Here at SBM, we have been very vocal during the last few months in stating our concerns about the disjoint beween financial markets and the real economy. With earnings rising only mildly, it is difficult to actually justify the kind of returns we have been observing for the S&P 500, and which ended the first half of the year accumulating a gain of 12.6%. With the Federal Reserve possibly reducing the amount of asset purchases during the third quarter, the “thin air” that has pushed equity prices higher may just vanish and at that time, only hard assets may be worth holding, with gold at the top of our preferences list. Aside from all the fundamentals pointing to a possible downturn during the next few quarters, what does the actual historical data say however?
It’s an interesting question, which we decided to address by looking at S&P 500 data for the past 60 years. We set up an experiment with the goal of trying to identify if a relationship existed between the first half and second half performances. Does the market tend to correct during the second half after a great first half? Or does it tend to rise even more?
In our view there are two main forces to consider. On the one hand, you would expect investors to take profits and so close some positions as the market rises. In this regard, the second half may very well underperform following a strong first half. On the other hand, households tend to wrongly time the market in aggregate and to enter it much later than institutional investors. They only enter the market after a period of good performance, and so they would probably assist with the theory of a continuation of a string second half performance.
Let’s set see what the data actually reveals…
Getting the data and building the datasets
We collected data for the S&P 500 from 1953 to 2012. Our dataset not only shows annual performance for each year, but it also details 1H and 2H performances. The main statistics show that on average the index rose 8.3% per year, with 1H returns having been slightly lower than 2H returns, 3.95% against 4.06%. The S&P 500 rose in 43 years, or 72% of the cases, and 1H performance was positive in 65% of the cases.
In order to implement our study, we need to split the data into different datasets. Our aim is to understand if there is a relation between 1H and 2H performances so we need to rank the data by 1H performance to try to identify a pattern. Does stronger 1H returns lead to strong 2H returns? Or does strong 1H returns lead to weak 2H returns as profit taking occurs?
After ranking the data by 1H performance, we split this into four datasets, each with 15 years of data. Dataset 1 shows the best overall 1H returns, dataset 2 shows the second best, and so on until dataset 4, which holds years with the worst 1H returns. The main stats for our four datasets are summarized below.
Dataset 1 shows a 1H average return of 18.1%, the highest of the four datasets, with a maximum of 38.8% and a minimum of 12.4%. All years in the dataset show a positive 1H return with 11 out of 15 continuing to show positive performance during the 2H. On average, performance for 2H was 6.7% in this dataset.
Dataset 2 has less than half the 1H performance of dataset 1. On average there is a rise of 8.15% but still all 15 years show positive 1H performance. On this dataset there are less negative 2H performances but the average 2H performance is 6.21%, a little less than in dataset 1.
The next dataset shows an average performance for 1H of 0.37%. In 9 out of 15 cases, the market rose during the 1H and subsequently rose 10 out of 15 in 2H to record an average 2H performance of 2.17%
Our last and worst performing dataset shows an average 1H performance of -4.96%. The market decreased in all 15 years during the 1H but then rose in 7 out of 15 cases to average a 2H return of 1.1%.
Interpretation of results
First of all, it is clear from the data that the higher 1H returns, the higher 2H returns tend to be. Past performance may lead to some profit taking but it also seems to encourage newcomers that will continue to drive the market higher. But, it is also important to note here that when 1H performance is high, 2H performance tends to be much lower. In fact, if we take a closer look at dataset 1, we identify 4 negative 2H returns and all those actually correspond to the five best 1H performances. It means that while on average 2H performance tends to follow the same direction as 1H performance, when 1H is extreme, the market usually corrects, performing negatively.
We reproduce here these dataset 1 results.
What this is telling us is that an investor should be more careful after a particularly good 1H performance as even though positive performance is likely to continue, the market may go sideways for longer periods and upside potential may be limited.
S&P 500 rose 12.6% in 1H 2013
With the S&P 500 recording a rise of 10% during the 1Q and an extra 2.4% during 2Q, the accumulated performance for 1H is 12.6%, which is inside our dataset 1, even if near the bottom of the table. This shows two important things. First, performance for the first part of 2013 has been very good but not extreme when compared with the past 60 years of data. If our study means anything, then the data seems to point to a high propensity for a good 2H performance albeit modestly lower than the 1H performance.
We decided to build an alternative dataset in which the number of years is still 15, data is ranked by 1H performance, but the average 1H returns are near our 12.6% seen during the 1H 2013. This alternate dataset is bounded by a 16.8% 1H return occurred in 1998 and an 8.9% return occurred in 1996. In all cases, 1H performance is positive. Interestingly, what this threw up was that in all cases, the 2H performance is also positive and the average 2H performance is 9.3%. This is a compelling signal we should not ignore.
In our view the reason behind these results is the fact this dataset shows strong enough performance to capture the attention of new investors in 2H but not extreme enough for many investors to take profits. This way 1H performance works as effective ‘bait’ for new entrants and thus explains why 2H returns are so strong.
While this would be general good news for what to expect in 2H 2013, we are a little cautious here. As we have seen above, extreme cases tend to revert into negative performance. Even though a 12.6% rise may not be an extreme return, we should take into account that S&P 500 has been on a bullish run since early 2009 and per our guide below, the likely end date will be spring next year. The 2 provide strong evidence that there is a little more gas in the engine but we are nearing spluttering speed and it may be time to look to sell calls on any further extension beyond the 200 day moving average.