pump.fun sniper bot settings: what 118,000 real bots show
Most advice on pump.fun bot settings is guesswork: a slippage number here, a take-profit there. We measured instead. For every wallet that traded pump.fun and PumpSwap between 11 and 17 October 2026, we computed the real result in SOL, fees included, and kept the 118,493 bots with at least 10 positions older than 24 hours. Then we grouped them by how they trade. Only 7.4% of them made money; the median one lost 22% of what it staked. Here is what separates the two groups.
1. Entry delay: the first seconds decide almost everything
Median delay between a token's creation and the bot's first buy:
| Entry delay | Bots | Median return | Profitable |
|---|---|---|---|
| under 2 s | 4,156 | −1.3% | 45% |
| 2 to 5 s | 1,481 | −8.0% | 22% |
| 5 to 15 s | 4,066 | −10.8% | 15% |
| 15 to 60 s | 11,898 | −16.6% | 10% |
| 1 to 5 min | 29,989 | −27.0% | 5.9% |
| over 5 min | 56,718 | −29.9% | 5.1% |
Every extra second costs. But the lesson is not simply "be faster". Many of the fastest wallets are part of the launch itself: the creator's own team, or snipers with dedicated infrastructure. A bot that buys 30 seconds after launch is usually buying from them.
2. The launch block: whoever is inside it wins
Share of the bot's entries made in the very block where the token was created:
| Launch-block entries | Bots | Median return | Profitable |
|---|---|---|---|
| under 1% | 98,710 | −27.0% | 6.1% |
| 1 to 10% | 3,238 | −14.7% | 12% |
| 10 to 30% | 2,296 | −12.1% | 19% |
| 30 to 60% | 1,302 | −5.1% | 30% |
| 60% or more | 2,762 | +0.6% | 52% |
Being in the launch block requires knowing about the launch in advance, or very specialised automation. Most bots cannot get there. The practical setting is the reverse: skip tokens whose launch block is already crowded, because those buyers are the ones who will sell to you. The hourly pump.fun weather lists the serial launch teams of the last 7 days.
3. Holding time: quick exits lose least
Median time between first buy and full exit, on closed positions:
| Holding time | Bots | Median return | Profitable |
|---|---|---|---|
| under 10 s | 4,758 | −4.2% | 28% |
| 10 to 60 s | 19,021 | −13.7% | 13% |
| 1 to 5 min | 31,332 | −22.2% | 7.4% |
| 5 to 30 min | 27,951 | −32.1% | 5.1% |
| 30 min to 24 h | 19,098 | −36.7% | 5.1% |
Most pump.fun tokens fade within minutes of launch. A bot that waits for a take-profit that never comes ends up holding the bag. If your bot holds for several minutes, a time-based exit usually matters more than the take-profit level.
4. Fees: watch the churn
A pump.fun round trip costs about 2.5% in protocol and creator fees before priority fees and tips. Bots whose fees reached 8% or more of the SOL they staked (4,696 of them, typically buying and selling the same tokens over and over) had a median return of −64%, and fewer than 1 in 200 made money. Our figures do not include priority fees or Jito tips, so add your own on top.
5. Volume: scale helps, but only the very top
| Positions | Bots | Median return | Profitable |
|---|---|---|---|
| 10 to 30 | 82,711 | −25.0% | 7.0% |
| 30 to 100 | 27,628 | −19.5% | 7.8% |
| 100 to 1,000 | 7,868 | −11.5% | 9.0% |
| over 1,000 | 286 | −1.5% | 32% |
The 286 busiest bots do far better than the rest, most likely because only operators with an edge can afford to keep trading that much. Running more trades does not create the edge.
What this means for your settings
- If you cannot enter within a few seconds, do not compete on speed: filter launches instead (crowded launch block, serial creators).
- Cap holding time. A time-based exit beats waiting for a take-profit on a token that is already fading.
- Count fees per round trip, and stop trades whose expected move is below about 3%.
- Change one setting at a time and measure again on the real result, not on the balance.
These are associations measured on past trades, not proof that a setting causes a result, and they describe six days of one market. To see where your own bot sits on each of these lines, paste its public address on Test my bot: free verdict in SOL and rank, with entry delay, launch-block share, fees and holding time in Pro.
How we counted
A position is one wallet on one token. Result = SOL received + tokens still held − SOL paid, fees included. Tokens still held are valued at the lower of their cost and their market price, so a dead token counts as a loss and a paper gain does not count until it is sold. Only positions opened more than 24 hours ago count; a bot is ranked from 10 such positions. Data: every pump.fun bonding-curve trade and every SOL-quoted PumpSwap trade, read from Solana's public chain.
FAQ
What slippage should I use on a pump.fun sniper bot?
Slippage only limits the worst price you accept; it does not make a trade profitable. In our data the entry delay and the crowding of the launch block mattered far more. A tight slippage on a late entry still buys from the launch team.
Do faster bots always make more money?
No. Bots entering within 2 seconds still had a slightly negative median result; only 45% of them were profitable. Speed helps, but many of the fastest wallets belong to the launch itself.
How do I know which group my bot is in?
Paste its public wallet address on Test my bot. You never need to share a private key or recovery phrase.