Testing Order Book Signals for ETHBTC Trading Strategies

Summary

This project studies historical ETHBTC order book data from Poloniex and tests two ways of using order flow in simulated trading. The first estimates prices from aggregated bid and ask positions and uses those estimates to filter trades generated by a moving-average crossover strategy. The project reports a moderate improvement for its best-performing simulation relative to using the crossover alone, though the supplied abstract gives no fuller performance statistics.

The second approach looks for imbalances in market orders and hypothesizes that traders caught on the wrong side may exit at a loss, pushing prices in the opposite direction. It reports a measurable advantage across a range of tested parameters, but estimates the average edge at about 0.01% per trade with market orders—below the stated cost of entering and exiting on the exchange. The author also warns that selecting only the best parameter simulation is cherry-picking; examining results across parameter settings gives a more realistic view. The findings are limited to the tested pair, venue, data, and simulations, and do not establish future profitability.

Key ideas

• The project tests whether aggregated bids and asks can filter moving-average crossover trades in ETHBTC.
• A second strategy seeks to trade against market-order imbalances when affected traders may exit losing positions.
• The reported per-trade advantage for the imbalance strategy is below the stated market-order commission cost.
• Comparing results across parameter settings helps reveal selection bias from reporting only the best simulation.
• The simulations do not establish that the signals will remain profitable in other markets or future periods.

Tags

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