Cross-Exchange Crypto Pricing with Returns and Order Book Signals

Summary

This draft presents a research framework for estimating short-horizon crypto prices from fixed-interval data. It resamples futures and spot mid-prices at 100-millisecond intervals, derives returns, and preprocesses order-book imbalance features. The proposed drivers include futures and spot returns, imbalance, and funding. It argues that a model trained on a liquid venue or trading pair may transfer to other exchanges and correlated assets, while emphasizing that venue structure, fees, microstructure, latency, and the strength of cross-market relationships affect transferability.

The notebook includes feature construction, backtesting examples, and information-coefficient calculations across forward-return horizons. It reports observations that combining market-level and venue-specific reversal signals with order-book features may help, while explicitly questioning whether the apparent improvement is genuine or overfit. The work is marked as a draft, uses a limited demonstration dataset, and leaves longer-period validation unfinished. Its transfer and performance claims therefore remain research hypotheses requiring robust out-of-sample tests and realistic latency and trading-cost assumptions.

Key ideas

• The framework uses fixed-interval resampling to prepare high-frequency futures and spot price data.
• It combines returns, order-book imbalance, and funding as candidate price drivers.
• Signals may transfer across exchanges and correlated pairs, but transfer depends on market structure and latency.
• The notebook evaluates signals with backtests and information coefficients over different forward horizons.
• The draft flags potential overfitting and leaves longer-period validation incomplete.

Tags

This summary was written by Stratmill's research agent from the original; it is not a copy of the source.