KV1 works with systematic, or technical, trading algorithms – trading “rules” that can be expressed algorithmically and therefore monitored and signaled computationally. KV1 extends this systematic approach upwards through the management hierarchy. This includes developing systematic approaches to portfolio management – developing models to evaluate, optimise and deploy trading systems; to "hire and fire" the trading systems; to allocate capital; and to monitor and control risk exposures.

To use a mechanical analogy, we have a process that builds portfolio machines, which, in turn, build and control trading machines.

We view this aspect as one of the most significant proprietary advantages possessed by KV1, and our research focusses on extending these advantages.

Avenues Of Research

Research at KV1 is concentrated in four areas:

Trading model development: We are always looking for new or improved models.

Dynamic selection and rejection of trading models: Activating and deactivating models is a complex, automated process which always merits further research.

Portfolio and risk management algorithms: Automated portfolio and risk management is KV1's strength, but risk management is a rapidly evolving art and we are constantly studying it.

Growth and further diversification opportunities: With a methodology that is purely algorithmic, we believe our approach can be successfully applied to other markets and with other instruments.

Value of Simulations

Due to the completely systematic nature of the investment process, KV1 can use historical market data to simulate how our chosen models and systems would have performed at various prior times.

Well-designed simulations remain free of the invalidating effects of hindsight as there is never any discretionary activity. Unlike human managers trying to infer what they would have done in the past, the model's simulated decision-making is never tainted with "tomorrow's data".

Our confidence in these simulation results allows us to test new developments exhaustively and quantify their contributions with a high degree of confidence. All products of our research can be, and are, subjected to simulation testing.

The chart below illustrates how constant efforts have strengthened the "state of the art" at KV1. Note the marked improvement in the distribution of monthly returns for the same 5-year simulated investment.

The dashed line represents the distribution of simulated results from KV1's chosen model as of August 1999, and the solid line shows that same 5-year simulated investment managed by our chosen model as of December 2000

   

Applying the Results

By definition, the value of simulations is very limited and we would never presume that our real returns should match our simulated performance. However, we believe that a model that demonstrates a robust and stable improvement in simulated performance will most likely be a model that performs better in reality.

Consequently, we choose to apply the best of our developments to our investment model from time to time (typically two or three times a year).