APPLICATIONS IN FINANCIAL MARKETS

ZODIAC

We have built a very fast multi-threaded risk management and trading framework. Zodiac uses a Hidden Markov Model (HMM) to recognize the current market "regime" and the probability  of "regime switch" out of the various possible different "regimes" of the world. For example, the model recognizes whether the financial markets are in a calm regime, tense regime, volatile but positive regime, volatile and negative regime. The risk management is context-sensitive. Conventional approaches like Value at Risk (VaR) will over-estimate the risk in a "calm regime" and over-hedge, leading to reduced profits and under-estimate risk  in a "panic driven regime" therefore, not applying enough risk-control and lead to larger losses than estimated. Our approach is able to recognize both the current regime and the probability of a "switch" to different market regime, and act accordingly - Zodiac's approach is novel in this regard. Zodiac uses probability density mixture models, unsupervised learning, Hidden Markov Models for effective risk management. These  are all sophisticated techniques,  we designed the Zodiac application after adapting the algorithms for our own specific use. Secondly, the Drawdown Control Optimization is also made by us and we had to do R&D/ try a number of different schemes before coming up with a good solution- this is not available in competing products.

Zodiac has an API for automated pattern recognition allowing the trader to flexibly define patterns for trading. There is already a rich set of patterns included for trading equity index futures, bond futures, commodity futures and volatility index futures. Three different data feeds can be utilised (Bloomberg, CQG and IB). The FIX protocol is utilised for position management and order execution.