The Order Book Imbalance strategy reads the Level 2 order book — the full depth of bids and asks at each price level — to detect when one side significantly outweighs the other. A 3x or greater bid-to-ask ratio suggests strong buying pressure is building, with substantial orders waiting to be filled below the current price. The strategy goes long, expecting the weight of resting buy orders to push price upward. The inverse signals short opportunities. This is a high-frequency strategy by nature — order book states change rapidly (sub-second), and imbalances can persist for minutes at most. It's designed for scalping small moves rather than capturing trends. The strategy cannot be backtested with the current data infrastructure because historical Level 2 order book data requires continuous streaming snapshots at sub-second intervals — terabytes of data that aren't practical to store and replay. The hourly OHLCV data used for other strategies doesn't contain order book information. In production, this strategy uses the Coinbase WebSocket Level 2 channel for real-time book snapshots. Key risks include: spoofing (fake orders placed to mislead), order book "walls" that get pulled before execution, and latency disadvantages against professional market makers.