Methodology
Every replay on FinFable is a reconstruction, not a live feed. This page documents how historical periods become interactive scenarios — where the data comes from, how it is normalized, what the indicators mean, and how written analysis is kept honest.
Candle data
- Each scenario stores one continuous OHLC series for a specific market and period. Daily series carry one candle per trading day; national property series carry one candle per month, dated from the first day of the month.
- A scenario’s sort date — the date shown in listings and timelines — is the first candle of its series. When a period starts mid-month on a monthly series, the month itself is the sortable unit.
- Series that traded in a local time zone are normalized to UTC so that events line up across markets. Where an exchange observes special trading sessions — overnight or holiday sessions — those dates are explicitly exempted from the weekday check rather than shifted or deleted.
Indicators and benchmarks
- Simple moving averages (such as the 20-period SMA) and benchmark comparisons are recomputed from the scenario’s own candle series, not sourced separately, so any number quoted in an article can be reproduced from the same dataset that draws the chart.
- Extremes quoted in analysis — largest single-day move, period high and low, volume spikes — are computed over the scenario window, never extrapolated from adjacent data.
Decision points and scoring
- Within each replay, decision points mark moments where a participant faced a meaningful choice. They are derived from the price series itself, and the replay scores decisions against what subsequently happened — simulated executions, no real capital, no order routing.
- Optimal decision zones are computed server-side for scoring and are never shipped to the page before a session is scored.
How written content is verified
- Editorial analysis is generated from a fact digest extracted from the same candle series, and every number in the output is machine-checked against that digest before publication.
- Facts that cannot be verified are marked as pending and kept out of published pages. Archive files containing pending facts carry a visible warning until each fact is resolved.
- Corrections change the underlying dataset and the affected pages at the same time; see the About page for how to report an error.