How We Classify Stories

A plain-language look at how AIOVEL turns a day's news into sentiment tags — and the rules that keep the process honest.

What the tags actually mean

Positive, Neutral, Caution-Negative, and Negative describe how the market reacted to a story, not whether AIOVEL thinks the news is good or bad, and not what happens next. A Positive tag means the market responded favorably — a stock rallied, a sector caught a bid. A Negative tag means the opposite happened. This distinction matters because a story can be objectively bad for a company and still get a neutral-to-positive market reaction if the damage was already expected, or vice versa.

Evaluating a reaction: direction, magnitude, corroboration

Three things go into classifying a reaction. Direction is the simplest — did the relevant price move up or down. Magnitude asks how much, since a 0.3% wiggle and a 5% swing are not the same event even if they point the same way. Corroboration checks whether the move and the story behind it are backed by more than a single source, so a tag isn't built on one outlet's framing of a number that other reporting doesn't support.

The sourcing rule: two or more, or it doesn't run

Every figure or claim behind a classified story is expected to be backed by at least two independent sources. A single unconfirmed number isn't enough to hang a sentiment tag on. This is a deliberately conservative bar — it slows things down in exchange for not building a classification on a report that later turns out to be wrong or overstated.

When the data isn't there: null, not a guess

If a figure can't be sourced to the required standard, or a data point simply isn't available, AIOVEL shows a dash — — — rather than filling the gap with an estimate. A missing number stays visibly missing. That's a small design choice with a real purpose: it's easy to make a dashboard look complete by quietly interpolating gaps, and the cost of doing that is that users can no longer tell real data from filled-in data.