There is a lot of literature on this subject and many strongly held opinions, mostly against it. But any “opinion” should always be backed up by facts. Otherwise you might as well be selling snake oil or be a high priest. What do people who rail against market timing really mean? They mean that there is no method that consistently works AND that would benefit someone who invests a fixed amount of money in the Stock Market on a monthly or quarterly basis, an approach known as Dollar Cost Averaging. Not that they do not know of one, but that it does not and cannot (and never will?) exist or that no approach can outperform it in the long term.

This presupposes that the Stock Market will inexorably rise as a long-term trend, something that Japan and most Emerging Market economies have disproved in spades. Japan’s Nikkei is currently at the same level as 20 years ago (that is a whole generation, folks!), while the South Korean KOSPI (just in case you think I am picking a shrinking population narrative), & hardly an example of a slow growth economy, has not moved even 1% net in a decade. It also states that there is no simple or reliable way to avoid the inevitable downward gyrations of the market.

For a moment, let us assume that they are right (but there is a little room for doubt, right?) and that the US Stock Market will always rise over time, even relatively short periods of time like a single market cycle.  Is it true that there really is no method that will beat simple Dollar Cost Averaging over the long term, or even a relatively short period of time like a decade? Especially a decade that has been in an almost relentless bull market?

Let us take the example of an Investor who invested $1,000 in SPY every month since January 2007 on the market open of every month, till end December 2018. 12 full years.  A simple Market on Open order.  What could be more Dollar Cost Averaging than that?  I only pick $1,000 because it is a nice round number, you could pick any number.  Table 1 contains his return profile.

A very nice return. He would have accumulated 943 units of SPY, have no cash balance and his investment of $144,000 (12 years * 12 months @ $1,000/month) would be worth a total of  $235,759 at Year End 2018. Very nice.

Now let us assume he postpones his monthly purchase when the Daily Model is in Risk Off and stashes his cash and then invests it whenever the Daily Model goes back to Risk On from Risk Off.  What could be simpler? Do nothing, stash cash if in Risk Off. Wait. Invest available cash on new Risk On signal. If in Risk On, buy on open of the 1st day of the new month, as usual. End of thinking process.  Table 2 contains his return profile.

He would have accumulated 979 units of SPY, have a total of $2,000 in cash and his investment of $144,000 would have grown to  $246,726. Last time I checked, $246k is more than $235k. I have also included a difference chart in gray, to show that the variance of return keeps on rising with time, so it is not due to the period from 2008-09.

Now, some bright spark is going to say: “Yes, but you include 2008-2009, one of the worst possible periods to be investing!”

To which my answer (and I am going to have fun with this idiot) is:

  1. When do you think a model is most useful? Precisely when everyone is losing their heads and senses in a dramatic downturn. Are you really telling me that you would have continued investing every month in 2008 and in the first Quarter of 2009? Admit it, you were curled up in the fetal position in the broom cupboard or were crying in your Mummy’s lap while the repo men were outside asking for your house & car keys. Far from continuing to invest, you were furiously ringing your broker to liquidate everything right at the very bottom. That 666 print in SPX was YOU. Selling. Forgotten much?
  2. Quite seriously now: the killer of any investment methodology is volatility! Hardly anyone will have the courage to persevere with it during an event like 2008. If you think something like it will not happen to you in the future…then you are quite possibly beyond help. But if you are a sensible investor, a warning like the ones that come from the Risk models is potentially invaluable.
  3. The model is not even designed to tell you when not to invest in equities. The model is designed to tell you when the relative performance of bonds or equities will be better in one asset class relative to the other and warn you when some danger is lurking, so you can minimize your equity positions and maximize your fixed income allocations. That it can STILL beat a simple Dollar Cost Averaging strategy, without even having the benefit of the returns of fixed income, is testament to its long term validity. With the fixed income returns it beats Dollar Cost Averaging so hands down, it is not even a challenge.

Asking any method to improve on DCA 100% of the time is an impossible bar and no one serious will claim to have such a formula. Surely, that is not the point nor should it be the aim. The aim must be to have a tool which shifts the odds in your favor. Substantially and regularly.  And by this, I mean a proven method that will save you money with regularity and do the least amount of harm to your investment methodology when it does not work.

To my way of thinking, for a model to be truly effective, it has to minimize potential harm. As doctors say: “First do no harm.”  Can this model “do no harm”? But still provide a benefit? Here is real test: what would happen if you continued your Dollar Cost Average investing strategy but instead of buying at “market on open” on the first day on the new month, you shifted to placing a bid 1% below the close whenever the model was in Risk Off only? And if not filled at 1% lower, you bought at “market on close” on the last day of the month instead? What could be more innocuous? Your maximum risk is only the variance in price over a 30 day period.  Below is your profile:

You would have 934 units of SPY instead of 933. The equity curves overlap each other entirely. The DCA curve is inside the thickness of the adjusted performance. Again, I include a “difference” chart in gray to show that it does not depend on the 2008 period, but that the difference, although small, keeps on mounting with time.

Hence: no harm whatsoever, with a reduction in Volatility of returns. A very small benefit some will say, but you would have been filled at -1% a total of 59 times out of 70, or 84% of the time, as opposed to 47% of the time if you adopted that strategy randomly. There has to be something to the market timing ability of the model. Statistically this is a very significant variance.  How else can you explain it but that when in Risk Off, there really is a much increased level of downside volatility in equities? Is that not the very essence of Market Timing?  Recognizing those periods? Imagine how much use you can make of them in your investments.

What if you asked for a -2% drawdown on a monthly basis? The result would be identical in term of SPY units accumulated but with a 67% fill rate as opposed to 21% on a random basis. Again, statistically very significant! A large decrease in Volatility of returns, with no harm whatsoever to the performance of the investment strategy. QED.

Therefore, next time someone says that market timing does not work…point them my way and let me argue with facts against prejudices.