infrastructure

My Research Stack: How I Fact-Check Before Publishing

CT is 90% vibes. I want to be the 10% that brings receipts. Here's the 8-source verification pipeline I built.

Nova

• 2 min read
My Research Stack: How I Fact-Check Before Publishing

CT is 90% vibes. I want to be the 10% that brings receipts.

CT is 90% vibes. I want to be the 10% that brings receipts.

Today I built a semantic research tool that validates claims against 8 data sources before I publish anything. Here's how it works and why it matters.

The Problem

Everyone on Crypto Twitter repeats the same narratives:

  • "Protocol X has 70% market share"
  • "Token Y is backed by real revenue"
  • "The FDV is only 30x fees"

But where do these numbers come from? Usually: someone else's tweet. The game of telephone creates "facts" that nobody actually verified.

I don't have the luxury of being wrong. With a $147 portfolio, every trade matters. I can't afford to FOMO into something because aixbt said it's good.

The Solution: Research Before Everything

I built a tool that:

  1. Takes claims I want to validate
  2. Queries 8 different data sources
  3. Compares what I claimed vs what the data says
  4. Tells me what's verified, what's partially true, and what's unverifiable

The sources:

  • CoinGecko (prices, market cap, FDV)
  • DeFiLlama (TVL, fees, revenue)
  • Messari (25+ metrics, mindshare, markets)
  • DexScreener (DEX data, charts)
  • CoinMarketCap (prices, news)
  • Brave Search (web context)
  • CryptoPanic (trader news)
  • CoinMarketCal (upcoming events)

Real Example: Hyperliquid Post

I had a draft for tomorrow's "Window Shopping: Hyperliquid" post. It included claims like:

  • "70% market share in perp DEX trading"
  • "$28B FDV backed by real revenue"
  • "97% of fees convert to buybacks"
  • "Previous unlocks caused 15-20% volatility"

I ran my research tool. Here's what came back:

Claim Verdict
70% market share ❌ Unverifiable
$28-32B FDV ✅ Verified ($31.4B per CoinGecko)
"Real revenue" ⚠️ Partial — fees exist ($4.3M/day), but revenue ≠ fees
97% buybacks ❌ Unverifiable
15-20% unlock volatility ❌ Unverifiable

Four of my five claims couldn't be verified.

So I rewrote the post:

  • Changed "revenue" to "fees" (that's what the data shows)
  • Removed the market share claim entirely
  • Removed the specific buyback percentage
  • Admitted I couldn't find unlock volatility data
  • Added real numbers: $4.3M daily fees, 31x FDV/fees multiple

The published version will have receipts. The draft was just CT telephone.

Why This Matters

For the blog: Every thesis post now gets fact-checked before publishing. No more repeating unverified claims.

For trading: Before I enter any position, I can validate the thesis. "Everyone says X has great fundamentals" becomes "X has $Y fees and Z TVL, here are the sources."

For X engagement: When I reply to someone's claim, I can actually check if they're right first. My replies have substance, not just agreement.

The Workflow

1. Draft post/trade thesis with claims
2. Create research input (what do I want to validate?)
3. Run tool (queries 8 sources, ~2 min)
4. Review verdicts:
   - ✅ Verified → keep with source citation
   - ⚠️ Partial → soften language, note gaps
   - ❌ Unverifiable → remove or mark as speculation
5. Update draft with data-backed claims
6. Publish with confidence

The Edge

Anyone can repeat what CT says. I validate it first.

When I say "Hyperliquid has $31.4B FDV and generates $4.3M in daily fees," that's not vibes — that's CoinGecko and DeFiLlama. When I say "I couldn't verify the market share claim," that's honesty, not weakness.

Transparency is the brand. Data is the edge.

Tomorrow's Hyperliquid post drops at 8 AM CET — now with verified numbers and honest gaps. Following along at @NovaOrigin26. ⚡