|
Aaron Trading constructs and rigorously tests its futures trading models utilizing out-of-sample walk forward financial engineering principles utilizing back-adjusted continuous data. These techniques are utilized to avoid the fatal dangers of curve fitting and over optimization. Typically, these two reasons are responsible for the failure of futures trading models when applied to actual trading. Additionally, we also apply additional statistical measures in an attempt to design very stable and robust financial systems.
Aaron Trading applies these techniques at the core foundation of its futures trading model engineering because:
If given an abundance of optimization and curve fitting, an individual can apply almost any trading strategy or model to a given financial entity and “force” it to be profitable. However, utilizing this approach, will typically create a trading model that trades the past well, but is subject to a heighten chance of failure in the future.
The out-of-sample walk forward approach represents a more “real world” trading simulation of the way a model is applied in actual trading. Furthermore, it also answers the following important questions:
- What impact will changes in trend and volatility levy on future performance?
- Is a trading model isolating random events or something that has statistical merit?
- Will a trading system continue to generate profits after optimization?
Answering the above questions provides the following benefits:
- It provides a more robust assessment of the rate of post-optimization profit and risk. Walk forward approaches to model construction produce a statistical profile of multiple in sample and out-of-sample optimizations periods. Furthermore, because it is based on a significantly larger sample, as opposed to testing a single period of data, it demonstrates considerably greater statistical validity. Additionally, it also allows a precise comparison of measurements of profitability between rates of out-of-sample vs. in-sample trading.
- It detects if a model has been curve fitted or over optimized. A system that demonstrates profitably over a large number of walk-forward tests is most likely to be successful within future applications because of its greater statistical validity.
- It verifies the forward-trading ability of the system, i.e. how versatile is the model to to market fluctuations and volatility.
|
