Forecasting does add value to the decision-making process. But it is not what you do that matters, it is how you do it.
Everyone Forecasts
Forecasting is the act of making predictions about the likelihood of something happening in the future. It is not a cognitive activity confined only to weather, politics, business, or sports. We all forecast in our everyday lives, to a certain extent, whether we are aware of it or not.
When we leave our house without an umbrella on a dark cloudy day, we forecast it won’t rain. We forecast less traffic when we choose one particular route to work among a set of alternate routes. We forecast bad grades and a dull future if we don’t study at school or college, so we put effort into our studies to avoid a bleak future.
The accuracy of our forecast determines if we have made a good forecast or a bad one. We left the house without an umbrella as we forecasted no rain; but it if rains, then we have a bad outcome and a forecast with poor accuracy. Similarly, a good outcome and an accurate forecast ensued if we reached work earlier than usual by selecting the particular route.
The accuracy of our forecast is determined by what we forecasted and what happened. The closer they are, the higher the accuracy.
We are also forecasting when we invest in a stock. We forecast the stock will give us high returns in the future. We forecast the stock’s earnings and cash flow to grow in the future. We forecast a more robust business and financial position in the future.
We want our forecast to be as accurate as possible so that we can make quality decisions based on that, and thus, later, benefit from the expected positive outcome.
The only thing we are sure about the future is: it is uncertain. Events can unfold in thousands of possible ways. Forecasting is the estimation of the likelihood of all these possible ways happening.
Now, our question is: Is forecasting possible? Can a forecast with reasonable accuracy be made with the information available to us? If it is possible, how do we make a good forecast?
And the most important question is, does forecasting add any meaningful value to our decision-making process?
Expansive, long-term studies done on forecasting finds it is possible to estimate the likelihood of an event in certain situations to a certain extent. As the future is uncertain, the act of forecasting can help reduce this uncertainty to some degree that enables us to take effective decisions. It is not an exact science. It is a mixture of both art and science.
In investing, the time frame matters when it comes to forecasting. It is difficult to estimate the stock price for the next day or next week. But if we move the time frame to 6 months or even 2 years, it is possible to estimate future stock prices to a certain degree. Again, when we move the time frame too far into the future, say 5 years or 7 years, the uncertainty is so high that effective forecasting is almost impossible.
Okay. Now we believe it is possible and productive to forecast stock prices for 6 months to 2 years. But how do we make a good forecast? What distinguishes a good forecast from a bad forecast? What are the ingredients needed to make a good forecast?
Low Accuracy Is Not That Bad
In the case of stock investing, even if the forecast lacked high accuracy, the act of forecasting itself can lead to improvement in the decision quality. Our purpose in building and managing an equity portfolio is to achieve superior investment returns consistently. And to that end, our forecast of stock prices need not be 100 percent accurate.
Suppose we estimated a fair value of ₹140 per share for a stock that is currently trading at ₹100 per share. We forecast the stock price to appreciate toward its fair value in six months. But after six months, the stock rose to ₹110 per share only, which means our forecast missed the mark and had poor accuracy.
But the story doesn’t end there. In the three months after the six months, the stock appreciates from ₹110 per share to ₹150 per share. We forecasted a 40 percent return (96 percent annualized return) in six months, received only a 10 percent return (21 percent annualized return) in six months, but a 50 percent return (71 percent annualized return) in nine months.
If we had sold the stock at ₹110 per share after six months, as things didn’t work out as forecasted, we will receive a 21 percent annualized return. If we had held onto the stock for another three months, then we would receive a 71 percent annualized return, and that is a significant difference. Selling at ₹110 per share after six months would have given a mediocre return, meanwhile holding on to it for another three months would have given a far superior outcome, even though both scenarios are farther from what we forecasted.
Consider another alternate scenario in which in the subsequent three months, instead of the stock appreciating from ₹110 per share to ₹150 per share, it depreciated to ₹85 per share. In this scenario, selling at ₹110 per share after six months looks like a good decision.
The above real-life-experience-inspired, but fictionalized situation might compel us to think forecasting is futile, useless, misleading, and confusing. But it is not, because it was the concreteness of our forecast (₹140 per share in six months; 40 percent expected return) that gave us the clarity and confidence to invest in the stock.
The problem is: what to do if things don’t work out as forecasted? Or, what to do if the stock price moves in dissonance with our forecast?
One method to deal with this problem is to consider every forecast as a work in progress. We can update our forecast, at least every three months, based on new information or insight.
This way, even though the stock price didn’t rise to the estimated fair value in six months if there still exists a profitable price difference between the current market price and the estimated fair price from the new updated forecast, then we can decide to hold onto the stock. This will prevent us from selling the stock and missing the subsequent return. Also, if the stock moves contrary to our forecast, then that could be considered an incentive to look for the information we might be missing. If you find one, update the forecast, and then take a decision.
What I intend to say through this discussion is that it is not necessary to be 100 percent accurate in all our forecasts. A 70 percent accuracy in maybe 60 percent of the forecasts should be considered an achievement. Forecasts can always be updated and improved based on new information and insights. At the same time, we can strive to improve our forecasting accuracy by learning and developing our forecasting skills.
Information Gathering
We need information to make forecasts. We analyze this information to gain insights and make forecasts based on it. The information we collect must be relevant to the situation or event on which we forecast. It must be useful and have a high degree of influence on the event outcome.
A good forecaster must be a good information gatherer, because however complex and detailed analysis system you have, if you feed wrong information into the system, you get a poor or mediocre output.
How much information should we collect? And what are they? The 80/20 principle states that you can know 80 percent of any subject from 20 percent of information about the subject. Being 80 percent knowledgeable about a subject is far enough for effective forecasting. The hard part is finding out which is the ’20 percent information’.
Our concern is investing in stocks and achieving superior returns. And for that, I believe competitive position, capital allocation, cash flow, earnings growth, and valuation are the principal factors that determine a stock’s future price movement. If so, then, any information that enlightens us on these parameters is the ’20 percent’ information.
Pattern Recognition
Once we have gathered a lot of relevant, useful information on the subject, the next step is to aggregate all this information to get a complete picture of the situation. A bird’s eye view of the situation.
The aggregate of information is analyzed to identify patterns. To do that, a clear, unbiased mind is imperative. We should use different analytical tools on the aggregate to recognize patterns that can lead to insight. Also, just like information gathering, we need to gather as many patterns as possible from the aggregate.
Obstacles in the Path
A major mistake we make, whether it is ‘information gathering’ or ‘pattern recognition’, is jumping to conclusions too soon. Don’t commit the mistake of reaching a conclusion based on just one or two patterns. Suppose that we observe that a stock has high earnings growth over the long term, but tepid earnings growth in recent quarters or years. The insight from this pattern is slowing earnings growth which is negative from an investment perspective. But avoiding the stock just based on that one pattern is a mistake.
But it is not that easy to avoid this mistake: anyone knowledgeable in human nature can know because of the way our brain works.
Confirmation bias creeps in when we jump to a conclusion or form a belief or opinion too early. When under the influence of confirmation bias, we look for information or patterns that agree with our conclusion or belief, and at the same time, fail to see or reject information or patterns that are in opposition to our beliefs.
It is not our fault; it is just the way our brain works. The brain craves certainty, but the world is an uncertain place. To overcome that, the brain creates a sense of certainty by reaching conclusions too early, maybe with the first bit of information itself, which unfortunately will not be coherent with reality.
How to Deal with the Obstacles
Accepting uncertainty could be a starting step in dealing with confirmation bias. Accepting uncertainty means being aware of our brain’s tendency to create certainty despite the world being uncertain. Once we have accepted uncertainty, then we start thinking in probabilities.
When we think in probabilities, we won’t say, “I am 100 percent sure of this event happening” anymore. Our information gathering and processing will not provide us with certainty but only help in reducing uncertainty to a certain extent. Now we say, “There is a 65 percent chance that the event happens” which means, despite all our efforts, there is still a 35 percent chance for the event not happening.
When we are aware of the “65 percent of happening” and “35 percent of not happening”, we take decisions differently, and in a better way. If it concerns investing, we won’t invest all our money in a single stock as there is a 35 percent chance for the investment not working out.
Skepticism can help us deal with confirmation bias. When the brain jumps too soon to conclusions or opinions, a skeptical mindset will make us doubt the quality of the conclusion or opinion, rather than accepting them. We are more prone to gullibility if the opinion comes from our minds. So, a skeptical mindset is imperative for effective forecasting and decision-making.
Perspective taking. There are always other knowledgeable people out there. We need to go out and find them, then qualify them, which means, are they qualified to speak on the matter: In our case, on the prospect of financial markets and stock prices.
Perspective-taking is the process of getting out of our heads and listening to what other people have to say on the issue. And trust me, it is not that easy a task. There are challenges to overcome.
Don’t commit the mistake of feeling good and sure when someone agrees with your view. Likewise, don’t antagonize people having perspectives different from ours. We should have an open and unbiased mindset. We are doing this to learn and improve, not to prove anything.
Don’t be gullible and make radical changes to the forecast if your view is in wide variance with most of the qualified subject matter experts. Stay humble, keep a balanced mind, and make a series of small incremental changes to your forecast, and that itself if you found that you have missed some relevant fact or perspective.
Open-ended Judgement
We have now collected a diverse set of information and then identified a diverse set of patterns from the aggregate information. By accepting uncertainty, through probabilistic thinking, and using skepticism and perspective-taking, we bring our forecast as closely as possible to the truth.
Finally, it is time to make the judgment. Make the judgment as precisely as you can in probabilistic terms. Ambiguity is the enemy of a good forecast. Ambiguous forecasts hinder effective decision-making and make it difficult to evaluate the forecast for accuracy later.
An ambiguous forecast looks like this: ‘The stock has high upside potential.’, whereas a precise forecast looks like this: ‘There is a 65 percent probability for the stock to appreciate towards ₹150 per share within the next 6-8 months.’
A precise forecast brings clarity that helps in fast, effective decision-making. It also helps in evaluating the accuracy of the forecast later.
The forecast should be updated based on new information or insight. A forecast that is updated regularly will be closer to the truth than one that is not updated. A good forecast has high accuracy and aids us in our decision-making process. It improves the quality of our decisions. Forecasting is a skill that any intelligent, hard-working person can learn, and then refine through practice. Like any skill, to be a good or even a great forecaster you need to: practice, practice, and practice a lot.
AMN CAPITAL 