"Absorption" on a retail chart: a story that fits any outcome
Posted: Sat Oct 31, 2026 8:00 am
The classic claim is that big volume at a level with little price movement means a large passive player is absorbing the other side, so price will reverse. It sounds convincing and in futures, with a real order book and traded size, the idea can be examined directly. On a spot FX tick-volume chart it is mostly storytelling.
Reason one: tick volume does not measure size. Fifty ticks might be fifty one-lot trades or one big trade sliced up by a dealer. Reason two: the claim is unfalsifiable. If price reverses after the high-volume bar, absorption worked. If price breaks through, it was "absorption failing" or "a bigger player arriving". Both outcomes confirm the theory.
To give it a fair test, define the pattern in numbers before looking at results: tick volume above twice its twenty-bar average, range below half of its average, at a prior swing level. Then record the next ten-minute move across a few hundred cases. If the direction is not clearly skewed, the label is not helping.
Treat it as a descriptive term for compression at a level, not as knowledge of who is on the other side.
Has a rule like that ever shown a measurable edge over random entries on your data?
Reason one: tick volume does not measure size. Fifty ticks might be fifty one-lot trades or one big trade sliced up by a dealer. Reason two: the claim is unfalsifiable. If price reverses after the high-volume bar, absorption worked. If price breaks through, it was "absorption failing" or "a bigger player arriving". Both outcomes confirm the theory.
To give it a fair test, define the pattern in numbers before looking at results: tick volume above twice its twenty-bar average, range below half of its average, at a prior swing level. Then record the next ten-minute move across a few hundred cases. If the direction is not clearly skewed, the label is not helping.
Treat it as a descriptive term for compression at a level, not as knowledge of who is on the other side.
Has a rule like that ever shown a measurable edge over random entries on your data?