Every year during January there is a special seasonal danger in the investment world, and that danger is the inevitable stream of predictions for the coming year. One of the most important things for a successful investor to learn is how to evaluate information, and particularly how to avoid being influenced by bad information or information that is not profitably actionable. Acting on bad information or bad actions based on good information is sure to lead to poor performance, and even though predictions are concentrated at the beginning of the year, a successful investor has to be on guard against the danger of predictions throughout the year. The way I have come to think of it is in order to be a good investor, one has to develop an immune system similar to a biological immune system that will quickly identify and reject bad information, or will sift through bad information to determine what is potentially profitable and what is not. To this end, here is a list of some of the questions that act as my immune system and that I ask about any prediction I hear.
Does the prediction acknowledge and address the strongest opposing arguments?
I have much greater confidence in an argument when the predictor acknowledges and refutes truly strong opposing arguments rather than stating one side of the case and leaving it to the reader to go out and research what the opposing arguments are. This is not to say that one can avoid independently determining and evaluating opposing arguments, but I find that the predictions from people who proactively acknowledge counterarguments are much more likely to have merit, and it helps me narrow down what to pay attention to for further investigation.
Is the predictor attempting to predict something that is currently unpredictable?
Most assets, most of the time, are not, in my opinion, sufficiently predictable to justify an investment. One of the keys to successful investing is to be able to identify when a clear direction or thesis emerges, and having the patience to wait to invest until this occurs. Warren Buffett and Charlie Munger are masters at this, and most of the people claiming to have identified "Buffett-like" investments forget that those two made many of their major investments during fairly short windows of opportunity when a clear extreme of risk/reward was present. Most predictions I see should really be "I don't know" or "I will make a prediction when I can be confident," based on the currently available data. Anyone can project a trend; however, most assets don't have a combination of foreseeable fundamental underpinnings and market recognition to justify a prediction.
The problem with hearing forced predictions is that it creates a mindset that suggests one should be expected to have a prediction on any asset at any time. However, this is exactly the opposite of a successful investor's mindset. When I have made bad trades it is often because I forced myself to develop a point of view on an asset when I should have said "not a fat pitch, too little clarity, move on until a better situation presents itself." This is also why it is important to not have too narrow an investing focus, so that you have enough assets to choose from and are less likely to force an investment that is simply too low of a probability.
Does the prediction respect the related trends?
If a prediction requires a trend to change (whether in the price of the asset itself or a related asset like a commodity price), the prediction is not actionable without a compelling scenario for how and when the associated trend and its underlying fundamental factors will change. Trending markets tend to take on a self-reinforcing nature that can persist far beyond what fundamental factors would suggest. A related question is whether the prediction is actionable until the related trends actually show signs of changing, and the answer is usually "no" -- unless there is the potential for a sudden and major gap-type change in trend. In short, it often pays to wait until what "should happen" shows sign of actually happening.
On the other hand, if a prediction depends on a mature trend continuing, the fundamentals behind the prediction can be sound. However, the asset price could already be exaggerated by investors who are simply following the trend and who will exit at the first sign of weakness, raising the risk of a sudden extreme reversal. Improving the degree to which I factored trend characteristics (e.g., maturity, amplitude, related trends) into my position sizing, investment time frame, risk management and market/asset selection has been one of the single biggest ways that I improved my investing performance, and I would particularly encourage investors interested in this topic to check out the interviews and other information available at michaelcovel.com that have been a big help to me in this regard (no affiliation).
What is the downside if the prediction is wrong? What is the risk/reward? What would invalidate the prediction?
I see many predictions that say something like "such-and-such has 30% upside." Let's assume that is correct and not worry about how everyone could possibly exit at close to a specific point of full valuation. The important question is: How much could I lose before the prediction is clearly invalidated? Answering this question requires two things: the criteria for disproving the prediction and the amount one might lose before this criteria is met. This is almost never present with any prediction, so an investor must determine if it can be calculated.
A closely related problem is that many predictions are inherently not actionable because they have too little margin of safety. An example would be "X stock has 10% upside." Well, if 10% really is the maximum gain and a reasonably expected gain is 7%, there is a scenario for most stocks that would see a gap down of 5% of more. So, by definition, unless the chance of the 10% gain is very high (highly unlikely), then the expected risk/reward does not provide a good margin of safety.
Is the prediction actionable? What profitable action can I take?
Most of the predictions I see are simply not actionable, or will be revealed to be actionable only by waiting for more reliable data, such as monitoring a specific upcoming fundamental event. For example, I have seen a number of predictions for a major correction in the broad averages in 2014. So what? That may well occur, but there is simply not sufficient broad-based selling pressure to far to justify anything more than tactical short positions in individual names with tight stops. And if selling pressure does develop, it will be readily apparent in the price action. When that will occur is a big unknown and one would leave significant upside on the table or suffer significant losses on their short positions by acting before decisive selling appears. Additionally, I have heard many predictions for a correction, however a correction is, to my way of thinking, only a correction in hindsight. Until an uptrend resumes, a correction is a downtrend.
Is the time frame meaningful, or is it arbitrary or irrelevant?
I frequently hear predictions based on data that has been unchanged for the past six months or a year. The obvious question is: Why now? What is different now compared to six months ago? This is a sure sign that a catalyst is missing from the prediction or that the prediction is already priced in by the market or is or simply wrong, and that a counter-bet may be the better play, if there is any play. In other cases the time frame itself may render the prediction meaningless. For example, I saw a prediction today about the relative performance of stock and bonds over the next decade.
While it may be reasonable to buy bonds with a duration of a decade or more as part of a diversified strategy, suggesting that someone should take action based on something that is unknowable (the world in 10 years) or, more importantly, not be ready to immediately change course as the fundamentals change is very counterproductive. Also, thinking in terms of a year is quite arbitrary, and is a distraction from identifying key milestones during the year based on actual events that should influence one's investment time frame, such as the debt ceiling deadline or the timing of the Affordable Care Act's impact on consumer spending.
Does the prediction seek attention by predicting an extreme outcome?
People seem to enjoy contemplating extreme outcomes and, as a psychology professor said to me in college: "The long shot is always overvalued." I agree that it is worthwhile to ponder tail events and to seek out points of view that reduce complacency, so in that sense these types of predictions can be valuable. But in my experience, it is psychologically tempting to bet prematurely on extreme events rather than waiting for confirmation that the event has become considerably more likely. Waiting will, of course, reduce the return if one is correct, but extreme events often take longer to happen than many prognosticators would have you believe. And, yes, I know this may appear to be counter to the idea of Black Swans. However, part of the problem with predicting "tail risk" events is that they have a tendency to morph into different risks and be harder to profit from than many people expect. Also, when extreme events become well-enough identified that people are predicting them, it often means that the real risk lies elsewhere.
Do I have to rely on the predictor to tell me when the prediction is no longer valid? Can the reasoning be independently verified?
Occasionally one will hear predictions (usually about obscure companies or asset classes) where there is almost no practical way to cross-check the thesis for reliability and, just as importantly, no way to easily tell when something that supports the thesis may have materially changed. This type of prediction is often associated with selling a pitch for some sort of newsletter or brokerage service. The problem here is not that the information is likely to be wrong, although that is certainly a possibility. The problem is that you have no way to know when the thesis has changed without relying on the predictor, so any investment is essentially blind.
In general, if you have a strong urge to invest in things that rely on a level of detailed knowledge that you don't have and can't realistically get, then you are better off investing through some sort of fund vehicle where there is a least a track record that you can verify and an ongoing performance record that will give an indication if the thesis is playing out. In this case, all buy and sell decisions for individual securities are made by someone who should know the details necessary to be successful. I don't necessarily recommend this approach either, but it is better than going into a situation where you are even less likely to make good decisions and have to rely on someone else who may have an incentive for you to buy, but not to sell in time to avoid a change in thesis.
Is there a "parlor trick" involved?
A big part of developing your investment immune system is becoming like a magician at a magic show who knows how every trick is done, but can appreciate creativity and can occasionally be inspired by seeing a new trick to develop something new for his or her own use, without becoming bored or cynical at seeing things that offer nothing new. Unfortunately, there is no substitute for time in the markets and experience with the pros and cons of different methods to be able to develop this sense. However, one indication that something is hidden in the picture is some aspect of the prediction that appears too good to be true. An example of this I hear from time to time is from people who site an extremely high percentage or streak of "winning trades" to justify their credibility.
There are various approaches that can be used to create a high win percentage, like option writing, for example. But in that case it is critical to remember the words of George Soros who said, "It's not how often you are right or wrong, but how much you make when you're right and how little you lose when you're wrong." Another example in this category are approaches that work based on "back testing" during a period of a bull market but that would be a disaster during a significant downturn.
Is the prediction using a benchmark instead of absolute return?
Most investors are absolute return investors rather than relative return investors (or they will wish they were absolute return investors in the event of a major market downturn.) In other words, beating a benchmark like the S&P 500 is acceptable if the market is up. But beating a benchmark by "only" losing 30% if the benchmark is down 35% is something most investors would not find acceptable if it actually happened. For this reason it is import to be on guard against predictions that are in the context of relative return, such as "Company X will outperform the market by 10%." While this may be true and have some value in terms of margin of safety, it does not address what risks might cause an outright deterioration of Company X's business or other factors that could cause a loss of capital.
Does the prediction add any value beyond a simpler decision-making approach?
In the spirit of Einstein's advice to make things as simple as possible but not simpler, it is wise to ask if a prediction adds any value over a simpler but still potentially valid approach. For example, any prediction about the S&P 500 needs to add some value beyond a simple approach such as "stay long as long as the uptrend channel is not broken." Most importantly, a prediction fails this test if it gives no compelling reason to stay long if the uptrend is broken.
Does the prediction identify conditions that are legitimate risks or opportunities that bear watching, even if the prediction itself fails other tests on this list?
From the preceding list it may appear that I find no value in predictions, but there are frequently predictions that I find very useful. Those are predictions that do not focus on a specific outcome but rather identify under-appreciated events or trends that can be monitored to determine if they are actually materializing in a way that creates risk or opportunity. I always think of Peter Brandt's counsel to focus on possibilities, not probabilities, and making a list of possibilities to be monitored is a useful part of prediction season for me. Examples could include making a list of commodities whose trends have extended to extreme points such that they may fall below the cost of production or are close to rising to a point that would encourage extensive switching. This would not be a reason to immediately take a position, but would rather be a reason to create a list of possible trend changes that could justify being on high alert, hedging against sudden changes or reducing the size of any positions taken.
A significant portion of my profits last year came from predictions that I thought were total nonsense, but that triggered enough of a new idea that by applying an investment approach that better fit my strategy, I could create a profitable investment in a company or market that I had not previously considered -- even if it was to do the opposite of the recommendation. Finally, I predict that over-relying on predictors or prognosticators will cause your investment performance to suffer. I guarantee it -- I've been right every year.