Trading accounts and event studies answer different questions.
The fixed Solana-token cohort supports three separate paper portfolios and a separate research lab. An event-study average is not an account return.
Reads
Forward price, reported liquidity, volume and buy/sell-count observations, with SOL as a reference. Eligibility also requires warmup, pool age and minimum activity.
Records
Portfolios record budgeted paper trades; the lab records later outcomes for individual conditions.
Momentum
Looks for
A token rising faster than SOL, with enough reported activity and buying pressure.
Acts when
The fixed momentum conditions and eligibility checks pass.
Why it might fail
Reported activity may not reflect independent demand, and the rise may already be exhausted.
Momentum + liquidity filter
Looks for
The same momentum signal, with added checks for retained liquidity and excessive turnover.
Acts when
Both the original signal and the additional liquidity checks pass.
Why it might fail
A reported liquidity number may overstate how easily a position can be sold.
Selloff rebound
Looks for
A sharp fall followed by a limited recovery.
Acts when
The registered shock and recovery thresholds pass with the common screens.
Why it might fail
A rebound can be temporary while the underlying token keeps weakening.
How the paper portfolios trade and exit
Each portfolio starts with $1,000, spends up to $50 per entry and has at most four positions. Exits request a sale after a 4% price decline, 6% gain or two hours. The model charges costs on both sides; unavailable exits remain unresolved.
What the separate research lab adapts
It tests quiet relative strength, price with liquidity expansion, and an orderly rebound. These are two-hour event studies without a shared cash budget or portfolio stop/take-profit rules. Events can overlap.
After sufficient warmup, at most one daily adaptive variant takes thresholds from the distribution of earlier observations, rather than choosing the most profitable backtest. Its rule is registered before accepting new events for 24 hours. Earlier variants stay in the record, and the trading portfolios keep their original rules.
Missing outcomes are counted separately; the mean of completed events can look too favorable when difficult exits are missing. Addresses and reported trade counts do not establish distinct people or genuine demand.
THE HYPOTHESIS
What we’re asking.
Momentum, added liquidity, or a recovery after a selloff might contain information about what happens next. We test each explanation instead of treating volume as proof of demand.
WHY THIS TEST
A manageable question.
A fixed cohort keeps the comparison honest. New names are recorded for future studies instead of replacing earlier disappointments.
FROM AN IDEA TO A RECORDED DECISION
How the study works.
01
Keep the cohort
Record the original category sample and keep missing members visible. The category sample is not the entire meme market.
02
Check the basics
Require forward observations, known pool age and minimum reported liquidity and volume before considering an entry.
03
Separate two kinds of test
Capital-constrained paper portfolios have budgets and exits. Independent two-hour event studies ask narrower questions and do not share a cash budget.
04
Retain the failures
Risk halts, missing quotes, losses and unfinished events remain part of the evidence. A halt is not reset to create more trades.
WHAT COULD CHANGE THE INTERPRETATION
The limits belong beside the idea.
Reported volume does not establish independent buyers. Sparse on-chain sampling cannot identify every linked wallet, wash trade or beneficial owner.
The paper model charges 1.25% fees and a 0.50% slippage allowance per side, plus size impact and a network allowance. Real execution may differ substantially.
Overlapping strategies can benefit or lose from the same price move. Their observations are correlated, and their independent balances must not be added together.
WHAT WE NEED TO LEARN NEXT
Does the evidence survive closer inspection?
Before proposing another rule, distinguish a bad price move from a cost problem, a missing-data problem, or repeated exposure to the same event.
Detailed recorded results are not publicly released yet. This page describes the experiment; it does not establish profitability.
