Most investors avoid forecasting (timing) methods because of the theoretical and practical difficulties that are associated with the transition from strategic to tactical asset allocation. If a transition is decided, assigning the task to a registered adviser with experience in trading timing models may be a prudent move.
Strategic Asset Allocation (SAA)
An example of SAA is the fixed 60%-40% portfolio in stocks and bonds. The portfolio is regularly rebalanced to maintain the fixed allocation. The allocation changes if the risk profile changes. This strategy may appear similar to passive buy and hold but in reality it is much more sophisticated than that.
Tactical Asset Allocation (TAA)
In TAA, forecasting models are used to determine the timing of adjusting portfolio allocations. Absolute (time-series), relative or dual momentum are often used for this purpose. The adjustments can also be discretionary based on fundamental analysis.
Why is the transition from SAA to TAA so difficult?
In a recent article I listed three reasons why investors neglect momentum strategies. One of them is technical and has to do with the high probability that any portfolio mix that is suggested in books or articles is the result of data-mining. While this does not invalidate the historical results, it raises the probability that the portfolio may not perform as expected in the future because of data-mining bias.
In a nutshell, data-mining bias results from the dangerous practice of reusing historical data and a large universe of assets in backtests in an effort to identify a strategy with desired performance parameters. The few good results that are observed may be due to data snooping and selection bias. Accepting the good results amounts to p-hacking, which is a polite way of saying that this process is stupid.
Is there an alternative way for low or no bias data-mining? Possibly, there isn’t any. In general, data-mining bias cannot be measured. In addition, those that claim that they can reduce data-mining bias by applying restrictions, or even fancy tests, such as for example outperformance in rolling 10-year windows or Monte Carlo analysis, ignore the fact that when those become part of the research and used repeatedly they lose their significance due to data-snooping.
However, the biggest problem in the transition from SAA to TAA is that it fundamentally changes the nature of operation from passive investing to market timing. The impact of this change is largely underestimated by authors of popular books and academic researchers that have little or no experience with the operational side of trading and investing. Timing requires constant monitoring of the markets and discipline for following the strategies and their signals.
Some have even suggested that even individual investors must apply timing because of the potential of achieving superior risk-adjusted returns. They have also gone as far as arguing that the models are so simple that anyone can do it. Although it is true that the models used for timing are trivial and usually amount to comparing monthly prices separated by a lookback period, one should not expect a retiree to do that while on vacation. Timing models are for traders no matter how slow the parameters are. But even traders have problems occasionally in adhering to system rules. Trading is a complex business no matter how simple the strategies are. Only someone who has little experience with trading can suggest that anyone can become a successful trader.
A successful transition from SAA to TAA is a dubious process when the above issues are taken into account. However, if a transition is desired, assigning the task to a registered adviser with experience in trading timing models may be a prudent move.
If you found this article interesting, I invite you follow this blog via any of these methods below:
If you have any questions or comments, happy to connect on Twitter: @mikeharrisNY