About This Dashboard
Measures cosine distance between consecutive regime fingerprints and uses CUSUM changepoint detection to identify regime transitions. Tracks drift rate, pole flip analysis, and similarity to known historical macro windows (e.g. 2008 crisis, 2020 COVID).
Today's market regime shows elevated drift at 2.63 normalized units, signaling meaningful structural shifts in asset behavior compared to recent trading patterns. The market is most similar to the post-GFC ZIRP environment (similarity score 21.0), suggesting current conditions reflect low-volatility, stimulus-driven dynamics rather than crisis or acceleration modes—the GFC regime scores negatively at -3.3, indicating we're materially distant from acute deleveraging stress. However, the gap between the top regime (21.0) and secondary regimes is stark; tariff shock concerns trail significantly at 2.9, while AI-bull growth and dot-com bust scenarios are minimal (.7 and 1.4 respectively), meaning the market is not pricing meaningful stagflation risk or tech rotation yet. The elevated drift tier combined with strong post-GFC similarity suggests the market is in a grinding, policy-dependent phase but with enough momentum shift that the CUSUM detector has flagged transition activity worth monitoring. Watch for whether drift continues climbing or stabilizes; sustained elevation above 3.0 would indicate a genuine regime changepoint rather than normal volatility.
For educational and informational purposes only. Not financial advice. Past performance does not guarantee future results.
Changepoints
33
Distance Max
0.0115
Distance Mean
0.0021
CUSUM Threshold
0.01
Macro Windows
10
Feature Dims
123
Drift Rate (ELEVATED)
2.63x
Most Similar
post_gfc_zirp
Similarity
21.0%
INFORMATIONAL ONLYScimode validation: regime distance does not predict forward SPY returns. Use for regime context, not as a trading signal.
NOV 2025 - FEB 2026:1 changepoint(s) detected between Nov 2025 and Feb 2026

Regime Distance Series + Changepoints

Cosine distance between consecutive 20-day fingerprints -- spikes = regime transitions
Distance (smoothed 3d median)
CUSUM statistic
CUSUM threshold
Detected changepoints
Drift rate (normalised)
SPY (normalized scale)
Historical events
Shaded bands = macro windows

Regime Similarity -- Current vs Historical Archetypes

Cosine similarity between today's pole fingerprint and each window's mean archetype

Current Pole Fingerprint

z-score momentum per pole -- Bull = positive / Bear = negative
WINDOW 20-day z-score momentum (short-term)

Detected Changepoints

Pole flips show which factors drove each transition (pre vs post 30 days)
DateCUSUMDistance Proposed WindowPole Flips (top 6)

Proposed Macro Windows

Compare detected breakpoints to proposed boundaries