge funds, has been growing at a linear rate since 1999, and will continue to do so in upcoming years.
At the same time, the vast majority of today’s algorithmic trading is founded on stochastic methods, probability functions, and other complex optimization functions. This means there is limited room for implementing new mathematical models into algorithmic trading. Also, with trading algorithms taking over modern finance, the future of investing requires strategies that can analyze other algorithms and predict their patterns and behaviors. More generally, there is now a need for strategies which can find new patterns embedded within the markets — Patterns which cannot be reached with conventional mathematical modeling.
This is not to discredit the use of traditional statistics and mathematics in trading. In fact, Carpe Ventures’s strategies rely heavily on Bayesian Inference and other probabilistic models. The key difference is that these models do not dictate our approach. The driving force behind what we do is natural language processing.
Specifically, we are harnessing the power of dynamic embeddings which have recently been introduced to us through unsupervised ML systems such as Word2Vec and BERT.
We are currently based in Portland, Oregon and New York City. We are always looking for driven new members who share our vision for the future of quantitative finance to join our team. If you are interested in our work and would like to get on board, fill out the form below or email: firstname.lastname@example.org.