What It Is, How It’s Used in Business and Investing

Hybrid or Combined Approaches to Forecasting

Researchers have long been interested in which kind of methods work best and in which circumstances. Of course, different parts of finance and other disciplines tend to side toward quantitative or qualitative methods (e.g., quants are unlikely to yield studies on the gains of market research or other judgmental methods).

But researchers have often found that determining which method to use depends on a variety of factors. For instance, a company might use time series analysis to identify historical trends and supplement this with insights from market research to account for recent changes in consumer behavior.

Given this, it’s no surprise that researchers have demonstrated empirically that combined or hybrid approaches often do best by pulling in the best of both worlds for more informed predictions. Neither human judgment nor quantitative methods are universally superior. Instead, their strengths are often complementary.

Quantitative methods do best in processing large data sets and identifying patterns in them, especially when augmented with AI and machine learning, while human judgment is best in interpreting ambiguous situations and incorporating contextual knowledge that may not be captured by data alone. Reviewing recent work in this area, we culled these key points about combining qualitative and quantitative approaches:

• Combine at the end, not the beginning: Each forecasting approach should be conducted on its own. This independence ensures that each method provides its own perspective on future outcomes, free from the influence of other forecasting techniques.
• Don’t use qualitative judgments to correct for data: The last point goes hand in hand with the errors often found by using qualitative expertise to correct for quantitative data. This is the most popular way to combine approaches, but it’s often the least accurate.
• Use diverse sources of information: An effective hybrid approach relies on methods that draw from distinct pools of data. For instance, statistical forecasts typically use historical data and quantitative metrics, while judgmental forecasts often incorporate qualitative insights, market sentiment, and expert intuition. This diversity in information sources enhances how robust the conclusions can be.
• Expertise matters: When incorporating judgmental forecasts, it’s crucial that they derive from individuals with relevant domain expertise. This criterion acknowledges that effective judgmental forecasting requires more than general intuitions about a field. It requires a deep understanding of the specific field or industry for which the forecast is being done. In short, while it might seem fine to include experts with more general knowledge, the more granular the expertise, the better.

Choosing the Right Forecasting Method

The correct forecasting method depends on the type and scope of the forecast. Qualitative methods are more time-consuming and costly but can make very accurate forecasts given a limited scope. For instance, they might be used to predict how well a company’s new product launch might be received by the public.

For quicker analyses that can encompass a larger scope, quantitative methods are often more useful. Looking at big data sets, statistical software packages today can crunch the numbers in a matter of minutes or seconds. However, the larger the data set and the more complex the analysis, the pricier it can be.

Thus, forecasters often perform a quick cost-benefit analysis—a mini-forecast—to determine which method will increase their chances of accurate predictions for the least cost in time and money.

Budgeting and Forecasting: What’s The Difference?

While often mentioned in the same breath, budgeting and forecasting serve separate yet complementary roles in financial planning, investing, and business management. Budgeting is primarily a planning tool, so it has some of the elements seen above when we covered planning as opposed to forecasting.

Budgeting is a detailed estimate of future income and expenses so you have a road-map for allocating resources and setting performance targets. Budgets are often more static documents, representing management’s commitments and expectations for the period. Here are the main characteristics of budgeting:

1. Goal-oriented: Budgets reflect specific financial targets.
2. Expense monitoring: They provide a basis for monitoring and controlling expenses.
3. Fixed time frame: Usually covers a specific fiscal period, often one year.
4. Internal focus: Primarily used for internal management and accounting.

Meanwhile, forecasting is about predicting financial outcomes based on present and historical data. So, when you forecast, you aren’t setting targets. Instead, you anticipate what will happen in the future and why you do it, helping organizations and investors adjust strategies and respond to changing conditions. Here are the main characteristics of forecasting:

1. Predictive: Aims to project likely outcomes rather than set targets.
2. Adaptable: Regularly updated to incorporate new information.
3. Various time horizons: Can cover short- or long-term periods.
4. Considers external factors: Takes into account market conditions and trends, not just the internal conditions of a portfolio, business, or economic sector.

12 Principles of Effective Forecasting

Effective forecasting is a critical skill in business and finance, providing a basis for the decisions that can make or break a business or portfolio. While no forecast is perfect, companies and investors with dependable forecasts are better equipped to navigate uncertainties, seize prospects, and maintain a competitive edge. The following principles, drawn from expert insights and those with practical experience, form the core of effective forecasting:

1. Be methodical: The best outcomes derive from forecasting that adheres to a systematic, well-defined process. Using a systematic, repeatable approach ensures consistency, allows for continuous improvement, and improves the reliability of your predictions.
2. Look back to look forward: A rule some researchers cite is looking back at least twice as far into the past as you are forecasting into the future. This simply means looking back at history. Forecasters can identify patterns that can help predict the future if they examine trends over a long period. The recent past is unreliable, so forecasters should look back at least twice as far into the past as they forecast into the future. The unexpected happens, and history doesn’t always repeat itself. Be mindful of history, not a prisoner of it.
3. Embrace uncertainty: Perfect forecasts are for the gods, not for you. All predictions carry a degree of uncertainty, reflecting the complex and dynamic nature of business environments.
4. Quantify your uncertainty where possible: Forecasts that can provide a “distribution” of possibilities—an airline predicting the price of jet fuel between minimum and maximum likely prices, for example—help the most in planning since that doesn’t make the prediction an all-or-nothing affair.
5. Be aware of wild cards: These are low-probability events with high potential impact at the edges of the possible range. Acknowledging outliers and unexpected events is crucial for comprehensive forecasting and, especially, for risk management.
6. Greater accuracy is found in the aggregate: Forecasts are more precise when applied to broader categories or groups rather than individual items. This principle, known as the law of large numbers, is at the heart of statistics and means that forecasts yield more reliable predictions for aggregate data.
7. Take heed of the “S curve”: In data science, the S-shaped curve occurs when some process starts slowly, speeds up quickly, and then levels off. According to Silicon Valley-based forecaster Paul Saffo, recognizing this pattern early can help you anticipate different stages of development in various fields and plan accordingly. Forecasters should identify the S-curve pattern as it begins to emerge, and they can then look for precursors to an inflection point rather than the inflection point itself (when changes made might be too late).
8. The longer the time, the more that can go wrong: The accuracy of forecasts typically diminishes as the time horizon extends. Near-term predictions generally offer greater precision than long-range projections. Unforeseen variables can simply compound their effects over time.
9. Look for the oddballs: Embrace things that don’t fit. New ideas often emerge as weak signals that can be hard to recognize because they seem strange or don’t fit into existing categories. These “oddball curiosities” sometimes are indicators of future trends. For example, the sale of virtual goods in online games in the late 1990s foreshadowed the rise of virtual world commerce like Second Life. Forecasters should be tuned to these indicators, which often appear as mere curiosities or failures but can actually be forerunners of significant change.
10. Hold strong views weakly: Don’t fall in love with the oddballs, though. The golden rule of forecasting, as one researcher puts it, is to “be conservative” and rely only on knowledge and methods consistent with the problem at hand. This means also that forecasters should be open to new information that contradicts their initial assumptions. Sometimes, substantial evidence can be misleading, while seemingly weak evidence can signify a future trend. Forecasters should always be willing to revise or discard forecasts when contradictory evidence surfaces. If forecasters adopt a process of strong opinions, weakly held, their range of uncertainty will be refined over time toward more accurate predictions. This rule encourages continuous refining rather than relying on a single, unyielding forecast.
11. Combine methods independently: When using multiple forecasting methods, ensure they are generated separately, based on different information sources, and incorporate domain expertise where relevant. Researchers have found that this makes for more accurate predictions.
12. Know when not to forecast: Sometimes, the future is too uncertain to make any predictions. During periods of dramatic, rapid transformation, a good forecaster will refrain from making definitive predictions and instead look for emerging indicators that can provide clues about the future. This is just a reflection of the reality that even in times of significant change, there are often more elements that remain constant than new elements that arise. Forecasters should be aware of the limits of their knowledge and avoid making pronouncements when there’s too much uncertainty.

In addition to these rules, you’ll want to practice the number one principle for what to do after you put a forecast into action through planning and budgets: Regularly reassess the forecast accuracy using appropriate measures and adjust your models as needed to improve performance over time.

The Bottom Line

Forecasts help managers, analysts, and investors make informed decisions about the future. Without good forecasts, many of us would be in the dark and would resort to guesses or speculation. By using qualitative and quantitative data analysis, forecasters can better understand what lies ahead.

Businesses use forecasts and projections to inform managerial decisions and capital allocations. Analysts use forecasts to estimate corporate earnings for subsequent periods. Economists may make more macro-level forecasts as well, such as predicting GDP growth or changes to employment. However, since we cannot definitively know the future, and since forecasts often rely on historical data, their accuracy will always come with some room for error—and, in some cases, may end up being way off.