Credibility beats volume: the accounts that actually move the needle
By Maya Koeva · July 9, 2026
We keep saying the same thing in different articles: credibility beats volume. It is the point behind the FuelCell autopsy, the smart-money explainer, and every "loud is not strong" piece we have run. This week we wanted to put numbers behind the slogan and show what the credibility data across the accounts we track actually looks like.
Most of the crowd has never been right
We took every account we track that has built up an actual track record, about 6,000 of them, and looked at where their credibility sits. The shape is not a bell curve. It splits in two. More than four in ten of these accounts sit below 0.2, close to the floor: voices that have made plenty of calls and gotten most of them wrong. Nearly as many sit at 0.6 and above. There is very little in the middle.
The concentration of correct calls is starker still. The credible accounts, the ones at 0.6 and up, produce roughly three quarters of every correct call in that graded set. The bottom group, the 44% near the floor, produce less than one percent of them, while still doing about a sixth of all the talking. A large slice of the crowd is loud and almost never right. A smaller, quieter slice carries nearly all of the signal worth acting on. That is not a knock on anyone. It is just what happens when you grade track records instead of counting followers.
Why volume is such a bad proxy
Because credibility is this concentrated, mention volume is a genuinely bad proxy for whether a signal is real. A name can pull enormous attention almost entirely from the low-credibility majority, and on a volume chart it looks like any other busy ticker. The counts tell you how many people are talking. They tell you nothing about whether any of them have been right before. That is the exact gap a mention leaderboard cannot see.
Two names, same board, opposite reads
Credibility does not just describe accounts, it describes the conversation around a stock. Here are two names from the last few weeks that make the point, each already pulled apart in its own autopsy. For reference, the average graded account we track sits around 0.46.
| Virgin Galactic (SPCE) | FuelCell (FCEL) | |
|---|---|---|
| Mentions | 134 in 48 hours | 123 accounts, all month |
| Trusted accounts | under 1 in 4 | 94% |
| Avg credibility | 0.15 | 0.62 |
| How it resolved | +23% pop, then halved | +92% off the low |
SPCE was a two-day confusion trade off the SpaceX IPO, one of the loudest names on the entire board for 48 hours, carried by accounts averaging 0.15 credibility. It popped once and slid to half the pop. FuelCell drew a fraction of that noise, but almost everyone talking had a real record, at 0.62 average, and it ran 92% off its low. One was loud and hollow. The other was quiet and trusted. The score told them apart while a volume chart would only have pointed you at the louder, emptier one.
The takeaway
This is why the product weights signals by credibility instead of ranking them by loudness. The crowd's size is not its accuracy, and most of the crowd, among the accounts we track, has not earned the benefit of the doubt. Find the credible minority and you find the signal. Count everyone equally and you find the noise. Loud tells you where the attention is. Credible tells you whether it is worth respecting.
Quantral surfaces signals and context from public sources to support your own research. Nothing here is financial advice or a recommendation to buy or sell. Past signals are not indicative of future results.