How finscans works

How the overall assessment is formed

Two independent sources feed every assessment. The technical side reads the trend across eight timeframes, from 15 minutes to one day, and is recomputed from the raw price candles every five minutes. The fundamental side reads the sentiment of news analyses over the past 30 days, weighted by how much impact each item was judged to have.

Both are normalised to a scale from −1 to +1 and then combined. The weighting shifts with the time horizon: an intraday trade barely cares about last week's news flow, while a position held for weeks is dominated by it. The overall figure is a weighted mean of the three horizons, centred on the medium one — that is what most visitors mean when they ask how an asset looks.

When both sides point the same way, that is noted as agreement. When they point in opposite directions and both are clear enough to matter, that is flagged as a divergence — usually the most informative state a page can show, because it marks the moment where price action and news have not yet resolved into one story.

Every assessment is also checked against the price. Each horizon is compared with the move of its own window: short with the last 24 hours, medium with 7 days, long with 30 days. The overall assessment is compared with 24 hours and 7 days. If a direction runs against a strong move, meaning at least twice what is normal for this asset, the page shows "Mixed" and states the reason instead of a direction. A horizon without price data shows no direction at all, and fewer than three news reports never produce a "strong" reading.

A warning sign "Possible turn" appears when the newest data point against the situation shown, without changing it. For the news: the last three or more reports with a direction, all from the past 7 days, run against the 30-day news situation; neutral reports do not count. For the technicals: the short timeframes 15M and 30M both show a real trend (ADX 20 or more) against the trend across all timeframes, or the two or more most recent active signals all point the other way. A neutral reading never gets the sign. In the signal history, the technicals appear as the trend across all timeframes at the end of each trading day, calculated exactly like the reading at the top; the last dot is the current reading.

The outlook: 7 days, 4 weeks, 3 months

The assessment for the next 7 days rests on the trend of the 4- and 8-hour charts (60 %) and the news picture (40 %); the one for the next 4 weeks on the 12-hour and daily trend (30 %) and the news picture (70 %). Our price history does not reach far enough for 3 months, so there we show only the news picture of the last 30 days, without a statement about the price. The overall assessment also mixes in the short charts from 15 minutes to 2 hours. Everything is calculated from data, not written by a language model, and every assessment is stored so that we can later measure how often it was right.

The typical move is the asset’s normal daily move, scaled to 7 or 30 days: in about half of all such periods the price moves less than that. A move of twice that size counts as strong. If an assessment runs against such a strong move, the page shows “Mixed”. The upcoming dates come from the economic calendar for the currencies that drive the asset.

What data coverage means

Data coverage says how much material is behind an assessment: half of it is how many of the relevant timeframes have price data, the other half how many news items there were in the last 30 days (10 or more count fully). The result is good, medium or thin. On news-only pages only the news count matters: 10 or more is good, 5 or more medium.

Data coverage is not a probability. It does not say how likely a direction is, only how well the assessment is supported. How often the assessments actually turn out right is something we will show once enough time has passed.

Trend direction and strength

Each of the eight timeframes is assessed with the Average Directional Index over 14 periods. The direction comes from whether upward or downward directional movement dominates; the strength comes from the ADX value itself, mapped to five steps: below 20 means no usable trend, above 60 means an exceptionally strong one.

At strength one the bar is shown grey regardless of direction. Below an ADX of 20 the method still names a nominal direction, but "no trend" is the more honest reading — colouring it green or red would suggest more than the data supports.

Trading styles and their timeframes

The eight timeframes are grouped into four styles: scalping (15 and 30 minutes), intraday (1 and 2 hours), swing (4 and 8 hours) and position (12 hours and one day). Each card shows the assessment for its own pair of timeframes, so a page can be bullish for swing traders and directionless for position traders at the same time — which is common and worth knowing.

How hit rates are calculated

A hit rate answers one narrow question: of the signals that ran untouched to either their target or their stop, how many reached the target? Signals closed early on a reversal are excluded, because an early exit says something about position management rather than about whether the direction was right.

The population is always stated alongside the figure. Where fewer than 20 decided trades exist for a timeframe, no rate is shown at all — a percentage built on a handful of cases is noise dressed as insight. There is no fallback to other assets: a figure on a Bitcoin page is always computed from Bitcoin trades.

Past results do not predict future ones. A hit rate describes what happened, under market conditions that will not repeat identically.

How the news situation is read

Every article is analysed for which assets it concerns, whether it argues bullish, bearish or neutral for each of them, and how much impact it carries on a scale of one to ten. The fundamental score is the impact-weighted balance of those judgements over 30 days, so one major central bank decision outweighs a dozen minor mentions. Press releases that a company distributes about itself through a newswire count with half their direction: they pull the score towards neutral instead of moving it as far as independent reporting.

The asset snapshot

The snapshot is generated automatically from the coverage statistics of the past year: how many news signals were recorded, how they split by direction, the average model confidence and impact. It describes how well an asset is covered, not what it will do.