The most repeated piece of PSX folklore
The story is elegant. Volume dries up, the float is locked away, nobody is selling, and the stock is coiled for a move. It is told constantly, and it sounds like insider knowledge.
So we tested it. Define dry-up as 5-day volume below 30% of the 60-day average, then measure how often those stocks lock:
5.5% locked, against a 6.2% base rate on the same definition, a lift of 0.88x. [LEDGER: project_research_findings]
A dried-up stock is less likely to fire than a randomly chosen one. Trading the folklore directly makes you slightly worse than a coin, before costs.
Where the folklore came from
It is not pure invention, and this is the interesting part. Dry-up is the best risk-adjusted operator archetype in a separate study [LEDGER: project_operator_forensic], and the sequence SHAKEOUT → LOCK is the strongest transition we have found (+9.32%). So there is something real in the neighbourhood.
What fails is dry-up alone, as a trigger. It is a condition, not a signal. The distinction is the whole lesson:
- A condition describes the state of a stock. It changes the odds of other things.
- A signal tells you a specific thing is more likely to happen soon.
Folklore is almost always a condition being sold as a signal.
A second warning from the same family
Dry-up and range-compression have been measured as INVERTED relative to the way they are usually described [LEDGER: project_lock_forensics]. If you have been using either one directionally, check the sign before you check the threshold.
Evidence · History · Confidence · Counterargument
Evidence, 5.5% vs a 6.2% base = 0.88x. [LEDGER: project_research_findings]
History. One large historical study on the PSX panel.
Confidence, high that dry-up alone does not predict a lock. Medium on the exact 0.88x, which depends on the 30%/60-day definition.
Counterargument. The strongest case against this rejection: dry-up may still be a valuable filter in combination (it is the best risk-adjusted archetype elsewhere in our own work), and this test measured only one threshold pair. It rejects "dry-up as a standalone trigger", not "volume behaviour is uninformative".