In our opinion, a key reason our long-term track record has been repeatable over several market cycles is that we harness quantitative models to provide discipline to our decision-making, particularly in the area of tactical rebalancing. Quantitative models represent part of the “Science” methodology we use in our “Art + Science” approach to investing. However, our experience suggests it is important to be aware of both the strengths and weaknesses of using quantitative techniques when investing. As Vin Scully says, “Statistics are used much like a drunk uses a lamp post; for support, not illumination.” For example, the best-selling book When Genius Failed chronicles how over-confident Nobel Prize winning quantitative investors permanently impaired investors’ capital through the failure of the hedge fund Long Term Capital Management. Therefore, Balentine’s Investment Strategy Team is mindful of five criteria when deploying quantitative models.
- Objective: An effective model is objective in that it strips out personal beliefs, behavioral biases and emotions and gives an unbiased assessment of the market. Key inputs to our quantitative models are index level returns across multiple asset classes. At Balentine, we like to say that the models are here to make us “feel uncomfortable,” because at times, a model may signal a change that is contrary to popular opinion. However, this data-driven objectivity is an important aspect of the “Science” portion of our thinking. In other words, quantitative models can be effective if they simply produce insight into how investors behave, not the reasons why.
- Simple: Models should exhibit simple logic and easy formulae. Maintaining simplicity in the models (but not oversimplifying) makes them more durable and able to adjust to changes in assumptions and different market environments while still producing realistic results. Balentine’s quantitative models are based on our observation that markets and asset classes move in long-term cycles with extended amplitude. Our momentum-driven models take advantage of these market cycles by identifying turning points and providing simple buy/sell signals across asset classes for us to implement. A simple, slow and low-turnover model capturing a large mispricing is often more effective than a higher frequency, intricate model that requires a lot of leverage to profit from a small mispricing.
- Intuitive: Every user of a model should be able to understand, interpret and clearly communicate the results of the model. The model should not be so overly complex that it can only be used by its creator or that it would require others excess time in “breaking the code.” Having an intuitive model is integral to Balentine’s team approach to investing. There are no points for elegance!
- Functional: Functional models allow for scenario and sensitivity analysis. This flexibility is instrumental in testing potential pitfalls. Creating an effective model also means knowing its limitations and understanding when it will and will not work under specific assumptions. Balentine’s quantitative models are momentum-driven in nature and are prone to lag market peaks and troughs. We understand this limitation in the models and consider it in our decision making process. It is often more important to spend time knowing when to turn the quantitative mousetrap on or off than trying to build a better mousetrap.
- Scalable: The structure of the model should be adaptable so that a large number of units can be added with minimal rework. The model should be robust enough so that asset classes with similar risk and return characteristics could follow the same methodology and still produce reliable recommendations. Our model methodology is currently being applied across many asset classes where liquidity is not a constraint to an investment idea.
For more on Balentine’s Investment Team, “Art + Science” approach to investing or our process, please email us.