Skip to content

Release notes

What changes for users in each version of Fairmark, in plain language. The technical history of every change is in the changelog. On the documentation site, the first section lists the changes not yet released.

Fairmark Unreleased

Action required

  • The DCF on free cash flows to the firm takes a new input. The single list of flows, discount rate, terminal growth and net debt are replaced by explicit stages, the WACC (one rate, or one per year with a terminal rate), the terms of the flows and of the rates, a required timing, the terminal technique and the net financial position with its components; what was called enterprise value (the present values only) is now the value of operations. The published examples give the same numbers in the new form. (#110)

Changes to results

  • Discounting at negative rates is now accurate to the last digit, and very long series no longer fail. Results of net present value and of the DCF with free cash flows can differ only in the last digits, and only when the discount rate is negative; all published examples give the same values as before. Both methods are now at version 2. (#95)

New

  • Fairmark now works with every major model provider: Anthropic, OpenAI, Google Gemini, OpenRouter and LiteLLM, and with open-weight models running on your own computer through Ollama, vLLM or LM Studio, so a valuation can stay entirely on premises. (#55)
  • Fairmark now works with dates and with rates that change over time. You can value dated cash flows from any valuation date (XNPV and XIRR, matching the spreadsheet standard), choose the day-count convention (Actual/365, Actual/360, Actual/Actual ISDA, 30/360, 30E/360) and the compounding, discount each year of a plan at its own rate, and compute MIRR, annuities and perpetuities. Every result states the conventions it used. (#95)
  • Fairmark now computes the discount rate from its components: cost of equity with the CAPM and explicit country, size and specific premia, unlevered and relevered betas (Hamada or Harris-Pringle), the beta of a group of comparable companies, the cost of debt and the WACC. Every premium and every choice appears in the result, and a specific risk premium must come with its written justification. (#99)
  • Goal seek on any method: find the value of one input that makes one output equal a target, for example the discount rate at which the DCF enterprise value equals an offer price, or the rate at which a net present value is zero. The search runs in a range you give, reports every evaluation in the audit trail, and stops with a clear error when the target is not in the range or cannot be reached. (#108)
  • The DCF now models a business plan in up to three stages: years of given flows, years of constant growth, and years in which growth declines in equal steps to its long-term rate. Each year shows its flow, growth, rate, discount factor and present value; rates can change every year; a plan in real terms discounted at nominal rates is refused instead of silently mixed; and the result moves from the value of operations to enterprise value and equity value with every item listed. (#110)
  • Financial ratios organised by the question they answer: profitability (ROE, ROI, ROA, ROS, margins, capital turnover, and the decomposition of ROE in three factors and in the leverage form with the leverage effect and the effective tax rate), capital structure, liquidity, working capital days and debt service (interest cover, net financial position to EBITDA, DSCR per period and LLCR, with an optional debt service reserve account). Closing or average balances and the 360 or 365-day basis are explicit choices, the net financial position is given with its composition, and a ratio that cannot be computed or is not meaningful, such as ROE with negative equity or a leverage multiple on negative EBITDA, is reported as not defined with the reason while the other ratios are still computed. (#114)
  • Terminal value as a family of explicit methods: stable growth (Gordon), growth earned by reinvestment at a given return on new capital, the H-model for declining growth, exit multiples with the growth they imply, the Italian finite-horizon and normalised-income forms, and liquidation value with every line shown. Each result states its assumptions and whether the growth respects a cap you set, and a comparison runs several techniques on the same rate and basis side by side, so the effect of the choice of technique on the valuation is visible. (#115)
  • Sensitivity tables, scenarios and tornado charts on any method. A one-way or two-way table shows how a result moves across a range of assumptions, for example the enterprise value across WACC and growth, with an impossible combination marked in its cell instead of stopping the table. Scenarios compare named cases, such as downside and upside, with the base case. A tornado ranks which assumptions move a result the most. Every evaluation is in the audit trail. (#116)
  • Fairmark now computes the constant instalment of a loan (with an optional balloon amount and an explicit rounding to the cent), the number of instalments needed to repay a loan with the amount of the last instalment, and the future value of a present amount and regular payments. The rate can be given as the annual borrowing rate with its compounding and payment frequency, or directly per payment period, and every convention is listed in the result. (#136)
  • Fairmark now builds the full amortisation schedule of a loan: for every period the instalment, the interest portion, the principal portion and the outstanding balance, with the totals. It covers constant-instalment (French), constant-principal (Italian-style) and bullet loans, an initial interest-only period and a balloon amount. Instalments can be rounded to the cent as banks do: the last instalment then absorbs the rounding so that the balance ends exactly at zero, and the adjustment is reported. (#137)
  • Covenant tests on the definitions of the loan agreement. For each covenant you give the numerator and the denominator as the agreement defines them, the threshold, whether it is a maximum or a minimum and whether the limit itself is allowed, and the test date. The rules that change from one agreement to another are explicit choices, with no default: what happens when the denominator is zero or negative (breach, pass or not tested), and, when the agreement says so, the rounding of the ratio before the comparison, for example to two decimals. Fairmark returns for each covenant whether it is passed, breached or not tested with the reason, the tested value and the headroom, in the unit of the ratio and as a share of the threshold, positive when the covenant is respected and negative when it is breached. The outcome is decided exactly, so a ratio that differs from the threshold only in the last digits is never misclassified. (#147)
  • Automation platforms such as n8n can now use Fairmark directly: the MCP server can run over HTTP, so an n8n AI Agent can call every valuation method as a tool.
  • The documentation is now available in English and Italian.

Improved

  • The internal rate of return no longer returns an arbitrary answer when a cash flow profile has more than one valid rate: Fairmark now reports the ambiguity, with the intervals where the rates lie, so that the analyst can choose a different measure such as MIRR. (#95)
  • The input definition of every method version is now kept, one file per version, with an index recording the release that introduced it and a fingerprint of its content. An integration can show which inputs a given method version accepted. Automatic checks refuse breaking changes to a version already published. (#103)
  • Fairmark now states clearly what it is and what it is not: a calculation tool, not an advisor. Every decision taken with its results belongs to the user, who verifies them before relying on them. The full disclaimer is in the documentation, in English and Italian, and AI agents connected to Fairmark are instructed to present results as calculations, never as advice. (#171)