> ## Documentation Index
> Fetch the complete documentation index at: https://docs-d0.donutbrowser.ai/llms.txt
> Use this file to discover all available pages before exploring further.

# Distilling Videos and Threads

> Drop in a two-hour keynote or a KOL's latest call. D0 pulls out the parts that affect a trade, then checks whether the market already moved on them.

## The Scenario

Most trade-relevant information arrives in formats that are hostile to traders: a 98-minute keynote, a 40-tweet thread, a podcast. Watching all of it is expensive. Skipping it means missing the one paragraph that matters.

A summary alone isn't enough either. Knowing what someone said doesn't tell you whether it's tradeable; that depends on whether the market has already priced it in. D0 does both halves: extract the signal, then cross-check it against flows, order books, and equities.

## Try It

* *Pull the trade-relevant parts out of this keynote and check whether the market has already priced it in:* *(paste a link after it)*
* *What is this KOL calling, and how accurate have his past calls been?*

## What You Get Back

### A long video, reduced to what's tradeable

> **You:** 6/1 Jensen Huang COMPUTEX 2026 keynote livestream runs over an hour. Pull out the trade-relevant parts (how this AI narrative hits crypto sectors), and check whether the market has already priced it in.

> **D0:** 📹 "Jensen Huang COMPUTEX 2026 Keynote", NVIDIA Keynote (98min)\
>
> **Key announcements**
>
> * Unveiled next-gen AI acceleration architecture, big jump in inference compute
> * AI data-center capex + sovereign-AI demand keep expanding
> * Heavy focus on Physical AI / robotics and AI-agent deployment
>
> **Trade-relevant extraction**
>
> * AI narrative bullish → crypto AI sector (RENDER / TAO / FET etc.)
> * Also watch NVDA and AI-compute-chain equities
>
> **Market cross-check**
>
> * ✅ Crypto AI sector net inflows warming up over the last 7d, narrative not fully played out
> * ⚠️ NVDA firm after hours, so part of the equities side may be priced in
> * ⚠️ On-chain AI tokens highly concentrated; chasing highs is risky
>
> **Conclusion:** AI narrative leans bullish medium-term; equities have partly reacted while the crypto AI sector lags, a catch-up candidate. Light position, avoid chasing.

A plain summary would have stopped at "bullish AI narrative." The cross-check adds where that's already in the price and where it isn't.

### A KOL call, with their track record attached

> **You:** This KOL is calling a trade again. Show me what he said this time, and whether his past calls were accurate.

> **D0:** 🐦 @CryptoHayes (Arthur Hayes · \~731K followers)\
>
> **This signal**
>
> * Asset: \$HYPE
> * Direction: Long
> * Catalyst: "Hyperliquid fee-buyback flywheel + HIP-3 perp expansion; \$HYPE's value capture is underpriced"
> * Timeframe: short-to-mid (1 to 3 weeks)
>
> **Hit-rate backtest (last 90 days)**
>
> * Similar calls: 11
> * Direction correct: 7/11 (63.6%)
> * Average return: +8.2%
> * Max loss: −22% (two high-beta alt calls drew down hard)
>
> **Assessment:** Above-average hit rate overall, but high-beta names (like \$HYPE) are volatile. Small position with a 10% stop.\
>
> \[Set as monitor] \[Copy trade]

A 63.6% hit rate sounds good until you see that the misses ran to −22%. The max-loss line is reported alongside it because the hit rate alone doesn't tell you how to size.

> [!WARNING]
>
> Hit rates, average returns, and backtest windows in these examples are sample output from a specific moment, shown to illustrate the format. They are not a performance claim or a prediction. A KOL's past accuracy says nothing certain about their next call.

## When to Use

* A keynote or earnings call you can't sit through: get the tradeable minutes without the other 90.
* A KOL you're unsure about: before you act on a call, see how the last ten went.
* A thread that's already moving price: ask whether the move has run its course.
* A narrative crossing from equities or macro into crypto: this is the fastest way to find which tokens it touches.

## Tips

* Paste the link and the question together. `"<link> — is this priced in?"` gets you further than the link alone.
* Name the sector you care about. "How does this hit AI tokens" gives a sharper answer than "summarize this."
* Ask for the counter-argument. Follow up with `"what would make this wrong?"` before you size a position on it.
* Turn a call into a monitor. If a signal looks worth watching but not worth taking, ask D0 to watch it; see [Scheduled Automations](https://docs-d0.donutbrowser.ai/use-cases/scheduled-automation.md).

## Related

* [Market Analysis](https://docs-d0.donutbrowser.ai/features/market-analysis.md): the data the cross-check runs against
* [Market Research](https://docs-d0.donutbrowser.ai/use-cases/market-research.md): token-level research workflows
* [Thorp Strategy](https://docs-d0.donutbrowser.ai/features/thorp-strategy.md): how extracted sentiment feeds the signal engine
* [Acting on Thorp Signals](https://docs-d0.donutbrowser.ai/use-cases/thorp-signals.md): what happens when a signal clears the bar


This documentation is built and hosted on [Mintlify](https://mintlify.com), a developer documentation platform.