What it is, how it is used in business and investments

What is prediction?

Forecasting is a technique that uses historical data as input to make informed estimates that are predictive of determining the direction of future trends.

Businesses use forecasting to determine how to allocate their budgets or plan for expected expenses for an upcoming period. It is typically based on the projected demand for the goods and services offered.

How forecasting works

Investors use forecasting to determine whether events affecting a company, such as sales expectations, will increase or decrease the price of shares in that company. Forecasting also provides an important metric for firms that need a long-term perspective of operations.

Stock analysts use forecasting to extrapolate how trends, such as gross domestic product (GDP) or unemployment, will change in the coming quarter or year. Finally, statisticians can use forecasts to analyze the potential impact of a change in business operations. For example, data can be collected on the impact of customer satisfaction by changing business hours or the productivity of employees by changing certain working conditions. These analysts then come up with earnings estimates that are often aggregated into a consensus figure. If actual earnings announcements miss estimates, it can have a big impact on a company’s stock price.

Prediction addresses a problem or states data. Economists make assumptions regarding the situation being analyzed that must be established before the variables of the forecast are determined. Based on the items determined, an appropriate data set is selected and used in the manipulation of information. The data is analyzed, and the forecast is determined. Finally, a verification period occurs when the forecast is compared with the actual results to establish a more accurate model for forecasting in the future.

Forecasting techniques

In general, forecasting can be approached using qualitative or quantitative techniques. Quantitative methods of forecasting exclude expert opinions and use statistical data based on quantitative information. Quantitative forecasting models include time series methods, discounting, analysis of leading or lagging indicators, and econometric modeling that can attempt to establish causal relationships.

Qualitative techniques

Qualitative forecasting models are useful in developing forecasts with a limited range. These models are highly dependent on expert opinions and are most beneficial in the short term. Examples of qualitative forecasting models include interviews, site visits, market research, opinion polls and surveys that may apply the Delphi method (which relies on aggregated expert opinions).

Collecting data for qualitative analysis can sometimes be difficult or time-consuming. The CEOs of large companies are often too busy to take a phone call from a small investor or show them around a facility. However, we can still sift through news stories and the text included in companies’ filings to get a sense of managers’ track records, strategies and philosophies.

Time series analysis

A time series analysis looks at historical data and how various variables have interacted with each other in the past. These statistical relationships are then extrapolated into the future to generate predictions along with confidence intervals to understand the probability that the actual outcomes fall within that range. As with all forecasting methods, success is not guaranteed.

The Box-Jenkins Model is a technique designed to predict data series based on inputs from a specified time series. It predicts data using three principles: autoregression, variance, and moving averages. Another method, known as rescaled series analysis, can be used to detect and evaluate the amount of persistence, randomness, or mean reversion in time series data. The rescaled range can be used to extrapolate a future value or average for the data to see if a trend is stable or likely to reverse.

Mostly, time series forecasting involves trend analysis, cyclical fluctuation analysis and issues of seasonality.

Econometric inference

Another quantitative approach is to look at cross-sectional data to identify links between variables – although identifying causality is difficult and can often be spurious. This is known as econometric analysis, which often uses regression models. Techniques such as the use of instrumental variables, if available, can help one make stronger causal claims.

For example, an analyst can look at income and compare it to economic indicators such as inflation and unemployment. Changes to financial or statistical data are observed to determine the relationship between multiple variables. A sales forecast can therefore be based on various inputs such as total demand, interest rates, market share and advertising budget (among others).

Choosing the right forecasting method

The right forecasting method will depend on the type and scope of the forecast. Qualitative methods are more time-consuming and expensive, but can make very accurate predictions given a limited scope. For example, they can be used to predict how well a company’s new product launch will be received by the public.

For faster analyzes that may include a larger scope, quantitative methods are often more useful. When looking at large 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 more expensive it can be.

So forecasters often do some sort of cost-benefit analysis to determine which method most efficiently maximizes the chances of an accurate forecast. Furthermore, the combination of techniques can be synergistic and improve the prediction’s reliability.

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 resort to guesswork or speculation. By using qualitative and quantitative data analysis, forecasters can get a better grasp of what lies ahead.

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