Methodology, data, and thinking.
How to value private companies, work with Czech registry data, and run a disciplined pipeline — plus the full valuation benchmark Misano works with, and essays by the founder. References carry an update date and stay current; essays carry their publish date.
Valuation multiples — the full benchmark, published
EV/EBITDA and EV/Sales bands for 82 NACE divisions and 62 curated niches — the numbers the Misano engine works with, in full, with evidence grades and machine-readable JSON.
Misano · Updated 3 July 2026Every NACE band carries an evidence grade.
Each NACE division carries an evidence quality tag (High … Low) for the strength of the underlying source data — not conviction about a target.
Misano · 18 September 2026Faulty years are dropped, never silently fixed.
A year where EBITDA exceeds revenue is excluded and flagged as provisional data — not silently corrected.
Misano · 17 September 2026Adjustments sit on the baseline, never on an override.
Heuristic shifts land on NACE or niche baselines only — a custom override stays final.
Misano · 16 September 2026Three verified niche deals before the median wins.
A niche multiple is either a median of verified deals or a labeled curated seed — never a quiet blend of both.
Misano · 15 September 2026Size and illiquidity discounts stack.
Size, illiquidity, and control-transaction basis stack — they are not one feel-based haircut on a US public multiple. Misano’s LMM SME control bands already sit on the right instrument.
Misano · 14 September 2026Damodaran DCF is a cross-check, not the headline.
Multiples set the headline. Damodaran FCFF@WACC is an independent cross-check on its own tab with a delta card — terminal g capped at Rf — not a silent blend with NAV.
Misano · 13 September 2026NAV reconciles the income value.
Multiples set the headline; book equity is the anchor. Misano compares income value to last year’s book equity on a 0.7–1.5× band, then recommends OK, anchor pull, or interval — and for asset-heavy archetypes makes NAV primary.
Misano · 12 September 2026Growth is not volatility.
One cyclical peak × a multiple is a lottery. Misano fits an OLS trend through the earnings series and treats residual noise — not the trend itself — as volatility. Above a threshold, the median goes into the multiple with a normalised (median) label.
Misano · 11 September 2026EV/EBITDA, then P/E, then EV/Revenue.
After Misano picks one source per metric family (custom → niche → NACE), a second cascade picks which metric to multiply. It is EV/EBITDA, then P/E, then EV/Revenue — one metric, not an average of three.
Misano · 10 September 2026Custom, then niche, then NACE. Never a blend.
Exactly one source prices each metric family. Misano picks custom override, then niche, then NACE — never a weighted blend of the three.
Misano · 9 September 2026Stale filings cannot make today's headline.
Old financials quietly inflate confidence. Misano's multiples engine applies a staleness gate so yesterday's EBITDA cannot silently price today's company.
Misano · 8 September 2026The value band is triangulated, not asserted.
A single number looks precise and is often wrong for SME screening. Misano prints low / mid / high from the multiples engine, uses mid as the headline, and triangulates with DCF and NAV on the football field.
Misano · 7 September 2026Hard gates can zero or cap a Score before the average runs.
Before the five-dimension average runs, hard gates can zero or cap the Score. Not operating, thin data, and sustained negative EBITDA margins take precedence over a charming composite.
Misano · 5 September 2026On Solo, Fit can move without rewriting every Score.
On Solo, a thesis move can change Fit without rewriting the Score. Watching and Passed keep prior Scores until you analyse again. Portfolio holdings can still be rescored. Retroactive Watching/Passed rescoring is Boutique and Institutional.
Misano · 4 September 2026Default Score weights are Finance-heavy. Custom ones live on the strategy.
Defaults weight Finance heaviest. Custom weights live on the strategy. Rescoring the book after a weights change is plan-gated — Boutique and Institutional yes, Solo no.
Misano · 3 September 2026Five-dimension averages are a market read, not a hurdle.
The statistics page prints five dimension averages. They are a read on the market Misano has scored, not a cut-off you have to clear.
Misano · 2 September 202699% filtered means not in diligence or portfolio.
The public statistics page says most companies are filtered out before you look. The number is live, defined, and not a marketing round-up.
Misano · 1 September 2026Fair market value is for a financial buyer.
We price the company a financial buyer would pay at the market median. Synergies of a specific strategic buyer, and the maximum in a competitive auction, are someone else's number.
Misano · 31 August 2026Upload works everywhere. Registry pull is a shortcut.
PDF, XLSX, or CSV up to 25 MB works for any company in any country. In Czechia you can paste an IČO instead and we pull the filings. You never wait for a market to be “supported.”
Misano · 30 August 2026When a last-round mark is older than 18 months.
Last-round marking is common early-stage. Misano stamps the round date. If the round is older than 18 months, it warns. A stale round is not a live mark.
Misano · 28 August 2026Strategy Fit is not the Score.
The Score answers “how strong is this company?” Strategy Fit answers “is it for you?” Mixing them is how a high-quality business that is not in your thesis still looks like a yes.
Misano · 27 August 2026Your old no is not always still a no.
Passing on a company is not the same as deleting it. Sectors move, the company grows, and your thesis moves. If the pass lives in a folder, none of that reaches you.
Misano · 26 August 2026Search reads websites. NACE codes don't.
Industry codes tell you the sector a company registered under. They do not tell you what the company actually does. Misano Search starts from a sentence (200 characters) and reads company websites.
Misano · 25 August 2026How we value companies at scale
“NACE comparables don’t really work” — and yet we build on them. The full methodology Misano runs across 100, 500 or 10,000 companies at once: six layers, each fixing the blind spot of the one before it.
Misano · 4 July 2026How to value a private company: multiples and DCF, the Damodaran way
Two methods carry almost every SME valuation: a market multiple on normalized EBITDA, and a DCF as the intrinsic cross-check. Here is the full working method — including the discounts everyone forgets.
Misano · Updated 3 July 2026Czech company registry data: what is public and how to extract it
Czech companies are unusually transparent on paper: identity, ownership, and years of financial statements sit in public registries. In practice the data is scattered, scanned, and full of format traps. A field guide.
Misano · Updated 3 July 2026Running a deal pipeline: sourcing, screening, and scoring against your thesis
The difference between a dealflow and a pile of PDFs is a pipeline with explicit stages and one scoring model applied to everything. The method, sized for a team of one to ten.
Misano · Updated 3 July 2026Valuing companies with AI: when ChatGPT is enough — and when it isn't
You can paste financials into ChatGPT or Claude and get a passable first opinion. Whether that is enough depends on what breaks: scanned filings, unit-scale traps, inconsistent scoring, and re-paying for the same documents. An honest map.
Misano · Updated 3 July 2026How I compute the Score: five dimensions, the full breakdown.
The Score is one number, 0–100. Instead of leaving it a black box, I show exactly how it's built: five dimensions, what's computed by math and what the AI judges, and the whole thing worked through on a real company.
Marek Kříž · 17 June 2026How to pay 90% less to analyse one company than in your own Claude/ChatGPT.
I wanted to drive down one number: what it costs to analyse a single company. Here, on real numbers, are the four places Misano cuts the cost of AI — and what that means for you.
Marek Kříž · 17 June 2026