Faber's buy-sell rules are ultra-simple: buy when the month-end price of the asset is greater than the 10-month moving average and sell when it is less, keeping the money in cash until the signal changes. That would have kept an investor in the five asset classes he looked at invested about 70% of the time and generated less than one portfolio turnover per year (i.e. minimal effort and low transaction costs required). It's a system akin to Larry Macdonald's One-Minute Portfolio for Canadian RRSP investors, which also escaped 2008 relatively unscathed with only a 10% decline according to his Dec. 18th update.
- Will the future be like the past and will the system continue to work? Fitting a system to past data can be made to work - it is a statistical version of "hindsight is 100% accurate". Faber's worked in 2008, after he had developed the system in years prior, but what about the future? One never knows for sure.
- In strong upmarkets such as the 1990s (Faber is upfront about this in the paper - see Appendix D on pages 39-40 for the S&P500 vs the timing system stats decade by decade), the system underperforms, and can do so by a lot. Could you stand being left behind over the two decades of the 1980s - by 2% annually - and the 90s - by 5% a year? How much return should one sacrifice for disaster crash avoidance? After 2008, many "deferred-retirement" folks might regretfully say Faber's system is worth it. Just because we've had one shock in 2008 and things seem to be on the rebound for now, that doesn't mean we couldn't have another big drop soon (e.g. suppose the swine flu gets deadly in the fall, then as they say, "you ain't seen nothin' yet" for market turmoil).