How Often Do Four Independent Reads of a Stock Agree?

Across 41,009 daily readings on 581 US stocks, taken between 22 April and 22 September 2026, all four of our data layers pointed the same way on 19.9% of them. Roughly one reading in five. Nearly all of that was agreement on the positive side: 19.7% of readings had all four layers above zero, and only 93 readings, 0.23%, had all four below it.

The number beside it matters more. If the four layers had nothing to do with each other, and each simply leaned positive as often as it does, chance alone would put all four on the positive side 19.8% of the time. We measured 19.7%. Four separate reads of the same stock line up about as often as four unconnected coin tosses, each weighted the way that layer usually falls.

That is not a defect in the measurement. It is the finding, and it says something uncomfortable about how the word consensus gets used.

What we counted

Every trading day each covered stock gets four readings, each on a scale from -100 to +100: price structure, news tone, analyst coverage and reported fundamentals. The categories and the sources behind them are set out on the methodology page.

For this piece we took every reading from 22 April 2026, the first scored day, to 22 September 2026, and kept only readings where all four layers were present. Funds were left out. A fund is scored on price structure alone, so it would count as agreement on days when nothing else was measured at all. Two stray weekend runs were dropped too, which leaves 106 trading days.

Direction here means the sign of each layer. A layer sitting at exactly zero is its own class and is not folded into either side, so "the same way" means all four above zero or all four below it.

The full distribution

• All four the same way: 19.9% (19.7% positive, 0.23% negative)
• Three one way, the fourth not: 53.1%
• Two against two: 17.6%
• Everything else: 9.4%, all of it readings where at least one layer sat at exactly zero and no three matched

The most common single state is not agreement. It is three positive layers with fundamentals as the odd one out, which accounts for 37.0% of every reading in the set. The next two are a long way behind: three positive with price structure dissenting (6.5%), and three positive with news tone dissenting (5.3%).

Why agreement looks like chance

The same pattern holds pair by pair. Here is how often each pair of layers pointed the same way, against what their individual leanings would produce if they were unrelated:

• News and analysts: 75.8%, against 71.8% by chance
• Price structure and analysts: 72.4%, against 72.2%
• Price structure and news: 63.4%, against 61.9%
• Price structure and fundamentals: 40.0%, against 41.8%
• Analysts and fundamentals: 38.3%, against 36.5%
• News and fundamentals: 37.4%, against 35.9%

The correlations between the underlying values say the same thing. The highest is between price structure and news, at 0.12. News and analysts come next at 0.09. Every other pair is within 0.02 of zero.

So a sentence like "analysts and the chart agree 72% of the time" describes a coincidence of base rates, not a relationship. The analyst layer was positive on 92.4% of readings and price structure on 76.9%, so the two land on the same side most days whether or not they are responding to the same thing. The one pair that agrees clearly more often than chance is news and analysts, and the only two correlations visibly above zero both involve news. That fits how news works: headlines report price moves and analyst notes, so the news layer partly echoes the other two.

What that means for consensus

We wrote about what market consensus actually means before, and argued that the spread around an average often says more than the average. This data pushes the point further.

If four sources are close to independent, then agreement among them is mostly a statement about how each one leans on a normal day. In a period when three of the four leaned positive most of the time, broad bullish agreement is the default setting, not a discovery. A stock where everything lines up positive is a stock in the most ordinary state this market had.

Our position is that the useful information sits in the departures. Which layer dissents, and is it the usual dissenter? Fundamentals breaking from three positive layers is the resting state here, more than a third of all readings. Price structure breaking from three positive layers is rarer, one reading in fifteen, and it describes a different situation: the professionals and the headlines still lean one way while the chart has turned. Neither tells anyone what happens next. They tell you which kind of disagreement you are looking at, and that is a more precise description than "consensus is positive".

What divergence does with this

Counting signs is a blunt tool. A layer at +2 and a layer at +90 both count as positive, and two layers can share a sign while sitting far apart. The divergence score exists to measure the size of the disagreement rather than just its direction.

It behaves the way the sign counts suggest. Average divergence across the whole set was 34.5 on its 0 to 100 scale. On readings where all four layers were positive it averaged 18.5. On readings split two against two it averaged 44.1. The divergence board ranks covered stocks by that number every day, from the widest disagreement to the closest agreement, which is the practical version of everything above.

Limits

• Five months, one market. The window covers a period when the composite Bias Score sat above zero on 87% of readings. A falling market could produce very different base rates, and with them a different chance level for agreement.
• A growing universe. Coverage grew from roughly 200 scored stocks a day in April to around 560 in September, so later months carry more weight in every percentage here.
• Direction only. Most figures above count signs, not sizes. That is why the divergence score, which does account for size, is quoted alongside.
• Descriptive. Nothing here says anything about what prices did afterwards. The reasons are in a score is not a prediction.

The exact queries behind every figure are kept with the article source, so each number can be rerun and checked.

EquityBias is a market data research tool. Nothing here is financial advice. These figures describe how our data layers behaved over a fixed past period, not predictions about any stock or any future period.