Methodology & data sources
Where every number on MarketQuill comes from and how the scores are calculated — so you can decide how much weight to give them.
Data sources and freshness
Prices can be delayed by the exchange or the provider. Every figure is shown in the currency it trades in; London, Johannesburg and Tel Aviv listings trade in pence, cents and agorot, and are labelled that way. Company accounts are often reported in another currency (TSM's in Taiwan dollars), and are converted at the live rate before they are compared with the share price.
| Data | Source |
|---|---|
| US stock quotes | Finnhub |
| Other quotes, crypto, FX | Yahoo Finance |
| Fundamentals & analyst targets | Yahoo Finance |
| Price history & technicals | Yahoo Finance |
| Company profiles & industry peers | Finnhub (Yahoo Finance outside the US) |
| Company & market news | Finnhub (Yahoo Finance outside the US) |
| Earnings dates & estimates | Finnhub (Yahoo Finance outside the US) |
| Economic release calendar | FRED, Federal Reserve Bank of St. Louis |
| Ask AI answers | xAI Grok, from the data above |
Fair value
Fair value blends up to three independent estimates. No single method is trusted: each is shown separately on the stock page, along with the ones that couldn't be calculated and why. The blend is a weighted average of whichever methods have data, with the weights rescaled to add up to 100%.
Labels. The gap between fair value and price reads as significantly undervalued at +20% or more, undervalued from +7%, fairly valued within ±7%, overvalued down to −20%, and significantly overvalued beyond that.
Confidence (0–100) rewards breadth and agreement: 55% comes from how many of the three methods could be calculated, 45% from how close together they landed. A single method can't be cross-checked, so it scores lower.
Wide disagreement. When the highest and lowest estimates are more than 60% of the blend apart, the average is an artefact rather than a view, so MarketQuill says so instead of labelling the stock: confidence is capped at 25 and you're shown the range, not a verdict.
Financial health score
A score from 0 to 10 across four equally weighted pillars. Each metric scores on a straight line between a weak and a strong value (shown below), with nothing below the weak end and full marks beyond the strong end. Each pillar averages the metrics that have data. A pillar with no data is left out rather than counted as zero, and the stock page shows what share of the 13 metrics were available.
- Net margin
- 0% → 25%
- Operating margin
- 0% → 30%
- Return on equity
- 0% → 25%
- Return on assets
- 0% → 12%
- Revenue growth
- −5% → 25%
- Earnings growth
- −10% → 30%
- Free cash flow
- positive = full marks, negative = none
- FCF margin
- 0% → 20%
- EBITDA
- positive = full marks, negative = none
- Debt / equity
- 3.0x → 0x (negative equity scores zero)
- Current ratio
- 0.8 → 2.5
- Quick ratio
- 0.5 → 1.8
- Net cash
- cash above debt = full marks
Labels: 8+ excellent, 6.5+ strong, 5+ good, 3.5+ fair, below 3.5 weak. Banks and insurers are scored on the same metrics, which suit them poorly — read their balance-sheet pillar with care.
How Ask AI is grounded
Each question is answered from a data block MarketQuill builds at that moment: the live quote, fundamentals, technicals, derived ratios, the fair value and health score above, analyst coverage, recent earnings, the latest five headlines from the past week and — if you hold the stock — your own position. The model is instructed to use only that data, to cite the figures behind each claim, and to say when the data doesn't cover something.
Fair value and the health score are calculated by MarketQuill, not by the model, so they come out the same every time. Each answer shows which data was available.
Language models can still misread or misstate a number. Check anything you would act on against the stock page.
Limits and disclaimer
Models like these are deliberately simple and use the latest reported figures: one unusual quarter can swing the growth inputs, and analysts' targets reflect their own assumptions. Data can be delayed, incomplete or wrong at the source.
MarketQuill provides information, not investment advice. Nothing on this site is a recommendation to buy, sell or hold any security. Do your own research, and consider speaking to a qualified adviser before making investment decisions.