Each quarter, my Newsletter features my overall Model Portfolio percent allocations to stocks, bonds, and cash for Moderate Risk Investors. I also present allocations for Conservative and Aggressive Risk Investors.

Since Jan. 2005, the Moderate Risk allocations for stocks have varied between 37.5 and 67.5% of one's entire portfolio for stocks and 22.5 to 50% for bonds. Of course, these allocations should only be viewed as suggestions since every investor has his/her own reasoning behind their allocation.

In my Newsletter, the reason these allocations typically change is because I believe there are better and worse times to invest in each of the above three asset categories. As a broad example, when I recommended a 65% allocation to stocks back during the first half of 2011, I was quite confident that stocks would perform well when considered over the next several years. When I recommended a 45% allocation in mid-2008 and early 2009, I was not at all confident of good performance over a similar period. The same type of approach was used to when judging the potential attractiveness of bonds.

Typically, past Newsletters have given readers an idea of what goes into making these judgments. For example, several times I presented the opinion that new Fed bond purchase programs (or "quantitative easing") would be a plus for stocks. Such programs often caused me to raise my stock allocations. More recently, as stocks seemed to become overvalued, I lowered my stock allocations.

Especially in recent years, I have tried to avoid letting shorter-term considerations exert a significant influence on these allocations. I have also frequently applied my own proprietary research, first described in 2008 Newsletters, to help define when stocks, in general, could be described as either a Buy, a Sell, or a Hold based largely on valuation measures. Given the successes, thus far, in using this research-based approach, it has become an integral part in allocation judgments along with any and all economic data.

Of course, the question that must be raised in the mind of anyone who sees my quarterly allocations percentages is "How likely is it that anyone can come up with frequently changing allocations that will prove to actually be predictive of stock and bond performance several years down the road?" While many experts might lead you to be skeptical that any such allocation numbers could actually be useful as a guide to forecasting returns, I decided to once again put my overall stock and bond allocations percentages to the test. (For those interested, I conducted and published some earlier, although somewhat different, tests in the Apr. 2011 Newsletter)

Subsequent Stock Market Performance as Predicted by Overall Stock Allocation

Many, and perhaps most, investors try to maintain just one constant allocation to stocks. For many, that figure may be roughly 60% of your total investment portfolio, with the remainder in bonds, and/or cash. Obviously, a constant allocation percentage says nothing about how well you think stocks will do other than to suggest that you have at least some confidence that stocks should do OK in the long run, otherwise, why would you invest that percentage in stocks at all.

But a changing allocation to stocks implies a changing outlook as well; a higher allocation should mean an expectation of better stock market returns ahead, and vice versa.

So how well have my changing stock percentage allocations done in predicting changing subsequent returns of the S&P 500 index? Since my allocations are designed to be a useful portfolio strategy tool over multiyear periods but not particularly likely to be on target for predicting short-term stock movements, we chose 3 years out as the period to measure how stocks had done over that entire period since the recommendation.

To better visualize the outcome of the above important question, we present the results in two tables. The first table shows all quarters going back to Jan. 2005, or 10 years, in which we, on the date shown, recommended a relatively high allocation to stocks. This was arbitrarily defined as 55% or higher for Moderate Risk Investors. The second table shows all quarters since 2005 in which we, on the date shown, recommended a relatively low allocation to stocks, which we defined as 52.5% or below.

Here are the results, followed by further analysis. Note that since we chose 3 years after an allocation was made to analyze results, the latest quarter to be included was Jan. 2012; for April 2012 and beyond, 3 years has not elapsed yet.

As can be seen, 3 years after relatively high allocation recommendations were made, most of the annualized returns on the S&P 500 index turned out to be excellent, ranging from over 20% to a little less than 9% per year. These recommended allocations, therefore, were highly successful in suggesting good future performance. Quarters highlighted in red show those where a high stock allocation did not produce a high 3 year annualized return, that is, at least 8% or above.

In numerical terms, the success rate of the high allocation recommendations in Table 1 was 77%, or 10 out of 13 comparisons.

In this table, you can see that 3 years after relatively low allocation recommendations were made, the majority of the annualized returns on the S&P 500 index turned out to be poor, ranging from -13% to a meager +3.3% per year.

Quarters highlighted in red show those where a low stock allocation did not produce a low 3 year annualized return. In fact, these quarters subsequent performance showed just the opposite, a high 3 year annualized return.

In numerical terms, the success rate of the low allocation recommendations in Table 2 was 75%, or 12 out of 16 comparisons.

Bond Allocations

Let's now look at a similar analysis of our quarterly bond allocations.

We arbitrarily defined a "high" allocation to bonds as 35% or higher for Moderate Risk Investors. A "low" allocation was defined as 32.5% or lower.

We used the performance of the benchmark iShares Core US Aggregate Bond (commonly referred to as AGG) as the measure of how well bonds in general performed.

When we compared the subsequent performance of the high allocation recommendations to the low ones over a two year period, we again found that the high allocation recommendation results were better than the low allocation results. (Note: the same outcome also held over 3 years but were stronger when a two year period was used.)

As can be seen, two years after relatively high allocation recommendations to bonds were made, most of the returns on AGG turned out to be good, ranging from an average of 9.4% per year for the period to 6.2% per year. These recommended allocations, therefore, were highly successful in suggesting good future performance. However, quarters highlighted in red show those where a high bond allocation did not produce a high 2 year annualized return, that is, of at least 6%.

In numerical terms, the success rate of the high allocation recommendations in Table 3 was 73%, or 8 out of 11 comparisons.

Table 4 shows that two years after relatively low allocation recommendations to bonds were made, most of the returns on AGG turned out to be decidedly mediocre, ranging from an average of 1.1% per year for the period to 5.8% per year. These low recommended allocations, therefore, were highly successful in suggesting less than optimum future performance. However, quarters highlighted in red show those where a low bond allocation did not subsequently lead to a low average yearly performance. Instead, these allocations produced relatively good performance.

In numerical terms, the success rate of the low allocation recommendations in Table 4 was 73%, or 16 out of 22 comparisons.

Further Analysis of These Results

These observations will help to put the data in the first two tables showing stock data in perspective.

• We recommended relatively high allocations to stocks early in 2005, in the latter half of 2007, and in the years following the end of the 2007-2009 bear market. Of course, as we began raising our allocations considerably beginning in 2010, no one could know for sure that a continued multi-year bull market lay ahead. But the factors that we regarded as important suggested higher allocations would enhance returns.
• We recommended relatively low allocations to stocks in the years leading up to the 2007 bear market and for its duration. As above, no one knew for sure in those years that a bear market was coming and when it came, when it would end.
• The average degree of difference between the subsequent returns of our higher and lower allocation quarters was substantial - nearly a 9.5% better annualized return in favor of the former (11.3 vs. 1.9%).
• However, we tended to make the wrong allocation judgment ahead of "turning points": When the market was about to go from bull to bear (2007), we were too positive; when it was about to go back into bull mode (early 2009) and a little beyond, we were too negative.
• Absolute percent level of allocation to stocks was not the key; what was key was a relatively high allocation vs. a relatively low allocation, as defined above. In other words, when we had a strong sense that stocks would do well, as compared to at other times, they usually did; when we had a weak sense that stocks would do well, as compared to at other times, they usually didn't.

Regarding our allocations to bonds:

• Our relatively low allocations to bonds tended to precede the 2007-09 bear market for stocks and its early months as well as from the start of 2011 and beyond; our relatively high allocations to bonds occurred during the intervening period.
• The degree of difference between the subsequent returns of lower and higher allocation quarters was considerable: there were more than 50% better average returns in the 2 years after our high allocations vs. those with low allocations (6.7 vs. 4.3%). The difference in average returns measured 3 years after the allocation were made dropped to 14% (5.7 high allocations vs. 5.0% low allocations). This suggests that our bond allocation recommendations successfully targeted performance in the bond market for a shorter period than for stocks.
• Generally speaking, we correctly anticipated lower bond returns toward the early part of the entire 10 year period, higher returns during the recession and its aftermath, and lower returns again starting around early 2011 up through the present.
• As with stocks, absolute level of allocation to bonds was not the key; what was key was our relatively high sentiment toward bonds vs. our relatively low sentiment.

What This Data Suggests for Future Returns

Of course, in investing, past data can never ensure or guarantee that future results will be similar.

But what I have demonstrated above is that you should not assume that pre-selecting a fixed asset allocation that seems appropriate for your risk tolerance and keeping your allocations at or near this allocation is a rock solid, guiding principle for managing your investments. Yet such a prescription typically forms the basis of how most investors indeed manage their investments from year to year.

A cursory look at the above tables shows the obvious - that investment returns, particularly for stocks, vary greatly from quarter to quarter and from year to year. Most of us have been told, though, by investment experts that it is extremely hard, if not impossible, to accurately forecast in advance what these changes will be.

But the above tables seem to demonstrate that even two or three years in advance, it is possible to use well-chosen information to get a good sense of what kinds of returns, generally at or above par, or below par, might be expected from stocks and bonds. The problem most people run into, in my opinion at least, is that they mainly focus on trying to predict how stocks or bonds will do for much shorter periods. It is mainly such predictions that have a much reduced chance of success.

Unfortunately, as you might have anticipated, I can't provide an all-in-one formula for attempting to make accurate predictions for two or three years down the road. But let me just reiterate that it is possible to successfully make such predictions, especially if one gives up on a) looking too much at short-term events that likely won't matter much in a year or two and focusing instead of matters that likely will matter; and b) expecting the predictions to always be right; if you are unwilling to proceed without near 100% certainly, you will miss out on many likely outcomes that will happen the majority of the time, but not 100% of the time.

In our most recent Model Portfolios, we have dropped back our allocations to both stocks and bonds, but especially stocks, from where they were several years ago. Take a look at the most recent allocations for Moderate Risk Investors:

These stock allocations suggest that if the predictive patterns of our earlier allocations hold true, upcoming 3 year returns for stocks may soon start to fall back considerably for periods beginning Jan. 2014. And if the results turn out matching pretty closely those reported in Table 2, it is possible that within the next few years, stock returns between 2014 and 2017 may wind up averaging in the low single digits when annualized over three years. This means that since 2014 was a pretty good year for stocks, especially the S&P 500, either 2015 and 2016 or both, will likely show considerably smaller, or even a year (or possibly two) of negative returns.

Two year average returns for bonds have already dropped off considerably since early 2011, as shown in Table 4, and our recent below average allocations suggest that the same will likely continue for the next few years.

Be assured the purpose of all this is not to "scare" investors; things over the next few years may not turn out as this data and by implication, my current allocations, might suggest. Rather, the data is presented to offer an alternative to the notion of always thinking that a fixed allocation to stocks and bonds, quarter after quarter, year after year, is the best way to manage your investments.