Skip to main content

Projecting Free Cash Flows

Updated Jun 24, 2026 at 2:35pm

Research Draft High 1,217 words

Projecting free cash flows is the first and most consequential stage of a discounted cash flow (DCF) valuation: building a multi-year, line-by-line forecast of the cash a business will throw off, before financing. It is where the analyst translates a narrative about a company — how fast it grows, how profitably, how much it must reinvest — into explicit numbers. The core tension is that these projections drive the bulk of the valuation yet are forecasts of an uncertain future; small, plausible-sounding changes in growth or margin assumptions compound into large swings in estimated value. The discipline of good FCF projection is therefore less about Excel mechanics than about internal consistency: every dollar of growth must be paid for with a dollar of reinvestment.

How it's calculated / formed

Most DCFs project unlevered free cash flow (free cash flow to the firm, FCFF) — cash available to all capital providers before interest, so it pairs with the WACC discount rate. The standard build, used in IB/equity-research training, is:

UFCF = EBIT × (1 − tax rate) + D&A − CapEx − Increase in net working capital

EBIT × (1 − tax) is NOPAT, the firm's after-tax operating profit as if it had no debt. Interest is deliberately excluded so the measure is capital-structure neutral — independent of how the firm chose to finance itself (Wall Street Prep; Macabacus). D&A is added back as a non-cash charge; CapEx and the change in working capital are subtracted as real cash reinvestment.

The projection is built top-down from drivers, not as a single growth rate:

  • Revenue — decomposed into volume × price, or by segment/geography, anchored to unit economics, company guidance, and industry data.
  • Operating margin — gross and operating margins modeled separately, linked to cost structure, scale effects, and competition, then applied to revenue to get EBIT.
  • Reinvestment — CapEx, D&A, and working-capital change, frequently forecast as a percent of sales when item-level detail is unavailable.

Damodaran's framing makes the consistency requirement explicit: growth is earned, not assumed — growth = reinvestment rate × return on capital, and the sales-to-capital ratio translates each dollar of incremental revenue into the capital needed to produce it. A model that forecasts rapid growth with little reinvestment is internally broken (Damodaran, NYU Stern lecture notes).

The forecast covers an explicit period — commonly 5 years for stable firms and up to 10 for high-growth firms (Wall Street Prep; Street of Walls). Cash flows beyond it are captured by a separate terminal value (a sibling node).

How it's used in practice

The projected FCF stream is the numerator of the DCF: each year's cash flow is discounted to present value at the WACC and summed, then added to the discounted terminal value. In practice analysts:

  • Build from the income statement down — revenue → margins → NOPAT → add D&A → subtract CapEx and ΔNWC. Tax rate is often seeded from the trailing-twelve-month effective rate, then trended toward a marginal/statutory rate.
  • Taper assumptions toward maturity — growth and margins should converge over the forecast so that, by the final explicit year, the business resembles a stable, terminal-state company. Terminal growth must not exceed long-run nominal GDP (typically ~2–3%), or the firm implausibly grows faster than the economy forever (Wall Street Prep).
  • Run reverse DCFs — back-solve from today's market price to find the growth/margin assumptions the market is implying, then judge whether those are achievable. This sidesteps much of the forecasting burden and is a favored sanity check.

Adoption, debate & evidence

DCF is the standard intrinsic-valuation method taught in finance and used across investment banking, equity research, and corporate finance; it is uncontested as a framework. The debate is entirely about the reliability of the inputs, and FCF projection is the largest source of error.

The empirical record on forecasting is sobering. Peer-reviewed work on analysts' long-run earnings-growth forecasts finds them, on average, inaccurate and optimistically biased, with limited usefulness for valuation; analysts overestimated growth for most firms over 1981–2011, with the bias strongest among large and growth firms (research summarized via ScienceDirect / ResearchGate). Notably, firms with the most optimistic growth forecasts have tended to underperform — the opposite of what the forecasts imply. Documented causes include incentive conflicts and anticipation of managers' earnings management. The practical lesson: projections built by extrapolating recent high growth are precisely where DCF goes most wrong.

"Folklore vs. measured": the folklore is that a careful 5-year model produces a precise fair value. The measured reality is that beyond ~2 years, point forecasts have weak accuracy, which is why disciplined practitioners treat a DCF output as a range (via sensitivity/scenario analysis) rather than a single number, and lean on reverse-DCF and reinvestment-consistency checks.

Strengths & limitations

Strengths. Forces an explicit, falsifiable economic story; makes value depend on cash, not accounting earnings or sentiment; and surfaces the reinvestment cost of growth that multiples hide. It is the right tool for businesses with reasonably forecastable cash flows.

Limitations. Highly sensitive to assumptions — value is dominated by hard-to-forecast revenue growth, margins, and the terminal value, the latter often >60–80% of total value in growth cases. It is unreliable for early-stage/negative-FCF firms, deeply cyclical businesses, and financials (where FCFF is ill-defined).

The #1 misuse: assuming growth without funding it — high revenue growth paired with low reinvestment, which inflates value through a logically impossible free lunch. The discipline of growth = reinvestment × return on capital exists precisely to prevent this. A close second is "reverse-engineering" a forecast to justify a price already decided on.

Sources

Dispute flagged: DCF as a framework is universally accepted; the reliability of multi-year FCF forecasts is genuinely contested by the academic forecast-accuracy literature. This doc treats that limitation as the central honest caveat, not a footnote.