Hi Guys,
Along with a long list of other famous quotes, John Maynard Keynes is usually credited with saying “ When the facts change, I change my mind. What do you do, Sir?”
Well the equity market facts are changing given that January 2016 was a down month. Consequently, the expectations for a positive annual return should be revised to reflect this additional data. The obvious question becomes “by how much”?
Statistically, when new information is added to existing info to update a projection, the procedure is called a Bayesian analysis. Simply put, an original baseline projection is multiplied by the additional (new) data to provide the posterior odds. The method is attributed to Thomas Bayes who did the math work in the mid-eighteenth century.
Historical equity market data provides a departure point. I did a quick and dirty analysis to make an estimate using the S&P 500 as a benchmark. I went back to 1979 data because of easy access and divided that data into a 2 by 2 matrix. Entries were made into 4 boxes: both plus and minus January returns, and, plus and minus S&P 500 annual returns.
In rough numbers, the baseline anticipated S&P 500 annual return is historically positive about 70% of the time. How does that change if the January return is positive? If it is negative?
For the timeframe that I explored, January returns were positive 23 times and that translated into a positive annual S&P 500 return 20 times. That equates to an 87% likelihood. Not bad since the supplemental January data enhances the probability of a rewarding annual return from 70% to the higher value. But that’s not what happened.
For that same timeframe, January returns were negative 14 times. In 9 of those instances, the year end returns were positive. Those outcomes represent 64% of the examination period for a negative January. From a Bayesian perspective, these data suggest that the likelihood of a positive annual S&P 500 return is reduced somewhat from the baseline data set.
That’s not so good, but it is not a disastrous probability downgrade. Given current market uncertainties and the limited data set, it could be assessed as a marginal downshift.
That’s exactly my interpretation. I am slightly decreasing the equity portion of my portfolio this year, but that decision is more dictated by my advanced age (81) then these specific calculations.
I hope you find these simple analysis interesting, and perhaps even actionable. What will you be doing?
Best Regards.
Comments
Thank you for your recommendations. After careful consideration I’ve decided to reject both of them.
Regardless of my age, I’m a night person so I don’t require or do an early bedtime.
Regardless of my age, my wife and I manage a far more aggressive portfolio than you suggested. We have merged our portfolios into a single coherent entity that satisfies our long term goals. Those goals include generous inheritances to our sons and a few worthy agencies. So our timescale remains rather long.
That’s the danger of making investment recommendations with limited knowledge of the individuals, their preferences, and their target goals. The time horizon of our composite portfolio is much, much longer than my age would suggest.
As I described my analysis, it was quick and dirty, but it also only examined a single parameter that might signal market direction. Given the complexity and the uncertainties of the marketplace, a single signal is not sufficient to make any major revisions to a portfolio.
At best, any signal should be interpreted with caution, and any resultant changes should be incremental in character. That’s precisely the way I reacted to this directional indicator.
Using a conditional probability (the Bayesian method) approach reflects the true nature of market uncertainty; it can never be totally eliminated. All MFOers accept that risk as market participants.
When making any decision it’s always good policy to know the score, to understand the odds, to estimate both the potential positive and negative outcomes, to recognize the timescale, and to acknowledge what is knowable and controllable and what is unknowable and uncontrollable, In a very crude manner, my simplistic historical Conditional Probability analysis nudged my wife and I along that challenging pathway.
Thanks again for reading and responding to my post. I do appreciate both efforts.
Best Wishes.
Thanks for posting your thoughts on the January effect on the markets (S&P 500 Index) along with explaining your reasoning.
If investors have invested based upon their risk tolerance, goals, time horizon, etc. then this past January is a good opportunity for them to review how they are invested and make adjustments if this volatility brings them pause and makes them uneasy. Things have indeed changed over the past ten years. Thinking back my portfolio now generates about half the income of what it did ten years ago. After all, back then, I was getting about 4% to 5% interest on my cash area investments alone. Today, zilch.
Perhaps some have taken on more risk than they realize, in an attempt, to maintain income levels.
Old_Skeet
Thanks MJG as always for your good insight.
This past year I've taken a buy-and-hold approach with my personal investments.
Invest in companies you want to own forever, ideally, right? Isn't that the way it should be? Attendant with personal investment horizon, risk tolerance, and income need?
That said, I think David is right to suggest our risk tolerance is never as great as we think, which of course forces us to make poor decisions.
So, better that you know.
And, as usual, you sound like you do.
Hope all is well.
c
The higher, new frontier research of "tactical" investing entails the use of quantitatively derived, statistically significant, multi variables combined into a cohesive, rules based, sequentially signaled asset allocation model.
As the January barometer may have statistical merit, one of the variables incorporated into my model, "similar" to the performance output of the month of January, calculates performance scores / "risk profiles" from a time series applied to the mid month periods of the months of Nov - Jan. These risk profiles then correlate to the forward 11 month returns of the market. Logic output from other priced based variables * come into play culminating in asset allocation decisions with high confidence.
The statistical distribution of variable # 2 mentioned above https://docs.google.com/document/d/17wSv7pobin6xgGXxe6MZSfZ6C-_ungJgMQBQd9O4CJY/edit?usp=sharing ( 2016 has been recently identified as a "high risk" profile year.
* price based variables at bottom: https://stockmarketmap.wordpress.com/2015/11/14/market-map-model-tactical-asset-allocation-using-low-expense-index-etfs-2015/
I have acquired / innovated variables geared towards the "long term" trend. This is important in the construction of a tactical model and geometric compounding of assets because, as the "short term" trend reflects noise and randomness, the long term trend reflects the assimilation and evolution of "robust" factors ( expected earnings, monetary flows, economic confirmations) leading to "persistence" of trend.
Thank you for reading and replying to my post.
Based on both your current post, and many earlier submittals, you seem to be married to Seeking Alpha’s Market Map technical timing approach. I wish you success with your application of their market timing methodology.
Whenever market “experts” are asked to identify their 10 golden rules for investment success, “understanding what you are doing” is always one of the rule set. I don’t attempt to time the market major movements (which is also frequently among the 10 golden rules) so I likely would not apply the Seeking Alpha scheme anyway, but I also don’t understand their methodology.
Not being a subscriber to that site, I don’t have access to their full disclosures, but their strategy seems to be a complex multi-parameter formulation. There surely are advantages to modeling a complex problem like market forecasting in several dimensions; but there are also hazards in dong so.
Curve fitting data can produce excellent results, but over-fitting the data set is a danger too. Given enough free parameters (like an infinite sine or cosine series), a curve fit can be made perfect. However, the ultimate test is if that curve fit can actually accurately project out-of-sample or future results.
I have no idea if the Seeking Alpha formulation is proving itself on this battlefield.
I have been exposed to many other correlations that were multi-parameter complex and attractive for the given data set, but failed miserably when challenged by future data.
Markets change over time. Technical analysts have been trying to discover the Holy Grail for countless decades with very unfortunate outcomes. Perhaps Seeking Alpha has uncovered the illusive pathway to market success. Maybe not.
I sure don’t know the answer. But I am suspicious given the numerous failures of earlier technical attempts. And I greatly resist being “obedient to a presumed authority”. Show me the confirmatory out-of-sample data.
Please continue to update us on your application of Market Maps. I do hope you achieve success on your venture in that arena. Your closing paragraph suggests you have made considerable progress following your pathway. Good for you!
Best Wishes.