When to Buy Pump.fun Coins: Fixed 53-Position Paper Cohort

Correction note (September 2, 2026): We corrected the prior execution label to “mechanically paper-executed trades using live market prices” and added subgroup, selection and execution limits. The cohort and figures are unchanged.

Where this Pump.fun entry data comes from

MemeAssist ran an automated paper strategy on freshly graduated pump.fun tokens using observed live market prices. For every simulated entry it logged the market conditions at that moment — liquidity, first-hour volume, buy/sell ratio and short-term price move. This article uses the fixed, strategy-selected 53-position paper cohort that closed between July 25 and August 2, 2026: the same cohort as the exit-timing study and rug-timing study. Winners, losers and recorded dead outcomes are all counted.

Is it better to buy early or wait after graduation?

Buying the graduation print itself means buying into heavy potential sell pressure — bonding-curve buyers, sniper bots and insider wallets all get an exit at that moment (why coins dump after graduation). Within the fixed, strategy-selected paper cohort, median modeled-position resolution was about 15 minutes and median time from detection to modeled entry was 3.4 minutes; neither observation is a population timing rule.

The question is not only "earlier vs later" but also structure. One pattern to investigate is a pullback that gets reclaimed: the first post-graduation dump arrives, buyers absorb it, and price recovers its ground. The fixed, strategy-selected paper cohort does not establish that pattern as a universal filter.

Should you buy a meme coin that is already pumping?

The fixed, strategy-selected paper cohort provides exploratory subgroup observations, not a general “no”:

• Within the fixed, strategy-selected paper cohort, the 10%+ short-term-move subgroup contained 13 modeled positions, averaged −8.3%, and had an average peak gain of +24.7%.
• Within the fixed, strategy-selected paper cohort, the 0–10% subgroup contained 36 modeled positions, averaged +0.9%, had an average peak gain of +36.4%, and included 24 profitable closes.
• Within the fixed, strategy-selected paper cohort, the red-momentum subgroup contained four modeled positions, all of which lost money, with an average peak gain of 0.0%.

Chasing a vertical candle means paying the price earlier buyers may be selling into. Within the fixed, strategy-selected paper cohort, all four red-momentum entries lost money, but that tiny subgroup cannot establish a general rule. A calmer structure is a condition to investigate, not a proven profitable zone.

Does high volume mean a good entry?

Not by itself. Within the fixed, strategy-selected paper cohort, the over-$100k first-hour-volume subgroup averaged −12.2% per modeled position, while the below-threshold subgroup averaged +10.4%. This exploratory comparison does not show that loud volume generally marks a peak or wash trading; the screen separately rejects volume-to-liquidity ratios that suggest wash activity.

Within the fixed, strategy-selected paper cohort, the first-hour buy/sell-ratio-of-4-or-higher subgroup produced 23 profitable closes among 36 modeled positions (average +1.7%), while the 2-to-4 subgroup averaged −10%. These selected, overlapping subgroups do not establish that either ratio predicts winners.

What is the best market cap to buy a pump.fun coin?

Market cap alone is a weak filter — a pump.fun token graduates around a ~$69k market cap by design, so fresh graduates start in a narrow band. The following characteristics separated modeled entries within the fixed, strategy-selected paper cohort:

• Pool liquidity, not market cap. The modeled strategy skipped pools under $50k. Within the fixed, strategy-selected paper cohort, median entry liquidity was ~$189k and seven of the eight rug-tagged modeled positions had over $150k at entry. Those cohort observations do not make either value a validated threshold.
• Very low caps are lottery tickets. In a separate sample of 37 sub-$30k tokens our screens rejected and then tracked, only one ever gained 25% — the "get in at $20k" dream mostly buys tokens nobody else ever buys.
• Compare structure with size. The fixed, strategy-selected paper cohort's entry subgroups are exploratory and do not prove that one structure consistently beats another.

The modeled Pump.fun entry checklist

1. Safety screen first. Mint/freeze authority, LP status, holder concentration including bundled wallets, creator history. Run the mint through the MemeAssist analyzer — this removes the free-square scams before timing even matters (full checklist).
2. Liquidity floor: at least $50k in the pool, and volume that isn't a multiple of liquidity (wash-trade smell).
3. Wait for the first dump to be reclaimed. The burden of proof is on the token. A dip that keeps making lower highs is the market telling the truth.
4. Investigate calm rather than vertical entries: the fixed, strategy-selected paper cohort's 0–10% and 4+ buy/sell-ratio subgroups were exploratory, not validated entry thresholds.
5. Size down on very young pairs. The modeled strategy halves position size on pairs only minutes old; this is a risk rule, not a finding that quantifies their population rug risk.
6. Have the exit written down before the buy. Within the fixed, strategy-selected paper cohort, median modeled-position resolution was ~15 minutes. The full exit ruleset is in the exit-timing guide.

Methodology notes

Cohort: 53 strategy-selected modeled positions mechanically paper-executed using observed live market prices, with no manual overrides, on freshly graduated pump.fun tokens from July 25 through August 2, 2026. Entry-condition metrics (liquidity, first-hour volume, buy/sell ratio and short-term price move) were snapshotted at simulated entry; averages are per position and unweighted. The separate sub-$30k observation contains 37 rejected-and-tracked tokens from the same period. The historical cohorts are fixed rather than refreshed with later outcomes.

Study limits: the strategy selected tokens that passed its rules, so this is not a random sample of pump.fun graduates. Several comparisons are small and underpowered: the 10%+ subgroup has 13 trades and the red-momentum subgroup has four; they are exploratory observations, not validated thresholds. The analysis does not adjust for liquidity, token age or overlapping entry signals. Paper observations can miss fill failures, fees, slippage and market impact, so reported returns are not executable net returns and do not guarantee future performance.