Can any one indicator predict, with a generally high degree of accuracy, whether stocks are going to do well or poorly over the next several years? Ditto for bonds? Most people would probably think not. Therefore, if you as an investor were to come across two such indicators, you would have to assess whether their successful results were indeed valid and likely to foretell future results, or on the other hand, perhaps be attributable merely to luck or chance, and essentially non-repeatable. Part of the obvious difficulty in finding such indicators is that both overall stock and bond prices cannot be attributable or influenced a single factor alone but a great number of factors. These factors interact, sometimes as might be expected, but at other times, differently, creating outcomes that few would have predicted. Some of these factors entail how certain aspects of the economy are functioning, but others would defy a straightforward relationship with the investing universe. Rather, the latter would likely require an estimate of hard to anticipate “psychological” factors that cause investors to either “love” or “hate” stocks and bonds, often over significant periods of time. At least, this is the conclusion I, and others, such as 2013 Economics Nobel Prize winner Robert Shiller have come to. In my case, this comes after approximately three decades of studying the movements of stock and bond prices and trying to understand what influences them. Given this, “single-dimensional” predictors of stock and bond prices are unlikely to be consistently successful, whether they be interest rates, gross domestic product, P/E ratios, consumer sentiment, or you name it. Instead, I’d rather study as many of the subfactors that potentially affect investment performance and come up with my own approximate assessment of potential future performance, even as highly subjective as that might appear. Narrowing this down further to a “composite” measure which might serve to help me predict future stock as well as bond performance might appear to be a nearly impossible task. But I will now present considerable evidence to the contrary. Specifically, the two measures I have come up with, one for stocks and one for bonds, have now been shown to have an impressive record going back to 2005. Note: Similar results were reported by me a little more than a year ago and even five years ago , and since then, new data continue to support the same findings. Of course, there can be no guarantees that these predictions will continue to prove accurate looking forward. Successful Stock and Bond Index Predictors Might Not Come From Where You Would Expect What I am referring to as “predictors” were not initially meant to be used to predict how the overall stock or bond markets would do. Rather, they were designed as recommendations, re-evaluated each calendar quarter, as to how much of a “moderate risk” investor’s portfolio should be allocated to stocks and how much to bonds in my Newsletter’s model portfolios. But, in a real sense, a relatively high allocation to stocks vs. bonds (or cash) should generally equate to an expectation that stocks will do relatively well, and the same for bonds. Likewise, relatively low allocations to either should generally suggest the opposite. Regarding a high or low allocation to stocks (or bonds), investors must ask what is the time frame involved. For example, in recommending a 100% allocation to stocks, does this mean one predicts stocks are going to be the best investment over the next few months, years, or over a lifetime? Each assumption has a totally different implication for judging the eventual success of the prediction. Again supposing an advisor recommends a 100% allocation to stocks. Question: Does this imply that he is quite bullish on the future prospects for stocks? Answer: Likely perhaps, but not necessarily. While this might be a logical conclusion, it only necessarily implies that he thinks stocks prospects are relatively better than the remaining alternatives, but not necessarily “high” in an absolute sense. For example: While stocks prospects might not be particularly bright, a 100% allocation would still make sense if one estimated that bond and/or cash returns were going to be even less. Thus, even expecting a 2% return in stocks should be preferable than, say, a negative return expected in bonds, or a near zero return in cash. If one had no expectation as to the future performance of either stocks or bonds, it would not appear to make any sense as a strategy to continually, or even on occasion, raise or lower one’s allocations. Rather, one would just select a single percent allocation that he was comfortable with given his risk tolerance, current financial position, age, years to retirement, etc. He would then only change that allocation when one or more of those variables changed, not because he thought it was a particularly appealing period ahead to hold more stocks, for example. Unlike aforementioned objective data such as current level of interest rates, etc., one’s percent allocations to stocks vs. bonds would seem to be totally subjective, and therefore, hardly useful as predictors. But in spite of the limitations, it does appear to make sense to consider strategic changes to allocations to either stocks or bonds as a type of numerically-based composite summary indicative of my level of confidence in upcoming future performance. I have chosen to define this measured future performance after three years for stocks and two years for bonds. How Do I Arrive at These Predictions? My recommended allocations (i.e. predictions) are based on almost all the information I can lay my hands on. This includes both economic data and, as noted above, whatever “psychological” inferences I can draw from observing how investors have behaved in the past, and therefore, are likely to behave like in the future. What are some examples of the latter? It has long been said that investors act in terms of greed and fear. This means that so long as the markets are going well, there is the tendency for investors to continue to invest accordingly, further pushing up prices. Obviously, the opposite is true as well: Fear, once aroused, can become the overriding emotion and keep on spreading to other investors. But “too much” of a good thing can lead investors to take profits. But “too much” of a reversal can bring out bargain hunters. And importantly, investors should be on the lookout for data that can cause what appears to be an ongoing trend to reverse. How These Past Predictions Have Fared Here are the data showing the effectiveness of using my overall allocation recommendations to stocks vs. bonds as predictors of subsequent stock and bond index performance. The data encompass all full three year periods for stocks and two year periods for bonds beginning back in Jan. 2005 and progressing to include the start of every subsequent calendar quarter. Using Stock Allocations The data is separated into two tables, making it easy to see two sets of outcomes depending upon high vs. low allocations. In Table 1 are shown all quarters in which, beginning the month shown, I recommended a relatively “high” allocation to stocks and the actual subsequent return on the S&P 500 index. A high allocation was defined as 55% or higher of an entire portfolio for moderate risk investors. Table 1: Annualized Returns for the S&P 500 Index 3 Yrs. After “High” Stock Allocation Recommendations Quarter Beginning Allocation to Stocks Annualized Return Quarter Beginning Allocation to Stocks Annualized Return Jan ’13 67.5% 15.1% Oct ’10 62.5 16.3 Oct ’12 67.5 12.4 Jul ’10 60 18.5 Jul ’12 67.5 17.3 Apr ’10 60 12.7 Apr ’12 67.5 16.1 Jan ’10 57.5 10.9 Jan ’12 62.5 20.4 Oct ’07** 55 -7.2 Oct ’11 60 23.0 Jul ’07** 55 -9.8 Jul ’11 62.5 16.6 Apr ’05** 55 +5.8 Apr ’11 65 14.7 Jan ’05 55 8.6 Jan ’11 65 16.2 A relatively high allocation to stocks made at the beginning of each quarter was predictive of a corresponding relatively high return on stocks as measured three years later in the great majority of cases (that is, 14 out of 17). The average annual three year return for all these high allocation predictions was 12.2%. Quarters marked ** show those three where a high stock allocation did not produce a relatively high 3 year annualized return; these returns were each below 6%. In comparison, Table 2 shows all quarters during the same span in which, on the date shown, I recommended a relatively “low” allocation to stocks along with the actual subsequent return. A low allocation was defined as 52.5% or lower for moderate risk investors. Table 2: Annualized Returns for the S&P 500 Index 3 Yrs. After “Low” Stock Allocation Recommendations Quarter Beginning Allocation to Stocks Annualized Return Quarter Beginning Allocation to Stocks Annualized Return Oct ’09** 50% 13.2% Apr ’07 52.5 -4.2 Jul ’09** 50 16.5 Jan ’07 52.5 -5.6 Apr ’09** 45 23.4 Oct ’06 52.5 -5.4 Jan ’09** 37.5 14.2 Jul ’06 50 -8.2 Oct ’08 42.5 1.2 Apr ’06 52.5 -13.0 Jul ’08 45 3.3 Jan ’06 52.5 -8.4 Apr ’08 47.5 2.4 Oct ’05 52.5 0.2 Jan ’08 52.5 -2.9 Jul ’05 52.5 4.4 A relatively low allocation to stocks was predictive of a corresponding relatively low return on stocks as three year subsequent stock index returns were noticeably lower than those shown in Table 1 in the great majority of cases (12 out of 16). The average annual three year return for the low allocation recommendations was a mere 1.9%. Notable exceptions are shown with ** for those four quarters where a low stock allocation did not produce a low three year annualized return, and in fact, where returns were quite positive. These exceptions, as well as those in Table 1, will be discussed in more detail shortly. Bottom line : The subsequent three year annualized returns in stocks originating from high allocation quarters were greater than 6 times more than those originating from low allocation quarters ( 12.2 vs. 1.9% ) in spite of the relatively small percentage of missed predictions. Using Bond Allocations Below are the data when the same type of analyses is applied to my bond allocations. For bonds, a “relatively” high allocation was defined as 35% or higher of an entire portfolio for moderate risk investors, while a “relatively” low allocation was defined as 32.5% or lower. If relatively high allocations to bonds were predictive of relatively high returns on bonds, one would expect to see that reflected in actual performance data shown in Table 3, likewise for relatively low allocations shown in Table 4 which should be associated with relatively low future returns. Bond returns were those reported for the standard bond benchmark, the Barclays Aggregate Bond index, by averaging the returns from year one and year two after the allocation recommendations. Table 3: Average Yearly Return for Bonds 2 Yrs. After “High” Bond Allocation Recommendations Quarter Beginning Allocation to Bonds Avr. Yearly Return Quarter Beginning Allocation to Bonds Avr. Yearly Return Oct ’10** 35% 5.3% Apr ’09 47.5 6.4 Jul ’10** 35 5.7 Jan ’09 50 6.2 Apr ’10 35 6.4 Oct ’08 40 9.4 Jan ’10 37.5 7.2 Jul ’08 35 7.8 Oct ’09 45 6.8 Apr ’08** 35 5.4 Jul ’09 45 6.7 The average yearly return for the high allocation recommendations in Table 3 was 6.7% . In 8 out of 11 cases, the return was at least 6%. Those 3 instances in which the return was less than 6% are marked with **. Table 4: Average Yearly Return for Bonds 2 Yrs. After “Low” Bond Allocation Recommendations Quarter Beginning Allocation to Bonds Avr. Yearly Return Quarter Beginning Allocation to Bonds Avr. Yearly Return Jan ’14 25% 3.3% Jan ’08 30 5.6 Oct ’13 25 3.5 Oct ’07** 30 7.2 Jul ’13 25 3.2 Jul ’07** 22.5 6.6 Apr ’13 27.5 2.8 Apr ’07 25 5.4 Jan ’13 27.5 2.0 Jan ’07** 27.5 6.1 Oct ’12 27.5 1.2 Oct ’06 27.5 4.4 Jul ’12 27.5 1.9 Jul ’06** 27.5 6.6 Apr ’12 25 1.9 Apr ’06** 27.5 7.2 Jan ’12 32.5 1.1 Jan ’06 30 5.7 Oct ’11 32.5 1.8 Oct ’05 27.5 4.4 Jul ’11 30 3.4 Jul ’05 30 2.7 Apr ’11 30 5.8 Apr ’05 25 4.5 Jan ’11** 30 6.0 Jan ’05 25 3.4 The average yearly return for these low allocation recommendations in Table 4 was 4.1% . In contrast to Table 3, in 20 out of 26 cases, the return was less than 6%. The exceptions are marked with **. Bottom line : Bond allocations formulated two years prior to actual bond market returns were available typically were able to predict how high or low bond market returns would be. In fact, high allocations were followed by bond market returns which were approximately 63% higher than when low allocations were recommended. Highly Accurate Predictions For both my stock and bond allocations going back to 2005, separating recommendations into those that were relatively high vs. low allocation would have been able to help investors capture high returns and avoid low ones. The results helped predict stock and bond market performance during both strong markets and weak ones over more than a 10 year period. These findings, along with those prior articles mentioned above, should be regarded as surprising, given what is regarded as the extreme difficulty of predicting stock and bond indexes using any number of other more objective measures. While the data show some exceptions to accurate prediction, even when including these exceptions, the average outperformance of the high vs. low allocations has been large enough to suggest that my allocations are, for the most part, anticipating correctly future strength and weakness within broad market indexes. Most and Least Successful Stock Predictions For stocks, the predictions for high returns were the most accurate from about 1 year after the beginning of the bull market which started in March 2009, a period encompassing 13 consecutive quarters. In the case of predicting low stock returns , they were most accurate for 8 consecutive quarters during the midst of the 2003-2007 bull market as they correctly anticipated that stock prices might begin to underperform for the next several years. They were also highly accurate in predicting low returns for 4 consecutive quarters after the 2007-2009 bear market had begun. The predictions for high stock returns were inaccurate only for a single quarter in the early part of 2005, and at the start of the two quarters preceding the start of the 2007-2009 bear market. The predictions for low stock returns were inaccurate only prior to the beginning of the 2009 bull market and for 3 subsequent quarters. Most and Least Successful Bond Predictions For bonds, the predictions for high returns were most accurate for the 8 consecutive quarters starting in the midst of the 2007-2009 recession and continuing for about a year beyond. They were most accurate in predicting low bond returns during the 12 consecutive quarters after the post 2007-2009 recession and economic expansion was well underway. They were similarly accurate during the ongoing economic expansion in 2005 for 5 consecutive quarters. Predictions of high bond returns were inaccurate during the 2 quarters US economy moved well past the 2007-2009 recession. There were several irregular periods of inaccuracy in predicting low bond returns during the 1 1/2 year period which preceded the 2007-2008 financial crisis. In summary, while my predictions were accurate the great majority of the time, they had the most trouble predicting subsequent returns when the economy “turned” in some significant way, such as when an ongoing bull trend turned to bear, or vice versa. But these inaccurate predictions were usually relatively brief as compared to the times when the predictions were accurate. What This Suggests for Future Stock and Bond Market Returns The above Tables do not show my most recent allocations to stocks and bonds. This is because not enough time has elapsed yet since Apr. 2013 for stocks and Apr. 2014 for bonds to see whether the longer term predictions will prove accurate. Table 5 shows these allocations; instead of showing three (stocks) and two year (bonds) returns, returns for just one year are shown. Table 5. Recent Quarterly Asset Allocations for Stocks and Bonds and Returns After One Year Quarter Beginning Allocation to Stocks S&P 500 Return 1 Yr. Later Quarter Beginning Allocation to Bonds Bond Index Return 1 Yr. Later Jan ’16 52.5% NA Jan ’16 35% NA Oct ’15 50 NA Oct ’15 35 NA Jul ’15 50 NA Jul ’15 25 NA Apr ’15 50 NA Apr ’15 25 NA Jan ’15 50 1.4 Jan ’15 25 0.5 Oct ’14 50 -0.6 Oct ’14 25 2.9 Jul ’14 50 7.4 Jul ’14 25 1.9 Apr ’14** 50 12.7 Apr ’14 27.5 5.7 Jan ’14** 52.5 13.7 Oct ’13 55 19.7 Jul ’13 65 24.6 Apr ’13 67.5 21.9 Note: NA signifies data not yet available. Although we cannot yet see if these predictions will be in line with the data in Tables 1 through 4, highly similar trends are already starting to emerge. For stocks, when allocations were high (55% and above), the average S&P 500 index return one year later was 22.1%; when allocations were low (52.5% and below), the average return one year later was 6.9%. The two instances out of 8 in which the predictions proved inaccurate are marked with **. For bonds, in the 4 instances where data currently exists, when allocations were low (32.5% and below), the average return for the Barclays Aggregate Bond index was 2.8%. Referring back to Table 4, you can see that bond returns have been consistently low for each quarterly two year period since April, 2011. Additionally, current three year returns Additionally, current three year returns on stocks still suggest that having a high allocation to stocks during the early months of 2013 would have been helpful to investors. However, since S&P 500 stocks have not shown any gains over the last a year and a half, it may be that 3 year gains will not be strong as we move forward. It appears that the trend for stocks indeed turned at that time and my allocations, as previously, had some difficulty at first in correctly predicting that turn. In Jan. 2014, my stock allocations dropped to 52.5% and have remained at that level or below ever since. Since stock returns, although initially good, have turned marginal since that date, it again appears that a relatively low allocation to stocks, although somewhat early, may turn out to have been a helpful move. Since Apr. 2014, my allocations to bonds have mostly been low. However, starting in Oct. 15, they turned high. As noted, over the entire period, bond returns have also been relatively low. But it should be pointed out that I raised my allocation to bonds not because I expected high returns on an absolute basis but only relative to cash. If the successful prediction demonstrated in Tables 1 through 4 shows these allocations are tapping into the potential performance of stocks and bonds, it would appear that we may be in for a continued period of low returns for both. Especially for investors in S&P 500 index funds, such as (MUTF: VFINX ) and (NYSEARCA: VOO ), the bond benchmark (NYSEARCA: AGG ), but also all other funds/ETFs that benchmark these two indexes, my findings may be of particular value. Disclosure: I am/we are long VFINX. I wrote this article myself, and it expresses my own opinions. I am not receiving compensation for it. I have no business relationship with any company whose stock is mentioned in this article.