Example: Treasury Yields vs VIX Volatility

This example shows a real research run on the Lomita platform. The entire process — from question to published report — ran autonomously.

The question

"Do rising 10-year Treasury yields predict VIX volatility changes?"

What happened

research("Does rising 10-year Treasury yields predict VIX volatility changes?")

The research team executed automatically:

Found FRED DGS10 (10-Year Treasury) and FRED VIXCLS (CBOE Volatility Index) — 9,084 daily observations from 1990 to 2026.
Verified data freshness and confirmed both series were aligned.
Ran 7 statistical tests:
- Pearson correlation (levels and changes)
- 2-week lag analysis (the critical test)
- Linear regression
- Regime analysis (high-yield vs low-yield environments)
- Structural break test (pre-2008 vs post-2008)
Compiled the findings into an executive report.

The result

REFUTED. The 2-week lag correlation is r = +0.009 — effectively zero.

Finding	Result	Confidence
Pearson correlation (levels)	r = -0.085, R² = 0.007	HIGH
Change-on-change	r = -0.157, R² = 0.025	HIGH
2-week lag (critical test)	r = +0.009	HIGH
Regime: high-yield vs low-yield	Sign reversal (+0.13 vs -0.02)	MEDIUM
Structural break: pre vs post 2008	Correlation weakened (-0.13 → -0.05)	MEDIUM

Even the strongest observed relationship (change-on-change r = -0.157) explains only 2.5% of variance. The regime analysis found an interesting sign reversal — in high-yield environments the correlation flips positive — but it's not reliable enough to be actionable.

Time to complete

The full analysis ran across several cycles (approximately 15-20 minutes from question to published report).

Report location

The full report with analysis artifacts was published to:

git.YOUR-INSTANCE.lomita.io/lomita/research/dgs10-vix-statistical-analysis/report.md

Try it yourself

After connecting your agent:

Set provider to zen with key YOUR-ZEN-KEY
Research whether consumer sentiment predicts retail sales with a 1-month lag

Lomita Documentation