Do not just use raw imbalance. Use weighted order flow where recent trades matter more:
flow_signal = sum( trade_volume[i] * weight[i] * aggressor_sign[i] for i in last_N_trades )
Where weight[i] = exp(-lambda * time_delta). This is the “daemon” recognizing that old flow decays. daemon goldsmith order flow trading for fun and profitpdf
A background daemon that:
Large limit orders stack on the bid (support) but keeps getting eaten by market sells. When the stacked bid vanishes, price snaps down.
Trade: Short the snap. The daemon tracks bid stack size over time. Do not just use raw imbalance
To understand the strategy, you must understand three pillars: Where weight[i] = exp(-lambda * time_delta)
For quant traders, building a daemon that watches the “digital gold” (crypto/futures) order flow is like solving a live puzzle. The satisfaction comes from seeing your bot catch a large player’s footprint before the crowd.