Skip to content

Correlation Regime Detection

Flags when assets move as one bloc and helps attribute which asset triggered a crisis.

Stars
37,433
Licence
MIT
Files
1 file

What it does

This skill teaches an AI agent to track whether a group of assets has become highly correlated, meaning they all tend to move together and diversification has weakened. It turns that into a monitorable on/off state using two thresholds so the signal does not flicker, and it avoids reading future data when smoothing. It also describes crisis attribution (identifying which asset moved first) and a leaderboard of assets whose correlations are shifting, with explicit verdicts like AMBIGUOUS or ABSTAIN when evidence is weak.

Great for

  • Quant or risk analysts monitoring portfolio diversification across a set of assets
  • Researchers studying historical correlation regimes and post-hoc market crises
  • Developers building a monitoring pipeline that needs a causal, hysteresis-based regime flag

Skip it if

  • Generating buy or sell trades: the file itself states this is not a trade-timing signal and that regime exits lost to a plain price stop in the author's tests
  • Relying on the unpublished headline results (13 regime cycles, zero wrong culprit names, FTT timing) as verified performance, since the file says they are not independently verifiable

Try asking

  • “Compute the edge density of my daily return matrix for these 8 crypto assets and tell me when the market entered a fused regime”
  • “Smooth this density series causally and run the hysteresis state machine with enter 0.65 and exit 0.45”
  • “Explain which asset moved first during the March 2020 selloff using the first-mover attribution method”
  • “Build a correlation-rewiring leaderboard for my watchlist to spot slow changes in relationships”
Step-by-step guide: how to use Correlation Regime DetectionInstall, first run, examples, tips and FAQs

Install in 3 steps

  1. 1

    Download the skill

    The ZIP contains the 1 original files, the MIT licence and an attribution note.

    correlation-regime.zip
  2. 2

    Add it to your agent

    Claude Code (all projects)

    mkdir -p ~/.claude/skills && unzip correlation-regime.zip -d ~/.claude/skills/

    Claude Code / Agent SDK (this project only)

    mkdir -p .claude/skills && unzip correlation-regime.zip -d .claude/skills/

    Claude apps (web/desktop): Settings → Capabilities → Skills → Upload skill, and choose the ZIP.

  3. 3

    Ask for the task

    No special command needed — the agent reads the skill's description and loads it when your request matches. Try one of the prompts above.

Get better results

  • Recalibrate edge_threshold and the enter/exit thresholds to your own calm-period density distribution rather than using the defaults.
  • Use trailing windows only for live monitoring; centered windows silently read future data, as the file warns.
  • Treat outputs as risk context and post-hoc attribution, not as trading signals.

What's inside

  • SKILL.md

Redistributed unmodified from HKUDS/Vibe-Trading under the MIT licence. The assessment was drafted with AI from the skill's files on 2026-10-29. Skills can instruct your agent to run code — read SKILL.md and any scripts before installing.