> ## Documentation Index
> Fetch the complete documentation index at: https://auramarkets.mintlify.site/llms.txt
> Use this file to discover all available pages before exploring further.

# Research overview

> Substack + X + YouTube ingested daily, scored, and searchable by ticker.

The Research section of auramarkets is reachable from the **Research ▾** dropdown in the top nav:

* **[Writers](https://auramarkets.io/research)** — curated Substack, X, and YouTube content (this overview's main subject), which opens on the **[Calls](/research/calls)** scoreboard by default
* **[13F Filings](/research/thirteenf)** — quarterly 13F flows from 34 tracked funds (Consensus + By Fund views)
* **[AskAura](/research/ask)** — Q\&A agent grounded in your portfolio, the corpus, 13F, filings, and live web/X data

This page covers the Writers feed. The others have their own pages.

The Writers feed is auramarkets' answer to the firehose problem. Curated researchers, analysts, and operators publish too much across too many channels for any human to keep up. Writers collapses it into one feed.

The Writers page itself has three tabs:

* **[Calls](/research/calls)** (default) — directional calls turned into scored positions; writer track records
* **Digest** — the day's **Top items**: consensus narratives plus *dedicated pieces*, each with up to four primary tickers, an LLM **verdict pill** (bullish / bearish / neutral / informational) and a one-line takeaway
* **Browse** — the full, filterable feed (covered on this page's [browsing guide](/research/browsing))

## What gets ingested

Three sources, all daily:

| Source            | What's pulled                                                                                                                                                                                                                             |
| ----------------- | ----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- |
| **Substack**      | Articles from a curated list of investing/macro publications. Both free and paid — paid subscriptions are picked up automatically and ingested with their **full archives**, not just new posts.                                          |
| **Substack Chat** | The **subscriber chats** of paid publications — weekly AMAs, live trade discussion, portfolio updates. Each thread is ingested as a full transcript and analyzed like an article, so the alpha that only lives in chat is searchable too. |
| **X (Twitter)**   | Posts and threads from a curated list of accounts. Original posts, threads, retweets with commentary, and quote-tweets all flagged.                                                                                                       |
| **YouTube**       | Videos from a curated list of channels (in development). Eventually with timestamp-anchored claim extraction.                                                                                                                             |

Currently **600+ pieces live** across all three sources, growing daily.

## What happens to the content

Each piece of content gets:

1. **Full-text ingestion** — the actual article/post body, not just headlines
2. **Embedding** — stored in pgvector for semantic search
3. **Ticker extraction** — every \$TICKER mentioned is tagged
4. **Author + publication** — for filtering and trust scoring
5. **Confidence score** — 0–1, based on author reliability, claim specificity, and corroboration → [Confidence tiers](/research/confidence-tiers)
6. **Lifecycle status** — fresh / aging / stale (claims have shelf life)

## How you can use it

<CardGroup cols={2}>
  <Card title="See who's been right" href="/research/calls">
    Every directional call, scored vs the market. Writer track records + live positioning. The default Writers tab.
  </Card>

  <Card title="Browse the feed" href="/research/browsing">
    Sort, filter, search across all sources at once.
  </Card>

  <Card title="Look up a ticker" href="/research/entity-pages">
    Every mention of \$NVDA, in one timeline. Across all sources.
  </Card>

  <Card title="See watchlist coverage" href="/research/watchlist-matches">
    Just the research touching names you actually own or follow.
  </Card>

  <Card title="Ask a question" href="/research/ask">
    Natural-language Q\&A across the corpus, 13F, filings, and live web.
  </Card>

  <Card title="13F Filings" href="/research/thirteenf">
    What 34 tracked funds bought and sold last quarter.
  </Card>

  <Card title="Daily recap chatter" href="/macro/daily-recap">
    A mini-feed of relevant research surfaces in the Daily Recap.
  </Card>
</CardGroup>

## What it's not

* **Not a hot-takes feed** — content is filtered for substance, not engagement
* **Not realtime** — daily ingest. For minute-by-minute, use the [Market Pulse](/macro/market-pulse) right rail
* **Not financial advice** — the system is a reading-aid, nothing more

## Related infrastructure (for the curious)

* The same pipeline powers a Claude Desktop MCP server — you can ask Claude to "find all Substack articles mentioning copper supply chain in the last 30 days"
* Confidence and lifecycle scoring is a separate pass that runs after ingestion; both are tunable per source