Algorithmic methodology
How the quantitative research that ZYXmon publishes under the MAR regime is computed.
Summary
ZYXmon produces per-instrument quantitative classifications by means of a deterministic synthesis of canonical signals computed from public data. The methodology is universal: the same instrument receives the same classification for every user at the same point in time. This page describes the inputs, the combination logic, the hysteresis bands and the known limitations.
Canonical signals (inputs)
Each classification is computed from the most recent state of the following canonical signals, persisted in a shared database (single writer, many readers):
- Overall score (overall_score) on a 0–100 scale, a normalised combination of the following pillars.
- Fair value (label and upside %): a Discounted Cash Flow (DCF) model, sector-relative multiples, and third-party analyst consensus, combined into a weighted consensus.
- Financial health (Piotroski F-Score, leverage ratio, interest coverage, free-cash-flow dividend coverage).
- Dividend safety (category: Safe / Watch / At Risk / Dangerous) based on payment history, free-cash-flow payout, prior cuts and coverage.
- Technical recommendation derived from RSI(14), MACD, moving averages (SMA 50/200), Aroon and chart patterns where applicable.
- Historical percentiles: position of the current price relative to its 52-week range and to its Average True Range (ATR).
- Instrument archetype (REIT, utility, bank, ETF or general) which adjusts thresholds to avoid apples-to-pears comparisons.
Synthesis logic
The synthesis is deterministic: the same inputs at the same point in time always produce the same output. No random component is involved.
Signals are combined through a weighted composition with discrete thresholds. Each signal contributes points to its corresponding pillar (valuation, quality, dividend, technical, risk) and the aggregate maps to a qualitative classification via calibrated thresholds.
Weights and thresholds are administrable parameters (they are not hard-coded in the binary): they are stored in a configuration table and are auditable. Every change is recorded with a date stamp.
Hysteresis bands
To reduce noise from day-to-day swings in price or input data, category transitions incorporate a hysteresis band: once a category is assigned, it only changes when the new score crosses the boundary by a configurable minimum margin. The default behaviour on a first computation (no previous category) is equivalent to a sharp classification with no hysteresis.
Hysteresis applies identically to every user and every instrument; it is not a user-tunable parameter.
Time horizon
The implicit time horizon of the algorithmic classifications is 1 to 3 months, unless expressly indicated otherwise. Beyond that horizon, the underlying information may have changed enough to invalidate the classification.
Refresh cadence
Recomputation runs overnight between 00:00 and 06:00 UTC. Every output carries a generation timestamp and a planned next-update timestamp.
Known limitations
Quantitative models have inherent limitations. The most relevant ones are documented openly:
- Inputs come from third-party financial data providers. Errors, delays or inaccuracies in that data propagate into the classifications.
- DCF models depend on assumptions about future growth and discount rates. For companies with very volatile earnings or cash flow, results may have wide variance.
- Historical percentiles assume structural stability of the quotation regime; deep changes in business model or regulation can make historical data less representative.
- Backtests of the historical behaviour of the methodology over past periods do not guarantee future results. Past performance is not indicative.
- Instrument coverage may not be complete for every market or for very illiquid tickers; some signals may be missing for such instruments.
Currency conversion (FX)
Quotes are stored in the trading currency of the instrument. Conversions to the user's base currency happen at query time using exchange rates from central banks (ECB) and public providers. Typical FX freshness is under 24 hours; on closed markets or low-liquidity pairs it can stretch to 48–72 hours. ZYXmon does not apply any irreversible currency conversion at ingestion.
Governance and oversight
The methodology is reviewed by the designated editorial responsible, Hugo Emilio Blanco, on behalf of Sixe Iberica SL. Any material change to weights, thresholds or signals must be approved and is recorded in the changelog at the end of this page.
A complaints channel is available at legal@zyxmon.com. Complaints are answered in writing within a maximum of 30 calendar days.
Changelog
Material changes to the methodology:
- 2026-05 — Initial publication of this methodology page, aligned with the MAR regime (Regulation (EU) 596/2014) and Commission Delegated Regulation (EU) 2016/958.
Last updated: May 2026. Version 1.0.