Do pump.fun dips bounce? Four in ten touch +30%. 95.2% of the coins halve again within a day
We tested the dip-buying rule chart traders teach (buy migrated pump.fun coins 40-75% off their high, ideally after four red candles and two green) on every coin that moved to PumpSwap from Oct 9 to Nov 7, 2026. After a 40% dip the price touched +30% 39.4% of the time, but 95.2% traded at half the entry within 24 hours, and a +30%/-30% bracket trade lost 15.7% on average after fees.
Quick answer
On pump.fun coins that had moved to PumpSwap, buying a dip of 40% off the high gave a bounce sometimes and a profit rarely. Of 14,273 such dips between Oct 9 and Nov 7, 2026, the price went on to touch 30% above the entry 39.4% of the time, and back to the old high 28.7% of the time. But 95.2% of those coins traded at half the entry or lower within the next 24 hours, and the median coin closed that day 91.9% under the entry. A trader who took +30% or cut at -30%, whichever came first, lost 15.7% a trade on average after pump.fun's fees. Buying at a random minute lost 7.5%.
What we tested
The rule comes from the dip-buying lessons small trading groups pass around: on a migrated coin, wait for a pullback of 40% to 75% from its high, and better still, wait for four to six red candles and then two green ones before buying; then scalp 10% to 30% back. We tested that literally, on the chain's own record.
- The coins: every pump.fun coin that migrated to PumpSwap from Oct 9 to Nov 7, 2026: 27,978 of them. 10,620 of those had their whole bonding curve bought within 10 seconds of their create, so they reached PumpSwap almost at once; we count them in, and show them apart below. 18,371 coins had a pool with real prices (10 SOL or more in the pool and 1% of the supply or more) in their first 48 hours: 10,586 bought at launch and 7,785 others.
- The price: the coin's own PumpSwap pool, trade by trade, in one-minute candles.
- A dip: the first minute in the coin's first 24 hours in its pool when the price traded 40% (or 50%, 60%, 75%) under the highest price before that minute. The entry is that line exactly: a resting buy there fills at it or better.
- The confirmed entry: after the 40% dip, the first time within six hours that four red one-minute candles were followed by two green ones; the entry is the second green candle's close.
- The baseline: each coin once, at the close of a random minute in its first 24 hours.
- What we measured, over the 24 hours after each entry: how high it went, how low, which came first of +30% and -30%, and where it closed.
What happened after the dip
All 18,371 coins:
| entry | dips | touched +10% | touched +30% | touched +100% | back to the old high | halved | median close after 24 h |
|---|---|---|---|---|---|---|---|
| 40% off the high | 14,273 | 47.7% | 39.4% | 22.2% | 28.7% | 95.2% | -91.9% |
| 50% off | 13,600 | 45.5% | 37.2% | 20.5% | 20.5% | 95.2% | -91.2% |
| 60% off | 12,837 | 43.4% | 35.2% | 18.3% | 13.7% | 94.7% | -89.8% |
| 75% off | 11,240 | 42.8% | 33.6% | 17.4% | 6.4% | 92.9% | -85.3% |
| 40%, then 4 red and 2 green | 2,981 | 79.0% | 59.0% | 28.1% | 41.1% | 78.1% | -74.1% |
| a random minute | 16,601 | 69.4% | 44.3% | 15.9% | 77.5% | -91.7% |
A plain dip buy touches +30% less often than a random minute does (39.4% against 44.3%), though it touches +100% more often (22.2% against 15.9%). Nearly every coin that dipped 40% went on to lose half the entry price within the day, and only 3.6% closed the day above the entry. Deeper dips change little: at 75% off, the price got back to the old high 6.4% of the time.
The confirmation candles help most. Waiting for four red and two green after a 40% dip found 2,981 entries, and 79.0% of them touched +10%, the nearest any version came to the "nine out of ten" such lessons promise. They still halved 78.1% of the time.
Bought at launch, and the rest
The two kinds of migrated coin behave differently enough to show apart. A dip on a coin whose curve was bought out within 10 seconds is the worse buy:
| entry | group | dips | touched +30% | halved | median close after 24 h |
|---|---|---|---|---|---|
| 40% off the high | bought at launch | 7,512 | 29.6% | 96.4% | -97.2% |
| 40% off the high | the rest | 6,761 | 50.3% | 93.8% | -86.6% |
| 40%, then 4 red and 2 green | bought at launch | 662 | 66.9% | 84.9% | -97.1% |
| 40%, then 4 red and 2 green | the rest | 2,319 | 56.7% | 76.1% | -67.9% |
| a random minute | bought at launch | 9,487 | 45.7% | 87.1% | -97.0% |
| a random minute | the rest | 7,114 | 42.6% | 64.7% | -67.6% |
On the other coins, a 40% dip does bounce more than a random minute (50.3% against 42.6% touched +30%). On the coins bought at launch, it bounces less (29.6% against 45.7%): their first drop is more often the start of the end.
The trade, with fees
A scalper takes +30% or cuts at -30%, whichever comes first, and sells at the 24-hour mark if neither came. We counted the stop as filled exactly at -30%, which flatters it: a falling pool often trades through the line. pump.fun's PumpSwap fee at these market caps is up to 1.25% a side, 2.5% a round trip.
| entry | coins | +30% before -30% | average trade | after fees | trades that made money after fees |
|---|---|---|---|---|---|
| 40% off the high | all | 27.8% | -13.2% | -15.7% | 28.0% |
| 75% off | all | 21.8% | -16.7% | -19.2% | 22.0% |
| 40%, then 4 red and 2 green | all | 42.7% | -3.6% | -6.1% | 43.0% |
| a random minute | all | 36.8% | -5.0% | -7.5% | 40.1% |
| 40% off the high | bought at launch | 20.4% | -17.6% | -20.1% | 20.5% |
| 40% off the high | the rest | 36.1% | -8.2% | -10.7% | 36.3% |
| 40%, then 4 red and 2 green | bought at launch | 54.2% | 3.3% | 0.8% | 55.0% |
| 40%, then 4 red and 2 green | the rest | 39.5% | -5.6% | -8.1% | 39.6% |
| a random minute | bought at launch | 38.8% | -4.3% | -6.8% | 42.4% |
| a random minute | the rest | 34.2% | -5.9% | -8.4% | 37.1% |
The plain dip buy loses more than buying at any minute, in both groups. The confirmed entry loses least. One cell came out ahead: the confirmed entry on coins bought at launch, 0.8% a trade after fees over 662 trades in one month. That is a thinner margin than the stop's flattering fill, so we read it as even, not as an edge.
The result held in both halves of the window: for 40% dips on coins migrated Oct 9 to Oct 23, 37.8% touched +30% and 96.6% halved; Oct 24 to Nov 7, 41.5% and 93.4%. The bracket trade lost 16.3% and 14.8% after fees.
What this means for a dip buyer
- A dip is a coin already falling. In this window most migrated coins were close to dead within a day of their pool opening; the median coin bought at a random minute closed the day 91.7% lower.
- Know which kind of coin it is. A coin bought out whole at launch reached PumpSwap without a crowd behind it, and its dips bounced less than chance. The coin check says whether a coin's curve was bought at launch.
- The bounce is short. A scalper who sells into it at once does better than one who waits, and the numbers above assume perfect exits at the line.
- Read the coin, not just the chart. Who holds it, whether the dev is a serial launcher, whether the volume comes from five wallets trading with themselves: the check reads those, and the screener lists young coins that pass the volume-and-holders preset now.
We describe what prices did on chain. None of this is advice.
How we counted
From our archive of every pump.fun and PumpSwap trade: each migrated coin's own PumpSwap pool, its price after each trade (SOL per token, from the pool's reserves), in one-minute candles over its first 48 hours, counting only minutes with trades. A coin counts as bought at launch when its curve completed within 10 seconds of its create. A dip had to happen in the coin's first 24 hours in its pool so that a full day followed it. Outcomes start at the minute after the dip. Prices are in SOL. The study script is scripts/research/dip_bounce.py; its output is the facts file this post is checked against.
Frequently asked questions
Do pump.fun dips bounce?
Sometimes, briefly. After a 40% dip on a migrated coin the price touched +30% 39.4% of the time within a day. But 95.2% of those coins also traded at half the entry price within the same day.
Is it better to buy a 75% dip than a 40% one?
Not by this test. At 75% off, 33.6% touched +30% and 92.9% halved; the bracket trade lost 19.2% after fees, against 15.7% at 40%.
Does waiting for red then green candles help?
Some. After a 40% dip, four red one-minute candles and then two green found 2,981 entries; 79.0% touched +10% and 59.0% touched +30%, but the bracket trade still lost 6.1% after fees.
Do coins bought out at launch dip differently?
Yes. A 40% dip on a coin whose curve was bought within 10 seconds of its create touched +30% 29.6% of the time and halved 96.4% of the time, against 50.3% and 93.8% for the other migrated coins.
Which coins were tested?
Every pump.fun coin that migrated to PumpSwap from Oct 9 to Nov 7, 2026: 18,371 with real pool prices, 10,586 of them bought out at launch.
Every number here comes straight from the blockchain, the same data behind the paid calls: the calls ($0.005 to $0.01 each, USDC over x402, no account), yesterday's /stats and the daily wire are free.