We Scored 100,000 Thesis Judgments on a Month of Stock News. Half of It Was Noise.

InsightDock does one unusual thing at scale: every news article we collect for a stock gets scored by an AI model against that stock's investment theses — the specific reasons an investor might own it. Not "is this bullish or bearish," but "does this story reinforce, challenge, or say nothing about each stated reason to hold."

That produces a dataset most newsreaders never see. Over the last 30 days (July 11 to August 10, 2026), across the 63 stocks we track, the pipeline collected 48,828 articles, scored 21,825 of them against active theses, and produced 101,195 individual article-versus-thesis judgments.

Here is what that month of judgments actually says about financial news.

Finding 1: Half of stock news bears on no thesis at all

Of the 21,825 scored articles, 50.3% touched no investment thesis on any stock they mentioned — every judgment came back neutral. These aren't junk pages or spam; they're ordinary financial media: price-recap posts, "3 Stocks To Watch" listicles, institutional 13F filing notices, analyst-price-target churn.

They mention the company. They say nothing about why you own it.

If you hold a stock for its services-revenue growth or its regulatory moat, a story reporting that a mid-size asset manager trimmed its position by 2% carries no information about your thesis. Half the feed is like this. An investor reading everything their broker's news tab surfaces is spending half their attention on stories that could not, even in principle, change their mind about anything.

Finding 2: The other half is a cheerleader

Among the judgments that did bear on a thesis, the skew is stark:

• 24,784 reinforcing judgments
• 3,441 challenging judgments

That's better than 7 to 1 in favor of reinforcement. Financial news, in aggregate, agrees with the reasons people already own stocks. Some of this is structural: companies publicize their wins, coverage follows momentum, and a rising stock generates admiring explainers about exactly the theses its holders believe.

The practical inversion: challenging evidence is the scarce, information-dense event. A story that genuinely cuts against a thesis is roughly seven times rarer than one that flatters it — which is precisely why it deserves seven times the attention, and why our alerts fire on materiality-gated challenges rather than volume.

Finding 3: Challenges cluster where you'd expect — and where you wouldn't

The most-challenged theses of the month:

Stock	Thesis	Challenging stories
TSLA	Market Sentiment and Institutional Support	197
NFLX	Recent Stock Price Correction	148
AMZN	Regulatory Scrutiny Challenges	43
MSTR	Bitcoin Investment Strategy	35
META	Data Privacy Regulations Risk	32

Some of this is intuitive: risk-shaped theses (regulation, privacy) naturally attract challenging coverage. The more interesting cut is by stock: the highest challenge-density names in our universe were MicroStrategy (22% of thesis-relevant coverage challenging), Tesla (21%), and Netflix (16%) — stocks where the narrative itself is the battleground. Compare that to the typical large-cap, where challenges run well under 10% of relevant coverage. If you own a battleground stock, thesis-tracking is doing real work; the evidence genuinely swings week to week. You can watch it live on our public trackers — for example Tesla's or MicroStrategy's.

Methodology and honest caveats

Each article is scored by a language model against every active thesis for the stocks it covers, producing a relevance score, a direction (reinforces / challenges / neutral), and a short written justification. Judgments cited here use our standard relevance threshold. The universe is the 63 stocks with maintained thesis libraries — large caps skew heavy, so this says more about widely-covered names than micro caps. Thirty days is one month, not a market cycle; the reinforce-to-challenge ratio likely compresses in drawdowns (we'll re-run this when we have one). And LLM judgments are judgments, not ground truth — we audit them continuously, and the scoring rubric is the same one that drives the product's alerts, so its biases are at least consistently applied.

Why we built it this way

The two headline numbers — half the news is thesis-irrelevant, and the relevant half skews 7:1 toward comfort — are the argument for the product in two statistics. Reading more news doesn't inform a holder; it mostly burns attention and confirms priors. What changes a decision is the rare story that cuts against a stated reason to own — so that's the only thing we interrupt you for.

Every stock's live thesis tracker is free to browse, no account needed: see the full list.