The calculation rules behind the Annolyse principles pages: thresholds, categories, denominator guardrails, exclusions, source timestamps, and current limits.
Rankings as of 08-06-2026 9:21pm NZT, based on Annolyse's coverage of 100 NZX companies.
Annolyse combines three comparable-period inputs: revenue direction, NPAT direction, and cash-flow direction. Revenue uses a 2% up/down band. NPAT and cash flow use a 5% up/down band.
At least two of the three inputs must resolve. If fewer than two resolve, the row is marked insufficient data. Sign-flips around zero are treated as unavailable rather than forced into large percentage changes.
Cash-flow direction uses FCF where available and operating cash flow where FCF is not disclosed clearly enough for the issuer.
Price direction uses the latest 12-month share-price change from the market-data snapshot. Moves inside +/-5% are flat; moves above the band are up; moves below the negative band are down.
The price direction is shown separately from fundamentals so readers can distinguish a business result from the market's response to it.
Opportunity signal means fundamentals improved while price fell. Aligned favourable means fundamentals improved and price was flat or up. Stable means fundamentals and price were both flat.
Decoupled upside and decoupled downside mean fundamentals were stable while price moved outside the flat band. Aligned unfavourable means fundamentals deteriorated and price was flat or down. Warning signal means fundamentals deteriorated while price rose.
Default sorting puts opportunity signals first, then favourable/stable/decoupled cases, with warning signals and insufficient data at the bottom.
Price volatility is the 52-week high-to-low range divided by the midpoint. Fundamental volatility is the larger absolute movement of revenue change or EBITDA-equivalent change over the comparable period.
Fundamental volatility treats sign-flips and changes above 1000% as not meaningful because those cases are usually denominator effects rather than usable business-volatility signals.
Noise ratio is price volatility divided by fundamental volatility. It is suppressed as n/m when fundamental volatility is below 2% because the denominator is too small to be analytically meaningful.
Displayed noise-ratio values are capped at >100x so denominator effects do not render as broken-looking six-figure percentages.
Euphoric Mr. Market means fundamentals were stable while the 12-month price move was at least +20%. Panicking Mr. Market means fundamentals were stable while the 12-month price move was at least -20%.
Underreacting market means fundamentals moved while price stayed flat. Aligned movement covers rows where price and fundamentals broadly moved together or where the move is not clean enough for the emotional Mr. Market categories.
The category is deliberately separate from the noise ratio: a suppressed ratio can still have useful directional context.
Listed investment vehicles are excluded from the principle rankings because operating cash flow, earnings yield, and FCF yield metrics work differently for those structures. Current exclusions: AFI, BAI, BRM, HFL, KFL, MLN, TEM.
Newly listed companies or companies without enough comparable data are retained only where the inputs support the calculation; otherwise they are marked insufficient data.
Earnings yield is trailing twelve-month NPAT divided by market capitalisation. FCF yield is trailing twelve-month pre-lease free cash flow divided by market capitalisation. Full-year rows use the latest FY result; half-year rows use latest HY plus the prior second half where the prior FY and HY are available. Negative or unavailable earnings and FCF are suppressed as not meaningful.
The default net-buyer table uses the 365-day primary lookback. Historical trailing fundamentals come from the latest Annolyse briefing available at or before that lookback anchor date.
Historical valuation rows store normalized lookback records, not only one set of 12-month fields. The public tables keep the 365-day primary lookback while the underlying history can support shorter or longer lookbacks.
The historical valuation layer also writes a price-observation store from each market-data snapshot. Lookbacks first use stored price and market-cap observations near the anchor date; the 365-day lookback can fall back to the provider's 12-month price-change field while the observation archive builds up.
Extreme yields above 30% remain visible rather than capped because they can be genuine distressed-pricing signals. Treat them as prompts to inspect the company's recent trajectory, not as automatic opportunity signals.
The historical valuation layer keeps the latest snapshot, a price-observation store, and an archive index. The default archive cadence is monthly.
Yield changes inside +/-50bps are stable. Yield improvement with stable or improving fundamentals is net-buyer favourable; yield improvement with deteriorating fundamentals is net-buyer caution.
Running examples are refreshed from the same principles snapshot that drives the ranked tables. Each principle ranks eligible rows by how cleanly they illustrate the category, then takes the strongest current examples.
Principles pages are refreshed after a briefing is published, so the examples move with newly published data instead of waiting for a manual content pass.
Commentary is prepared from the stored table metrics and category labels only. It deliberately avoids target prices, forecasts, or buy/sell language.
Principles rows can show same-reporting-shape historical context from the Annolyse historical-observation layer. Only preserved outlier classifications are surfaced: below normal range, above normal range, unprecedented low, and unprecedented high.
Normal-range, edge-of-range, insufficient-history, and not-applicable observations are suppressed on principles tables. Absence of a historical-context note does not mean a value is normal; it may mean there is not enough comparable history or the observation is not relevant to that principle.
Historical context is evidence only. It does not change the category definitions, does not alter default ranking order, and is not a recommendation.
Principles snapshots are prepared from the latest Annolyse company coverage plus the latest market-data snapshot. Market-data refreshes and newly published briefings both refresh the principles snapshot and running examples. Each ranking page displays the snapshot timestamp near the table.
The pages show analytical categories, not recommendations. A company illustrating a principle is not a buy, sell, hold, target price, or forecast.