About Prediction Machines

Prediction Machines builds intelligent and intuitive algorithms to capitalize on the commercial opportunities across trading venues, market places and other transacting portals.

At Prediction Machines, our research team is adept at working with data to identify the salient features relevant for making commercial predictions. We deploy a broad spectrum of techniques within machine learning theory for classification and regression when working with such data.

Deep domain knowledge can provide a crucial advantage when attempting to identify the relevant features of a commercial exploit. Prediction Machines works closely with domain knowledge experts to identify and characterize such features which might not often lend themselves to traditional classification methods in machine learning.

Learning From Simulations And Games
Exploiting a commercial opportunity will often take the form of a transaction, buying or selling something, as a result of a prediction. There is an action (the transaction) from which there will in the future be a potential commercial benefit.

Learning to make optimal transactions is a nascent extension of machine learning theory. The application of simulations or games which model the commercial environment and embed the salient features being exploited provides a setting for our proprietary algorithms to hone their skills for completing optimal transactions.

Our researchers and technologists work with the latest techniques in deep learning theory and reinforcement learning to develop algorithms that effectively and profitably exploit commercial opportunities.

Our technologists have significant skills and experience in automating data capture, processing, analysis, integration with predictive algorithms, transaction processing, reporting, and record keeping; including where necessary, regulatory reporting and compliance monitoring.

We don’t just build the algorithms of predictions; we build prediction machines.

Prediction Machines - Blog


Our blog is an excellent source of current research and developments in deep learning and other active machine learning topics.


Academic Outreach

Prediction Machines uses methods that are at the forefront of academic research into the fields of deep learning, reinforcement learning, and simulation/game-based learning. We value our research collaborations and relationships with academia.


If you are a startup business making use of predictive methods we are interested in providing advice, consultancy, and seed-investment capital into your venture.


2 Battery Road #28-01
Singapore 049907
Tel: +65 6532 1252


15 Stratton Street, Green Park House
London W1J8LQ, United Kingdom
Tel: +44  203 036 0256


7 Chome-20-5 Roppongi, Minato-ku
Tokyo 106-0032, Japan