Volume pulls price toward it
Heavy volume at the current price, or along the route back, goes with a return. Thin volume goes with escape. This reversed our own starting hypothesis.
The question
A common reading says that heavy volume outside value means the market accepts the new prices and will escape, and that thin areas are where price slides through. We started with that hypothesis ourselves. Then we measured it.
Result: outside value
When price trades outside the value area, heavy volume at the current price (relative to the point of control) goes with a return to the point of control. Thin volume goes with escape.
| Outside value, ES, held-back year | Returned to the point of control |
|---|---|
| Thinnest volume at price | 26% |
| Heaviest volume at price | 63% |
New volume at that price pushed the same way as old volume. It behaved like absorption, not acceptance.
Result: between two balance areas
The volume along the route back to the area price came from carried the same lesson. Under a test whose pass rule was written in advance, route volume improved the forecast beyond distance and time in all three markets in the test years, and in two of three (ES, NQ) in the held-back year. It was the largest gain we found in that situation.
| Route back to origin | Returns above the random walk |
|---|---|
| Thinnest fifth | about +1 pt |
| Heaviest fifth | about +10 pts |
What it means
Volume is a magnet, not a road. Heavy areas pull price toward them; thin areas are where price is reached less often. This matches the classic reading of high- and low-volume nodes, and it reverses our own starting hypothesis. We report it the way the data came out.
Limits
- Outside value, the volume effect is clear across all moments, but at the moments a model actually makes a call it adds little beyond distance and time. Most of that model's gain over pure distance comes from time.
- A broader claim, that volume pulls price toward any level, did not survive. See What we withdrew.
How these numbers are produced and compared: our method. Every change to this finding is dated in the research log. Historical results only; not a forecast and not a trading system. See the disclaimer.