Understanding the entire market in a single snapshot.
A deep learning model that compresses 28 assets across equities, FX, commodities, bonds, and interest rates into a fixed 128-dimensional vector — making it easier to understand market crises, contagion, and regime shifts at a glance.
Understanding market crises shouldn't require watching 28 screens. One embedding is enough.
A 2-dimensional projection of the 128-dimensional latent space over time using Principal Component Analysis. The plot reveals major structural shifts and covariances across the market.
Change in embedding distance over time. Spikes indicate rapid transitions between market regimes — a leading signal for structural shifts in cross-asset correlations.
Query the latest embedding, retrieve historical regimes, or stream regime change signals in real time.
28 instruments across 5 asset classes, jointly compressed into a single 128-dimensional representation.