Accuracy of Historical Data
Financial forecasting often is performed using historical results as a proxy for the future. You can do this by analyzing historical income statement and balance sheet items for trends, such as growth trends, and applying these figures going forward. For example, if a company achieved stable growth averaging 5 percent per year for the past five years, you could forecast next year's sales using a 5 percent growth rate. While widely used, this approach can be problematic. If the company's results are erratic from year to year, historical averages may not provide good indications for the future. If the company is a start-up, historical results may not be available at all. In addition, external market conditions can affect financial results in a way that would not be captured by analyzing historical results.
The longer the time frame, the more difficult it will be to accurately forecast financial results. It is less difficult to forecast next year's financial results than anticipate numbers for the upcoming decade. For example, if you are extrapolating trends using five years of historical data while preparing 10-year financial projections, the applicability of a five-year trends would likely be lower to a 10-year period. As more time elapses, the probability of events occurring which could affect the company's financial results increases. Market share can increase or decrease, or economic conditions can change substantially. As a general rule, shorter projection periods are more accurate.
Problems With Input Data
Besides using historical data, forecasts often are performed using linear analysis, which pegs future financial performance to various dependent variables correlated with the underlying financial figures. This can be highly problematic -- best captured by the expression garbage in, garbage out. Your forecast's reliability is only as good as the inputs used to calculate it. This leaves room for errors caused by mistakes made in collecting or interpreting the data, or human error in entering data into the forecasting model. Also, humans are subject to various biases, such as confirmation bias, which occurs when the forecaster's judgement is skewed by predisposed notions about the projected results. This can cause the forecaster to place too much emphasis on less relevant data items, or vice-versa.
Even if you perfectly perform the quantitative and qualitative forecasting methods, it is impossible to foresee the unforeseeable. These elements can vary in nature, but can be risks based on competition, the economy, and external shocks to the market. For example, after many years of growth, Blockbuster was blindsided by the performance of Netflix, which very quickly eroded Blockbuster's market share and sales. A retail outlet can open a new location and project strong financial growth, only to have a direct competitor open up across the street, affecting sales and earnings.
Furthermore, a Black Swan event can easily render well-prepared financial forecasts obsolete. A Black Swan event is a highly unlikely occurrence that occurs, exhibiting three factors -- it is impossible to predict, it carries a massive impact, and its shock value is stunning, because people could never conceive of such an event occurring.