InvestNot yet confirmed elsewhere1 publisher2 min readPublished
Solana sandwich bots grossed $29.7 million over 11 months, about $4.34 an attack
Bitquery estimates Solana sandwich bots took $29.7 million gross over 11 months, before any costs. Jupiter trades allowing 20% to 50% slippage on Pump.fun venues were sandwiched 14 times as often as those under 3%.
The Investor · Invest desk
Four figures from Bitquery: 6,850,144 sandwich attacks counted; 96% of detected attacks hit Pump.fun's launch curve or PumpSwap; other bots' trades supplied 39% of the gross take; Jupiter trades on those venues allowing 20%-50% slippage were hit 14 times as often as those under 3%.
| Figure | Value | Note | Claim |
|---|---|---|---|
| Sandwich attacks counted | 6,850,144 attacks | 6 | |
| Attacks on Pump.fun venues | 96% | 5 | |
| Gross taken from other bots | 39% | 7 | |
| Sandwiched at 20%-50% vs under 3% | 14 times as often | Jupiter via Pump.fun curve or PumpSwap | 1 |
What happened
- Bitquery counted 6,850,144 attacks by 2,545 bot wallets on 2,154,994 distinct wallets between October 30, 2025 and October 7, 2026, and says the addresses are not a count of people.
- Attack rates did not climb with every wider slippage setting: trades below 0.5% tolerance were hit more often than those at 1% to 3%, and the 50%-plus band less often than 20% to 50%.
- Bitquery values SOL at the minute of each attack and says the figure does not measure net profit.
Why it matters
- cost Other bots' trades supplied 39% of the gross, about $11.6 million, so at most about $18.1 million came from trades not attributed to other bots.
- decision Traders choosing a slippage limit on a volatile launch pay either way: a tight limit can fail the swap when the price moves, and a wide one leaves room for a worse fill.
- exposure Moving to protected order flow is an unsettled answer: the authors of a preprint submitted September 23, 2026, covering six blockchains, concluded such protections can offer weaker guarantees than users expect.
Divide Bitquery's $29.7 million gross [4] by its 6,850,144 attacks [6] and the unit is small: about $4.34 an attack [16]. That is an average. CryptoSlate's account of the report does not include a median or a size distribution, so the average can hide a long tail.
Bitquery's cost sample points below the gross. In a separate sample of 63 attacks, bots retained 68% of aggregate gross proceeds, and 18 of those attacks lost money after costs [9]. Applied to the whole period, 68% would leave about $20.2 million [19]. Bitquery says the sample does not establish net earnings for the full period [10], and 63 attacks is under 0.001% of the 6,850,144 counted [20]. We would use $20.2 million only to size the gap between gross and net.
The slippage result is the part a trader can act on, and it has the narrowest coverage of the report's figures. The comparison is observational and covers only Jupiter routes through Pump.fun's launch curve or PumpSwap, and Jupiter carried only a small share of all victim trades [3]. An observational comparison cannot separate the setting from the kind of trade that needs it. A trader who allows a very wide tolerance may be buying a thinner, faster-moving token, and the token could account for part of the exposure.
The total could be wrong in either direction. Bitquery excluded round trips that begin in a token's first 10 slots and reports gaps in exchange and block coverage [12], and inconsistent Pump.fun sale-fee recording affects the gross estimate by about 3% [11]. Against that, it identified bots from repeated trading patterns [12], and chain records cannot prove a bot saw an order beforehand, so some qualifying wallets may be fast traders [13].
Our view is that the hidden cost is real, that the bots' take was below $29.7 million, and that the slippage finding is the better guide to action. The counter-case is the undercount from the exclusions above. A net-of-cost sample far larger than 63 attacks [9], showing bots keeping close to all of the gross, would break the second claim. A comparison outside Jupiter routes in which the 14x ratio [1] shrinks toward 1 would break the third.
What to watch
- A net-of-cost estimate for the full 11 months; the 68% retention in a 63-attack sample is the only cost figure so far.
- A slippage comparison on routes outside Jupiter, since the 14x ratio rests on a router that carried a small share of victim trades.
- Whether the September 23, 2026 preprint's finding on protected order flow is tested against Pump.fun routes specifically.
Clarity's read
What the record supports and how the coverage leans. The claims behind it follow.
Reality
- Evidence55
- Adoption
- Insufficient
- Hype gap−10
- Incentives
- Insufficient
- Confidence50
Claim ledger
Ranked by verification strength, evidence, and original report placement.
- [1]
Among trades sent by Jupiter through Pump.fun's launch curve or PumpSwap, trades allowing 20%-50% slippage were sandwiched 14 times as often as those allowing under 3%.
ReportedSupportedSource: Bitquery, as reported by CryptoSlate2 sources— create a free account to open themView cited source - [2]
Attack rates did not rise uniformly with every wider slippage setting: trades below 0.5% tolerance were hit more often than those at 1%-3%, and the 50%-or-more band was hit less often than the 20%-50% band.
ReportedSupportedSource: Bitquery, as reported by CryptoSlate2 sources— create a free account to open themView cited source - [3]
The slippage comparison is observational and covers Jupiter routes through Pump.fun's launch curve or PumpSwap; Jupiter carried only a small share of all victim trades in the investigation.
ReportedSupportedSource: Bitquery, as reported by CryptoSlate2 sources— create a free account to open themView cited source - [4]
Sandwich bots extracted about $29.7 million before costs from Solana trades over an 11-month period, according to blockchain data provider Bitquery.
- [5]
Bitquery found 96% of detected attacks on memecoin platform Pump.fun's launch curve and its PumpSwap exchange.
- [6]
The investigation, covering successful swaps, counted 6,850,144 attacks by 2,545 bot wallets affecting 2,154,994 distinct wallets between October 30, 2025, and October 7, 2026; those addresses do not represent a count of individual people.
- [7]
Other bots' trades accounted for 39% of the estimated gross take.
- [8]
The $29.7 million estimate values SOL at the minute of each attack and does not measure net profit.
- [9]
In Bitquery's separate 63-attack fee sample, bots retained 68% of aggregate gross proceeds, and 18 attacks lost money after costs.
- [10]
The 63-attack sample does not establish net earnings for the full period.
- [11]
The report identifies inconsistent Pump.fun sale-fee recording that affects its gross estimate by about 3%.
- [12]
Bitquery used repeated trading patterns to identify bots, excluded round trips beginning in a token's first 10 slots, and reports gaps in exchange and block coverage.
- [13]
Chain records cannot prove a bot saw an order beforehand, and some qualifying wallets may be fast traders.
- [14]
Slippage tolerance sets how far execution can depart from a quote; a price move beyond the limit can cause a swap to fail, and a wider limit permits a worse fill.
- [15]
A research preprint submitted September 23, 2026 examined protected order flow across six blockchains; its authors found sandwich patterns extending across trades and blocks and concluded these protections can offer weaker guarantees than users expect.
- [16]
$29.7 million of gross extraction over 6,850,144 attacks is about $4.34 per attack on average.
- [17]
39% of the $29.7 million gross, the share from other bots' trades, is about $11.6 million.
- [18]
Subtracting the other-bot share leaves about $18.1 million of the gross on trades not attributed to other bots.
- [19]
If the 68% retention seen in the 63-attack sample held across the whole period, bots would have kept about $20.2 million of the $29.7 million gross; illustrative only.
- [20]
The 63-attack sample is about 0.0009% of the 6,850,144 attacks counted, under 0.001%.
Sources
1 independent publisher whose own reporting we read for this story.
- cryptoslate.comSolana sandwich bots took $29.7 million before costs, mostly targeting Pump.fun
1 article · October 12, 2026
Topics and entities
Follow any of these and your For You feed starts watching them — no settings page required.