What happens to pump.fun coins after they migrate (graduate)?
Of the 13,188 pump.fun migrations Belphor detected between 28 Aug 2026 and 27 Oct 2026 that have a price reading at least six hours later, 93% were at 20% of their migration price or lower when last checked, 19% reached +50% after migrating, 5.6% reached +200%, and 5.2% were above the migration price at the last reading. Updated 2026-10-27.
Cite as: Belphor Research (2026). What happens to pump.fun coins after they migrate (graduate)? Belphor Data, https://www.belphor.com/data/migration-outcomes. Updated 2026-10-27. Inputs 12b16e0b6831, formula 1.5+t34-corrected, covered 13,188, missing 69. CC BY 4.0.
Marks for migrations detected 1 to 7 October 2026 were corrected on 8 October 2026 after a price-source defect froze them near the migration price; the correction ledger is public.
By liquidity at migration
| Liquidity | Migrations | With a later reading | Down 80% or more | Reached +50% | Reached +200% | Above entry now |
|---|---|---|---|---|---|---|
| $15k to $30k | 6,293 | 6,293 | 94% | 14% | 5.1% | 2.1% |
| $30k to $60k | 3,596 | 3,596 | 98% | 21% | 6.5% | 1.4% |
| over $60k | 3,299 | 3,299 | 84% | 27% | 5.5% | 15% |
887 rows excluded: price not verified.
By week
| Week of | Migrations | With a later reading | Down 80% or more | Reached +50% | Reached +200% | Above entry now |
|---|---|---|---|---|---|---|
| 2026-08-24 | 539 | 539 | 88% | 14% | 4.5% | 8.5% |
| 2026-08-31 | 1,315 | 1,315 | 89% | 21% | 4.1% | 8.5% |
| 2026-09-07 | 1,398 | 1,398 | 91% | 25% | 5.5% | 6.7% |
| 2026-09-14 | 1,321 | 1,321 | 91% | 14% | 4.2% | 6.9% |
| 2026-09-21 | 1,556 | 1,556 | 95% | 8.3% | 3.0% | 3.7% |
| 2026-09-28 | 1,872 | 1,872 | 95% | 6.4% | 3.1% | 3.5% |
| 2026-10-05 | 2,015 | 2,015 | 95% | 14% | 4.8% | 3.2% |
| 2026-10-12 | 1,567 | 1,567 | 93% | 29% | 9.0% | 4.5% |
| 2026-10-19 | 1,420 | 1,420 | 91% | 39% | 11% | 4.8% |
| 2026-10-26 | 185 | 185 | 90% | 38% | 16% | 6.5% |
887 rows excluded: price not verified.
Time to the peak and to an 80% fall, by liquidity at migration
| Liquidity | Migrations | With a minute path | Median peak (x) | Median time to peak | Paths that fell | Fell to 20% or lower | Median time to that reading |
|---|---|---|---|---|---|---|---|
| $15k to $30k | 5,633 | 1,863 | 1.19x | 6 min | 1,725 | 93% | 9 min |
| $30k to $60k | 3,346 | 1,448 | 1.39x | 10 min | 1,397 | 97% | 16 min |
| over $60k | 2,807 | 1,307 | 1.39x | 1h 32m | 1,095 | 84% | 1h 13m |
These two tables read the migrations with a minute-level price path for their first 24 hours, recorded between 7 Sep 2026 and 26 Oct 2026 and computed 27 Oct 2026 05:12 UTC (inputs d907239076bc, formula 1.2). "With a minute path" is the n of every figure to its right except the last, whose n is "Paths that fell". How the paths are chosen is under How this is measured.
Time to the peak and to an 80% fall, by week
| Week of | Migrations | With a minute path | Median peak (x) | Median time to peak | Paths that fell | Fell to 20% or lower | Median time to that reading |
|---|---|---|---|---|---|---|---|
| 2026-09-07 | 1,503 | 726 | 1.47x | 1h 12m | 656 | 90% | 1h 13m |
| 2026-09-14 | 1,441 | 675 | 1.35x | 12 min | 607 | 90% | 21 min |
| 2026-09-21 | 1,683 | 747 | 1.30x | 8 min | 701 | 94% | 14 min |
| 2026-09-28 | 1,974 | 629 | 1.19x | 8 min | 592 | 94% | 9 min |
| 2026-10-05 | 2,099 | 480 | 1.24x | 8 min | 443 | 92% | 12 min |
| 2026-10-12 | 1,600 | 640 | 1.31x | 9 min | 572 | 89% | 15 min |
| 2026-10-19 | 1,451 | 704 | 1.32x | 9 min | 632 | 90% | 15 min |
| 2026-10-26 | 35 | 17 | 1.24x | 8 min | 14 | 82% | 16 min |
Alive a day later, by liquidity in the first 10 minutes
| Liquidity | Migrations | Read at 24 h | Alive at 24 h | Below a fifth at 24 h | Not read at 24 h | Median peak (x) | Reached 2x |
|---|---|---|---|---|---|---|---|
| Under $10k | 690 | 137 | 18 (13.1%) | 119 (86.9%) | 553 (80.1%) | 0.05x | 12 (1.7%) |
| $10k to $20k | 853 | 812 | 99 (12.2%) | 713 (87.8%) | 41 (4.8%) | 1.27x | 216 (25.3%) |
| $20k to $50k | 940 | 828 | 42 (5.1%) | 786 (94.9%) | 112 (11.9%) | 1.53x | 328 (34.9%) |
| $50k to $100k | 1,106 | 1,064 | 18 (1.7%) | 1,046 (98.3%) | 42 (3.8%) | 1.59x | 385 (34.8%) |
| $100k or more | 1,019 | 916 | 201 (21.9%) | 715 (78.1%) | 103 (10.1%) | 1.39x | 246 (24.1%) |
| No reading in the first 10 minutes | 5,713 | 265 | 16 (6.0%) | 249 (94.0%) | 5,448 (95.4%) | 0.04x | 8 (2.2%) |
The band is the pool's liquidity at the first reading inside 10 minutes of the migration, not the detection figure the tables above use. Belphor records a migration only with $15k or more of liquidity, so a pool under $10k at that reading lost liquidity after the migration. "Alive at 24 h" is a reading 23 to 25 hours after the migration (the one closest to 24 hours) at or above a fifth of the migration price, and "Below a fifth at 24 h" is that reading under it. Both shares are over the migrations with such a reading, "Read at 24 h"; "Not read at 24 h" is a share of the band. A pool that drains early stops being read, so "Not read at 24 h" is mostly pools that were already down, and the share alive therefore reads high for the bands where that column is large. Minute readings exist mostly for the migrations Belphor alerted on. Readings stop 24 hours after a migration, so nothing here is said about later days. The median peak and "Reached 2x" are the highest reading in the first 24 hours over the migration price, over the migrations with any reading in them. A share over fewer than 10 migrations prints as too few readings, and a row with fewer than 30 read at 24 h is marked small sample. Recorded between 7 Sep 2026 and 26 Oct 2026, computed 27 Oct 2026 05:12 UTC (inputs 6d2ea6549a7b, formula 1.2). Left out: 5 migrations with a market cap over $250M at the migration, 602 whose detection price and market cap cannot both be right and 863 whose price marks a price-source defect froze in early October 2026.
How this is measured
A migration is a pump.fun coin (mint ending in "pump") appearing in a fresh Raydium, PumpSwap or Meteora pool younger than 30 minutes with at least $15k liquidity; Belphor records the price at that moment.
The window is the 60 days ending when this page was recomputed, read from both the live table and the archive of older signals, so the first and last migration it holds are the dates printed above. "Down 80% or more" means the most recent price reading, at least 6 hours after detection, was 20% of the migration price or lower. "Reached +X%" uses the highest price observed after migration. A share over fewer than 10 readings prints as too few readings.
A migration is left out when its price cannot be verified: a market cap above $250M at detection or a peak of 100x or more with no matching price snapshot in Belphor's record. The count is printed under each table.
A migration's price path is followed for 24 hours after the migration: its price is read again through that day, and a minute-by-minute path is kept mostly for the migrations Belphor alerted on. After that the scanners stop reading it, so the latest reading behind "Down 80% or more" and "Above entry now" is the last one inside those 24 hours; the daily recompute rebuilds this page from those readings and takes no new ones.
Belphor's own migration-snipe strategy was withdrawn from members on 2 September 2026 because of this record. The page stays up because the number is useful whether or not it flatters the product.
Median peak and median time to peak are measured on the migrations with a minute path: pool price readings in the 24 hours after the migration (a reading that carries the pool's liquidity figure; a leftover bonding-curve listing is not a reading), the first of them inside ten minutes, and then 30 or more readings, or fewer when one of them is already at or below 20% of the migration price, because a pool that drains inside its first half hour stops being read and leaving those coins out would hide the fastest deaths. The peak is the highest of those readings and its time is counted from the migration; a coin that never rose has its peak at its first reading, and it is counted.
Paths with fewer than 30 readings are included only when a reading is at or below 20% of the migration price; 266 short paths that never were are excluded.
"Fell to 20% or lower" is the share of those paths with at least one reading at or below 20% of the migration price inside the 24 hours, and the median time is the time to the first such reading, among the paths that have one. It is not the "Down 80% or more" column, which reads the latest price of every migration with a later reading; a pool drained in a single step can leave no reading at or below 20%, and that coin is then not timed.
Belphor snapshots a price every minute for 24 hours mostly for the migrations it alerted on (3,866 of the 4,618 paths here), and keeps those readings for 45 days, so most other migrations and every older one have no minute path. The time columns therefore cover 4,618 of the 11,786 migrations recorded between 7 Sep 2026 and 26 Oct 2026 whose 24 hours were over, and say nothing about the rest; a figure over fewer than 10 paths prints as too few readings.
Readings excluded, price unverified: 874. A detection whose recorded price and market cap cannot both be right (together they imply a token supply outside 0.7 to 1.3 billion, and a pump.fun coin is minted with one billion) is kept on the record with its price exactly as it was read, flagged, and left out of every multiple, rate and list on this page, so the denominators above are the verified rows only. From 12 October 2026 the scanner reads both figures from one pair document at detection and flags a disagreement as it writes the row.
Frequently asked
Is buying a pump.fun coin right after migration profitable?
Not in this record: 93% (n 13,188) of the migrations with a later reading went to 20% or less of the migration price, and 5.2% were above it at the last reading. The migration moment is a volatility event with the odds against the buyer.
How many pump.fun coins graduate per day?
Belphor recorded 14,144 migrations between 28 Aug 2026 and 27 Oct 2026, 60 days, roughly 236 a day, counting only pools that reached $15k liquidity within 30 minutes of creation.
How long after migration does a coin usually fall 80 percent?
Among the 4,618 migrations recorded between 7 Sep 2026 and 26 Oct 2026 that have a minute-by-minute price path for their first 24 hours, 91% (4,217 of 4,618) had a reading at 20% of the migration price or lower inside those 24 hours, and the median time to the first such reading was 16 min (n 4,217). The same 4,618 peaked a median 10 min after migrating, at 1.33x the migration price. These are the migrations Belphor snapshotted every minute, mostly ones it alerted on (3,866 of the 4,618 paths here), 4,618 of the 11,786 it recorded in that span, so the time describes them and not every migration.
Does a pump.fun coin's liquidity at migration tell you whether it is still alive a day later?
In Belphor's record of the pump.fun migrations recorded between 7 Sep 2026 and 26 Oct 2026 that were read again 23 to 25 hours later, the share still at or above a fifth of the migration price was 13.1% under $10k (n 137), 12.2% at $10k to $20k (n 812), 5.1% at $20k to $50k (n 828), 1.7% at $50k to $100k (n 1,064) and 21.9% at $100k or more (n 916), banded by the pool's liquidity at the first reading inside 10 minutes of the migration. The share does not rise with liquidity. It is computed over the migrations with that reading, and a pool that drains early stops being read, so the share reads high where many were not read. It describes these migrations, not the next ones.
Where does this data come from?
From Belphor's migration detector (GeckoTerminal new-pool data) and DexScreener prices, recomputed daily from the raw record. The JSON and CSV files on this page contain the same numbers.
Belphor records pump.fun takeovers on-chain, scores each signal 0 to 100 and publishes measured outcomes with their sample sizes, windows and missing readings. Reflex Bot is coming soon and priced separately.
Explore Belphor →