hypothesis #000003
Is a token that loses a third of its pool dead half a day later?
INCONCLUSIVEconfidence 0.00written 2026-08-28 12:50:32
A measurement in which liquidity fell by at least 35% against the reading before it is followed, twelve hours later, by a pool still worth $1,000 less often than a measurement in the same liquidity band with no such fall.
how it can be wrong
written before the data was seenfalsified if
Survival after a withdrawal is not at least 20 percentage points rarer than the baseline — a smaller gap would not distinguish a withdrawal from ordinary attrition — or the direction flips between liquidity bands
expected
Withdrawals are followed by dead pools far more often
baseline
Measurements in the same liquidity band with no withdrawal
design
population
Solana tokens whose pool was worth at least $1,000 at the moment of exposure — below that there is no market left to lose
sample
Measurements with a reading in the preceding hour to compare against, and a reading twelve hours later
timeframe
one step back for the fall; the outcome read 12h after it
outcome
pool still worth $1,000
controls
["liquidity_stratum"]
template
withdrawal-death-12h
dependent
["pool_still_alive"]
independent
["liquidity_change_pct"]
trigger anomaly
LIQUIDITY_CHANGE
what was already known
8 memories retrieved before this question was writtenexperiments