Blue Fin’s strategies are systematic in order to remove subjectivity from the trading operation and to participate efficiently in market dynamics that our research has identified as potentially profitable. The specific market behaviours Blue Fin participates in can be seen on intraday and 1-3 day time frames (sudden range expansion) and more medium to longer-term time frames up to six months (serial correlation and trend persistence).
To ensure our approaches are robust, Blue Fin employs the following research and development cycle for constituent models:
Stage 1: Both internal ideas and information from the public domain are accumulated and ranked according to potential benefit. The highest ranked ideas are then selected for prototyping. Once prototyping shows potential opportunities for alpha generation and diversification, they are taken to a second stage.
Stage 2: Ideas are refined into models that have costs and execution assumptions applied which approximate our experience of real life. Since most of our models have some control parameters, their values have to be carefully selected. Blue Fin ensures that selection of parameters explicitly avoids 'curve-fitting', or 'over-fitting', historical data. Numerous in-sample and out-of-sample tests are used to establish the most statistically 'robust' parameter values. Stress testing with simulated risk events ensures models can cope with situations of extreme price movements and market dislocation.
Stage 3: Once control parameter values have been selected (determined from Stage 2), trading models are then submitted for paper-trading. This step allows the detection of any inconsistencies in the models. If inconsistencies are detected, models must go through Stage 2 again to accommodate any changes or adjustments.
After a successful paper-trading run is passed, models are submitted to the real-time trading environment for systematic trading.
Risk is carefully and systematically monitored from both an individual market and entire portfolio perspective. At the portfolio level, a predetermined risk budget is allocated to each market/system. Position size is restricted to preset computational limits and capital allocation to each market can be increased up to those limits or decreased with reference to the individual strategy’s longer-term performance in any particular market, considering both absolute return and volatility of return. Trade size may also be decreased on a discretionary basis prior to the release of major economic numbers and reports.
