Imbalance Momentum

What it trades

QD·1 looks for pressure that persists. A single imbalanced tick means almost nothing; the same imbalance held across several hundred milliseconds, confirmed by signed taker flow and a micro-price that has already started to lean, is a different object. The system scores that state on a bounded scale and enters when the score holds above a threshold for long enough to rule out flicker.

Exits are mechanical: the score decays back through zero, the take-profit is hit, or the stop is hit. There is no discretionary layer.

What is in the box

• Signal definition. The exact state variables, their normalisation, and the composite, with the parameters the research pass settled on.
• Tuned configurations for BTC and ETH perpetuals, with the sample and the label each one was validated on.
• Execution logic. Entry and exit rules, cooldowns, and the cost hurdle the signal has to clear before an order is allowed.
• Monitoring. The health checks the lab runs on the live signal, so you know when the regime has moved.
• The operators' workspace. A private channel with the people running the system, and the lab's changelog as it evolves.

Where it came from

The state variables are taught in the open research. Read the queue-imbalance and order-flow-imbalance notes for the atoms, and the memory note for how short-lived pressure becomes a state that lasts long enough to trade.

What the preview shows

The terminal above runs the same signal against live Binance Futures data with simulated taker fills. Entries cross the spread and fees are charged both ways. Queue position and market impact are not modelled, so treat it as a behaviour demo, not a performance estimate.