Per-bar order-flow delta predicts the next bar
The delta bar is the price bar restated, so the information lives in cumulative flow measured against price, not in the per-bar figure.
- Verdict
- Settled by the data
- Decided by
- Redundant signal
- The number
- Delta vs. own-bar return: near-collinear
The hypothesis, as pre-registered
Signed volume within a bar reveals aggressive participation and should therefore forecast the following bar's direction better than the bar's own return does.
What the test found
Per-bar delta and the bar's own return are close to the same variable. Buyers lifting the offer is what an up bar is. Conditioning the next bar on the previous bar's delta is conditioning it on the previous bar's return with extra steps and a worse name.
Once the previous bar's return is included as a control, the marginal information in per-bar delta collapses toward zero on every timeframe tested.
This is not a statement that order flow is useless. Cumulative signed volume measured against price — divergence, not level — survives where the per-bar figure does not. The failure is specific: the per-bar delta as a standalone trigger adds nothing to the price bar sitting next to it.
What would make this work
Delta would have to carry information after controlling for the same bar's return — for example as a divergence measured over a window, or sourced from a different tape than the one being traded.
Every insight ends with this line: the exact condition that would turn the finding into a tradeable edge. That is what makes each entry a next experiment rather than a full stop.
Screened June 14, 2026
Testing something similar?
The screener puts your version of this idea through the same gate stack, on your market, and answers in seconds. Your first verdict is free.