Methodology

What this page is

ValuJack publishes a company as a card. Behind every card is an estimate of what that company is worth, a rating describing the gap between that estimate and the price, and a quality score. This page sets out how each of those is produced.

It describes the process, and the assumptions that affect how a card should be read. It does not publish the thresholds inside the filters, the formulas inside the valuation, the weights inside the scoring, or the band boundaries behind the circles. Those are the part of the method that took work to arrive at. Nothing withheld changes how you should interpret what a card says.

A one-minute animation of the steps below, using real figures.

The funnel

Five stages, from the whole market to the two cards dealt today.

The set narrows from roughly sixteen thousand listed companies to a few hundred that pass the filters, to around thirty approved in a given week, to two dealt each day.

Stage 1 — Selection

Every company is assessed on figures from its own financial reports. A common quality base applies to all six strategies, and each strategy then adds the factors that define it.

The common base

• Real profitability — a positive return on invested capital. Companies that destroy capital never enter the funnel.
• Balance-sheet safety — debt to equity capped, more tightly for the value strategies than for momentum and growth.
• Earnings reality — positive average earnings across five years. No perpetual loss-makers, whatever the story.
• Tradability — listed in an approved market and reporting in a major currency. A company nobody reading the card could actually buy is worse than no card.

Geography here is a tradability filter, not a judgement about quality. Some good markets are excluded simply because they are awkward to buy.

The six strategies

Each strategy has its own page with the original rules, the evidence, and every point where ValuJack differs: see the six strategies. The underlying approaches are the published work of their authors. ValuJack claims no ownership of them — only of its own implementation.

Stage 2 — Estimating fair value

Every surviving company is valued by a discounted cash flow model. This is the single most important number in the product: the margin of safety on every card is computed from it.

Base earnings

Five-year average earnings per share for the dividend, value and contrarian strategies, which smooths single-year distortions. Momentum and growth strategies use current earnings, because their case rests on what the company is doing now.

Margin normalisation

The hardest problem in valuing a cyclical company is that its earnings are least representative exactly when they look best. Base earnings are therefore adjusted by reference to the company's own long-run operating margin, within bounds.

A company earning far above its historical margin has its base earnings reduced; one earning below has them lifted, within limits. The correction is made company by company rather than by industry label.

Growth

Growth is not taken from any single source. Several independent estimates are blended, and every one of them is capped before it enters the blend, with the blend itself capped again. No single optimistic number can carry a valuation, and there is a ceiling on how much growth any company can be credited with however strong its recent history.

Companies in commodity-cyclical industries are additionally prevented from receiving an above-market growth uplift — otherwise the largest boost would go to precisely the companies whose earnings are already inflated by a cycle peak.

Projection

Seven years of projected cash flows discounted at 10%, plus a terminal value at 2.5% perpetual growth. A small residual reduction is applied to commodity-cyclical industries as an allowance for commodity price uncertainty; the substantive cycle correction is the margin normalisation above.

Two rules that do not bend

• The published figure is the model's own output. External estimates may be recorded for comparison but never enter the calculation. If the model disagrees with an outside source, that triggers a review, not an adjustment.
• The estimate is converted into the currency the share trades in, so the figure on the card and the price beside it are comparable.

Stage 2b — The star rating

Each company receives five criterion scores from 0 to 10, computed by fixed formulas that recalculate on every refresh. The criteria mirror each strategy's own logic, weighted so that a strategy's defining factors count for more than its supporting ones. The weighted result maps to one to five stars.

The star rating contains no price input. A company keeps its stars in a crash — which is precisely when its margin of safety improves. Stars describe the business; the rating on the card describes the price relative to the estimate. They are separate on purpose.

Stages 3 to 5 — Approval and dealing

The full scored set is loaded into the app rather than pruned to a shortlist, and the daily gates surface the eligible best continuously.

What a card must survive

• Data checks. Ticker must resolve to the company's primary listing, the currency must match the exchange, and figures must reconcile with the company's report. A card that fails is blocked, not footnoted.
• A business review. What the company actually does, how it earns, and what could go wrong — written for the card and checked against the company's own reporting.
• The Edge Check. Every trading day an AI research agent reads the latest reports, official registers of insider trades and short positions, and news for the best-ranked companies, and asks each strategy's own question: whether a dividend is safe, whether a recovery is real, whether there is news behind a trend. It can stop a card or change which company is dealt; it never changes the fair value or the stars. The six questions and the checks are on the strategies page.
• A price gate. Each strategy sets how much overvaluation it tolerates. The dividend strategy must be genuinely undervalued; value strategies accept very little overpay; momentum and growth accept more, because paying up for a leader is part of the thesis. A card with no resolved price is skipped for that day.
• Concentration limits. Cards dealt on the same day are kept apart by sector and by economic cycle exposure, so the deck does not quietly become one bet.

Cards are dealt every trading day regardless of news. ValuJack does not rush a card out because something moved.

What the circles mean

Most figures on a card are shown as a five-step circle rather than a number. The steps are ValuJack's own bands, set so they divide evenly across the companies that actually reach a card. A full explanation, including the the circle that does not mean "more is better", is on the disclosures page.

What this method does not do

Stated plainly, because a method described only by its strengths is not much of a disclosure.

• The estimate is sensitive to its assumptions. Moving the discount rate or the growth blend by a point or two changes the fair value materially — in some cases by a third. It is one view under one set of assumptions.
• It assumes reported earnings are representative. Margin normalisation corrects for cycles, not for accounting that flatters.
• It says nothing about timing. The estimates rest on multi-year cash flows. A gap between price and estimate may close slowly, or never.
• Quality scores compress at the top. Every carded company has already passed filters selecting for these qualities, so scores cluster high and some circles read full on most cards.
• It is a starting point. A card is where research begins, not where it concludes.

Changes to this page

The method changes as it is improved, and this page is updated when it does. Cards already published are not revalued to match: the card history records what each card claimed on the day it was dealt.