TPT’s mission is to help bring asset management into the 21st century.
We focus all of our efforts on one task: uncovering persistent inefficiencies in financial markets. The client keeps the assets, so that all our resources are directed towards alpha generation. This, combined with our relentless automation, allows us to deliver performance at much lower costs than those charged by traditional asset managers.
For the past two decades, TPT’s researchers have made some of the most impactful innovations in financial machine learning, covering every stage of the investment process. These contributions include:
Feature engineering: Order imbalance sampling; the triple-barrier method; trend-scanning labels; volume-synchronized probability of informed trading
Strategy development: Stochastic flow diagrams
Bet sizing: Meta-labeling
Backtesting: Combinatorially-purged cross-validation; Monte Carlo backtests
Strategy selection: The deflated Sharpe ratio; the probability of backtest overfitting; the “false strategy” theorem
Portfolio construction: Hierarchical risk parity; nested clustered optimization
Execution: Optimal execution horizon
OUR EXPERT NETWORK
Conventional methods yield conventional outcomes. Beating the collective wisdom of the crowds requires innovative approaches, produced by best-in-class talent. For this reason, TPT partners with domain-area experts for each specific project.
Through our connections with the leading universities and National laboratories, we assemble the right team for every particular mandate. Our network includes 30 of the best-known authors in mathematical finance, machine learning and supercomputing. We have access to hardware and software that is years ahead of what commercial services can offer.
This flexible team structure enables us to deliver answers that are novel, insightful, authoritative, and scientific.
TO LEARN MORE
Advances in Financial Machine Learning (Wiley, 2018) explains in detail how TPT’s strategy factory approach works. One year after its publication, this graduate textbook has been translated into Chinese (China Citic Press), Russian (Progress Knijga), Japanese (Kinzai Institute) and Korean (Acorn Publishing), and it is taught at leading universities worldwide.
Its follow-up, titled “Machine Learning for Financial Researchers”, is scheduled to be published by Cambridge University Press at the end of 2019.
These two textbooks form the academic core of the ORIE 5256 course, at Cornell University’s School of Engineering.