How Zortix counts
Zortix is the public index of record for what finance podcasts are saying about stocks: it transcribes the shows, extracts every attributed investment idea, and publishes the statistics — mention counts, discussion depth, top shows — while the app is where you read, hear and track the ideas behind the numbers.
Every number on a Zortix page comes from the same pipeline: episodes are transcribed with timestamps, an extraction step identifies each specific investment argument made on the show, tickers are fact-checked against a curated registry (so a host saying “apple” the fruit never becomes AAPL), and the results are quality-checked and de-duplicated before anything is counted.
How are the shows selected?
Zortix monitors a curated roster of finance podcasts — shows with a regular cadence whose episodes carry substantive investment discussion rather than headlines alone. The roster is public: every monitored show has its own page with per-show coverage statistics, and it grows deliberately rather than automatically.
How does transcription work?
Every episode is machine-transcribed with timestamps, so each sentence stays anchored to the second it was spoken. The transcript is the working material for extraction — and the anchor that lets a reader in the app verify any idea against the original audio.
What is an attributed idea?
A specific argument a named speaker made on a named show, tied to the episode it came from — never an aggregate opinion, never Zortix’s view. Ideas that fail quality control, duplicate an existing idea, or have been superseded are excluded from every public statistic. In the app, each idea carries its decoded thesis and links to the exact moment in the episode.
What counts as a mention?
One attributed idea naming a ticker. “Mentions · 30d” is the count of attributed ideas about that name extracted from episodes that aired in the last 30 days — always the episode’s air date, never the date Zortix processed it, so working through a show’s archive can never make an old conversation look current. An idea that names several tickers counts once toward each.
How are the statistics windowed?
Ticker statistics are counted over rolling 7-, 30- and 90-day windows of episode air dates. “Vs prior week” compares the last 7 days with the 7 days before that; the weekly Signal counts ISO calendar weeks and compares each week with the one before it. Every page states the date its numbers are current to.
What do the discussion-depth bands mean?
How thoroughly an idea was argued on the episode — a description of the conversation, not a rating of the stock. Each idea is graded from a passing remark to a fully argued core thesis, and public pages show the resulting band as a word: Brief mentions, Notable commentary, Substantive discussion, In-depth coverage, or Core-thesis conviction.
- Brief mentions
- The name came up in passing; no argument was developed.
- Notable commentary
- A short but substantive comment — a view stated with at least one reason.
- Substantive discussion
- A developed segment: the argument, some evidence, a risk or counterpoint.
- In-depth coverage
- An extended discussion that examined the thesis from several angles.
- Core-thesis conviction
- The idea was a centrepiece of the episode, argued at length.
How is disagreement between shows handled?
It is kept, not averaged away. Where shows or speakers take opposing views on a name, both sides are extracted and attributed, and nothing combines them into a verdict — a count that grew is a fact, a “winner” would be an opinion Zortix does not issue.
How current is the index?
Episodes are processed as they publish; ticker pages carry an as-of date, the Signal publishes weekly at a permanent URL per edition, and the daily edition surfaces one lead idea each morning at a dated permalink. If nothing qualifying aired, an edition is skipped rather than padded.
What are the known limitations?
Machine transcription can mishear; speaker identification is best-effort; extraction can occasionally miss an idea or scope one imperfectly; and coverage extends only to the monitored roster. That is why every idea links back to its source — the recording, not Zortix, is the authority. Verify at the source before relying on any claim.
How does Zortix handle rights to the shows’ content?
Zortix publishes statistics and short attributed excerpts, always linked to the original episode, and sends listeners to the shows rather than replacing them. Rights holders with questions, corrections or removal requests can reach us at hello@zortix.app.
Does Zortix rate or recommend stocks?
No. No ratings, no scores presented as investment merit, no price targets, no buy or sell language, and no aggregation that turns attributed speech into a verdict. Mention counts and depth bands describe how much and how thoroughly the shows talked — research signals about a conversation, not advice about a security. Zortix is a research and education tool; verify anything you read here before any investment decision.