Finora AI Search markets
Language: English
Open bot

Should you buy after a pump? What followed 15-minute pumps on Binance futures

Across 90 days of Binance USDT-M perpetuals, buying the close of a pump did worse than buying any 15-minute candle: 36 in 100 pumps were higher a day later, against 47 in 100 of any candle. A pump here is a 15-minute candle that closed 5 percent or more above its open. Most went higher first, and within four hours 80 in 100 traded back to the middle of the candle’s body and 51 to its open.

On this page (8 sections)

We counted 864 pumps, 978 dumps and 1.03 million 15-minute candles in all on 120 liquid Binance USDT-M perpetuals over 90 days.

What followed 864 pumps

15-minute candle, per 100Higher after 1hHigher after 4hHigher after 24hBack to the body’s middle within 4hBack to its open within 4h
Pump, up 5 percent or more (864)4541368051
Dump, down 5 percent or more (978)5451436032
Any candle (1.03 million)484847n/an/a

An hour later, 45 in 100 pumps were higher, against 48 in 100 of all candles on the same coins, a gap small enough to be chance. After four hours the gap was about seven points and after a day about eleven, and both held when we resampled the coins.

The usual path: higher first, then back to the middle

Most pumps traded above their high first: after 65 in 100, the next candle traded above the pump’s high, and for 55 in 100 price took out that high within four hours before it reached the middle of the body; for 27 the middle came first. Over those four hours the median high came 6.2 percent above the close and the median low 7.0 percent below it.

Green 15-minute candles of 1 to 2 percent made the same trip: 86 in 100 got back to their middle within four hours. Size did change the next day: price was higher 24 hours later after 46 in 100 green candles of 1 to 2 percent, 42 of 2 to 3 percent, 39 of 3 to 5 percent and 36 of 5 to 7.5 percent. Above 5 percent the share stopped falling, with 35 in 100 pumps of 7.5 to 10 percent and 39 of 10 percent or more higher a day later.

Too late? The run before the candle

Of the pumps, 280 came after the coin had already gained 10 percent or more in the previous 24 hours. AKE on September 2 is an extreme case: after gaining about 150 percent in a day, it closed a 15-minute candle 59 percent up, the next candle closed below that candle’s open, and a day later AKE was 57 percent under the pump’s close.

Within four hours 61 in 100 of these pumps went all the way back to the open, against 45 in 100 of the 550 without such a run. That gap held without the ten busiest coins (62 against 38), with every coin weighted equally (64 against 33), and both before and after October 7. These pumps also swung wider both ways: 88 in 100 traded above the pump’s high within four hours, against 81. A day later 30 in 100 were higher, against 38, but without the ten busiest coins that gap all but closed (31 against 32).

Pumps that closed at a fresh 24-hour high did slightly worse: 42 in 100 were higher an hour later, against 47 for pumps that closed inside the prior day’s range. On the 15m such a candle is often a break of structure, and it closes far from the low a long would rest on.

Chasing the close or waiting for the middle

For a buyer at the close, the middle of the body sat about 3 percent lower on a typical pump. A limit there was usually reached, but most of the time the middle didn’t hold. Of the 693 pumps that reached it, 63 in 100 carried on to the open within those four hours, and 41 in 100 closed above the middle a day after the pump. Of the 350 pumps that stayed above their middle through the first hour, only 35 in 100 were higher a day after the pump than at the end of that hour.

If you still want the coin, plan a structure trade on the 15m. The long rests on the last confirmed swing low. The low the move launched from, at or below the pump candle’s open, only becomes that once five candles have printed on each side of it, 75 minutes on this chart; until then the last confirmed low is an older one. The invalidation is a 15m close below that low, and the stop goes a buffer further down. On the median pump that stop sits more than 6 percent under the close, so the size has to shrink; the position size calculator works it out.

Why this doesn’t make pumps a short or dumps a buy

A short at the pump’s close needs a stop above the candle’s high, and price traded above that high within four hours after 83 in 100 pumps, by a median of about one more body when it did. The average works against the short too: the median pump was 4.4 percent lower a day later but the average only 1.1 percent, because some kept going. BULLA closed a 15-minute candle 55 percent up on September 5, didn’t trade back to the middle of its body in the next 24 hours, and a day later was 96 percent above that close. TUT was more than three times its pump close a day after a 6.7 percent candle on August 8.

Dumps behaved differently, partly because about one in six came from five market-wide candles, 95 coins in a single candle on August 22. Those mostly kept falling: 37 in 100 took back half the drop within four hours and 68 were lower a day later. Of the other 811, 65 in 100 took back half, against 80 for pumps, and 55 were lower a day later, close to the 52 for any candle.

Ten coins made half the pumps

AKE alone produced 91 pumps, BTW 65 and US 56, and ten coins made up just over half the total. Without those ten, 40 in 100 pumps were higher an hour later and 31 a day later, and 79 came back to the middle within four hours.

The gap to any candle also held as the market turned. BTC rose from about 64,900 on August 8 to between 85,000 and 87,400 from late September to October 7, then fell to 76,000 by November 6. Before October 7, 37 in 100 pumps were higher a day later, against 50 for any candle in that stretch, and after it 34 against 43.

Where Finora fits

Finora’s Pump & Dump alerts and /pumpwatch, which messages you when a coin you name moves 5 percent either way within about 15 minutes, tell you that a move like these is under way, not whether to buy it. To see where that candle sits in the structure, send the coin and timeframe to the bot, say AKE 15m: the chart table in the analysis shows the last confirmed swing high and low and the close that would flip the trend. Watch levels then messages you when price touches a plan level, though a level only counts as broken on a candle close (see how Watch alerts work).

How we measured

  • Window: candles opening from 00:00 UTC on August 8 to 00:00 UTC on November 6, 2026 (90 days), plus 15-minute candles to November 7 for the 24-hour outcomes.
  • Universe: Binance USDT-M perpetuals on crypto coins, stablecoins excluded, listed before August 1 and trading on August 8, ranked by average daily volume over August 1 to 7 and cut at 120 (the last about 10 million USDT a day), so no coin got in through a pump we count. STG, EPIC and PTB, delisted during the window, count until trading stopped; contracts already dropped from Binance’s list are missing.
  • Events: a close at least 5 percent above the open is a pump, at least 5 percent below a dump. Repeats on the same coin within 4 hours are skipped.
  • Outcomes: higher compares the close 1, 4 or 24 hours on with the event candle’s close. The middle and the open are the body’s, reached when a low (for dumps, a high) touches them within 16 candles. The baseline is every traded 15-minute candle on the same coins.
  • Checks: 2,000 resamples of whole coins, the 3, 8 and 10 busiest coins removed, every coin weighted equally, a split at October 7. The prior-day comparison skips the 34 pumps from the window’s first day.
  • Limits: one exchange, closing prices with no fees, slippage or funding, and events on different coins overlap in time. This describes what followed and is not a tested trading rule.
  • Reproduce: Binance’s public futures endpoints exchangeInfo and klines (1d for the ranking, 15m for the rest).

Today’s trend and key levels: Akedo (AKE) · Bitway (BTW) · Bitcoin (BTC)