Core Research Approach

Our AI system is designed to study and exploit funding rate differentials across perpetual swap markets. Perpetual swaps pay or charge funding periodically, and different exchanges often show different rates for the same asset at the same time. The model captures that differential by:

Long Leg

Taking a long perpetual position on the exchange where funding is most favorable to longs.

Short Leg

Taking a short perpetual position on the exchange where funding is most favorable to shorts.

Delta Neutral

Sizing both legs so the portfolio holds near-zero net directional exposure to the underlying asset price.

Result: Model performance is driven primarily by funding received minus funding paid, rather than by price direction. This is a market-neutral research framework.

Sources of Model Returns

Funding Differential

Collecting on the leg that receives funding while paying less on the other leg. The spread between the two rates constitutes the core research signal.

Rate Convergence

Optional upside when funding rates normalize over time, improving the carry profile of existing positions.

Microstructure Effects

Secondary contribution from execution efficiency and market mechanics observed during live operation.

Risk Controls

Delta Neutral at Portfolio Level

Across all open positions, the model targets net USD notional long approximately equal to net USD notional short. Significant directional price moves are therefore designed to have minimal impact on the portfolio. The system does not attempt to predict market direction.

Beta Hedged at Venue Level

Even with overall delta neutrality, individual exchanges can accumulate residual directional exposure due to operational constraints and market mechanics. The model independently manages net exposure per venue, maintaining neutrality at both portfolio and exchange level. This reduces venue-specific risk during volatile market conditions.

System Architecture

The research system operates through five integrated processing layers, from raw data ingestion through to execution and governance:

1

Data Collection and Normalization

Continuous ingestion and normalization of market data across exchanges, including order book depth, funding rates, and trade flow.

2

Deterministic Analytics and Feature Generation

Raw data feeds are transformed into structured analytical features and actionable metrics used as inputs to the AI layer.

3

AI Interpretation of Analytics

AI models process the feature set to detect patterns, anomalies, and market regime shifts that inform model decisions.

4

AI Portfolio Construction and Optimization

AI-driven optimization generates portfolio allocation plans subject to strict risk constraints defined at the research design stage.

5

AI Execution and Risk Governance

Final execution decisions are made within a comprehensive risk governance framework, with real-time controls and dynamic position management throughout.

Review the Model Results

See how the AI system has performed across different market conditions during our internal validation period.