How to Build a Simple Earnings Forecast Model

Building your own earnings forecast model is one of the most valuable skills you can develop as a fundamental investor. While analyst estimates and consensus forecasts are readily available, there is no substitute for building your own model. The process forces you to understand the specific drivers of a company's business, make explicit assumptions about the future, and quantify the impact of those assumptions on earnings. It transforms you from a passive consumer of other people's research into an active, independent analyst.

Your own model also gives you a framework for evaluating new information. When a company like Bajaj Finance reports monthly business updates, your model tells you what those numbers mean for the quarterly forecast. When management provides guidance on NIM or loan growth, your model shows you the earnings impact. When a competitor enters the market or a regulatory change is announced, your model helps you quantify the potential effect. This ability to translate news into earnings impact is a powerful analytical edge.

Before building your model, make sure you understand the financial statements that will feed into it. See our guides on How to Read an Income Statement and The 3 Financial Statements Every Investor Must Know for the prerequisite knowledge.

Revenue forecasting is the foundation of any earnings model. Get the revenue forecast right, and the rest of the model flows from it more naturally. The approach to revenue forecasting depends on the type of business you are modeling. For a retailer like Titan, the revenue model breaks down into two components: same-store sales growth (revenue from existing stores) and new store contributions. For a bank like HDFC Bank, the model breaks into loan volume growth and net interest margin.

Volume-Price-Mix Analysis

For most product-based companies, revenue can be decomposed into volume, price, and product mix effects. Volume is the number of units sold. Price is the average selling price per unit. Mix refers to the proportion of high-margin vs low-margin products sold. For Maruti Suzuki, you would forecast total vehicle sales volume (using monthly sales data as a leading indicator), average selling price per vehicle (considering the shift toward higher-priced SUV models), and export volumes. Changes in any of these three drivers have different implications for margins, so the decomposition is valuable beyond just the revenue forecast.

Top-Down vs Bottom-Up

Top-down forecasting starts with the overall market size and applies market share assumptions. For Britannia, you might start with the Indian biscuit market size, estimate category growth, then apply Britannia's market share to derive revenue. Bottom-up forecasting starts with specific operational drivers — number of stores × average revenue per store for a retailer, or number of subscribers × ARPU for a telecom company. The best approach often combines both: use a top-down check to validate that your bottom-up forecast is reasonable relative to the industry context.

Once you have a revenue forecast, the next step is modeling the cost structure. This requires understanding the company's operating leverage — the relationship between fixed and variable costs. Companies with high fixed costs (like airlines, cement manufacturers, and telecoms) have high operating leverage: a small change in revenue leads to a large change in profits. Companies with primarily variable costs (like IT services firms) have lower operating leverage but also lower business risk.

Cost Line Item Analysis

Rather than forecasting a single margin percentage, break costs into their components. For Hindustan Unilever, this means separately forecasting raw material costs (linked to commodity prices like palm oil and crude oil derivatives), employee costs (typically growing at 8-12% annually), advertising and promotion spend (often a percentage of revenue), and other expenses. Each cost line item has different drivers and different predictability. Commodity-linked costs require tracking global input prices. Employee costs are more predictable based on headcount growth and salary inflation assumptions. This granular approach produces a more accurate and useful model.

Depreciation, Interest, and Taxes

Below the operating profit line, three items complete the earnings model. Depreciation is linked to the company's fixed asset base and capital expenditure — it can be modeled as a percentage of revenue or using a more detailed fixed asset roll-forward schedule. Interest expense depends on the company's debt level and borrowing costs. For companies with significant debt (like infrastructure or telecom firms), interest modeling is critical. The effective tax rate for Indian companies is typically 25-30%, but can vary due to tax incentives, deferred tax adjustments, and one-time items. Use the company's normalized tax rate rather than the reported rate from any single quarter.

With revenue and cost forecasts in place, calculating EPS is straightforward: subtract all costs from revenue to get net income, then divide by the number of diluted shares outstanding. However, several adjustments are needed for a meaningful EPS forecast. First, decide whether you are forecasting reported GAAP EPS or adjusted (non-GAAP) EPS — adjusted EPS excludes one-time items that management considers non-recurring. Second, account for share count changes due to buybacks or ESOP issuance.

Example: Forecasting Infosys EPS

Let's walk through a simplified example for Infosys. Assume you forecast USD revenue growth of 10% with an INR/USD exchange rate of 83, giving INR revenue of INR 1,60,000 crore. Employee costs (the largest cost item) are modeled at 55% of revenue, other costs at 15%, and depreciation at 3%, yielding an operating margin of 27%. Apply other income of INR 2,000 crore, interest of INR 500 crore, and a 25% tax rate. This produces net income of approximately INR 32,000 crore. With 415 crore diluted shares outstanding, the EPS forecast is about INR 77 per share. Compare this to the analyst consensus to gauge whether your assumptions are more or less optimistic than the market.

Common Modeling Pitfalls

Beginners often make several common mistakes in EPS forecasting. First, they assume historical growth rates will continue indefinitely — no company grows at 20% forever. Second, they extrapolate margins without considering operating leverage or competitive dynamics. Third, they ignore the impact of share buybacks on per-share earnings. Fourth, they use overly precise assumptions (forecasting revenue to the nearest crore) when the inherent uncertainty makes broad ranges more appropriate. The mark of a good model is not perfect precision but a clear articulation of assumptions and their sensitivity.

No forecast model is complete without sensitivity and scenario analysis. Since every assumption in your model is subject to uncertainty, you need to understand how changes in your key assumptions affect the EPS output. This analysis transforms your model from a single-point forecast into a probabilistic framework that acknowledges uncertainty — the hallmark of sophisticated investing.

Building a Sensitivity Table

Create a sensitivity table that shows how EPS changes as your key assumptions vary. For an IT company like Infosys, the two most critical and uncertain variables are revenue growth rate and operating margin. Build a matrix with revenue growth ranging from 8-12% on one axis and operating margin ranging from 25-29% on the other axis. The resulting grid shows EPS under each combination. If EPS ranges from INR 70 to INR 85 depending on these two variables, you know that 80% of the uncertainty in your forecast comes from these two inputs. This insight focuses your research effort on the variables that matter most.

Scenario Planning

Beyond sensitivity analysis, construct three discrete scenarios: a bullish case, a base case, and a bearish case. The base case should represent your most likely estimate (the mode of your probability distribution, not necessarily the mean). The bullish case assumes favorable outcomes for your key assumptions — stronger growth, better margins, favorable currency. The bearish case assumes headwinds. Assign subjective probabilities to each scenario. For example: base case 60%, bullish 20%, bearish 20%. The probability-weighted expected EPS gives you a more robust estimate than any single scenario and helps you avoid anchoring to one optimistic projection.

A forecast model is not a one-time exercise — it is a living document that should be updated regularly as new information emerges. The discipline of maintaining and updating your model is where the real learning happens. Each quarter, when the company reports actual results, compare your forecast to reality and analyze the differences. Were your revenue assumptions too high? Did margins come in differently than expected? Did the company buy back more shares than you anticipated?

The Post-Earnings Review Process

After each earnings release, follow a systematic review process. First, record your previous forecast and the actual results in a tracking spreadsheet. Calculate the variance for each major line item. Second, identify the specific assumptions that caused the variance — was it a volume miss, a pricing issue, or a cost overrun? Third, update your forecast for the upcoming quarters based on the new information and any guidance management has provided. This post-earnings review is the quickest way to improve your modeling skills, because it exposes the specific weaknesses in your analytical framework.

When to Rethink Your Model

If your model consistently produces forecasts that are significantly different from actual results, the problem may not be your assumptions — the model structure itself may need to change. Business models evolve, and your forecasting approach must evolve with them. If a subscription-based company shifts to a usage-based pricing model, your subscriber × ARPU framework may no longer be appropriate. If a manufacturing company outsources production, the cost structure changes fundamentally. Periodically review whether your model's structure still reflects how the business actually operates. For guidance on analyzing changing business models, see our articles in the Industry Analysis section.

Building your own earnings forecast model is a skill that improves dramatically with practice. Start with one company you know well, build a simple spreadsheet, and refine it after each earnings release. For more context on leading indicators that feed into your model, see Leading Indicators to Predict Earnings. This content is educational and does not constitute financial advice.