Worked example · 2021-20

Slippage-manipulation sandwich MEV cohort — EVM (primary), Solana — 2021–2025

Loss: Unquantified per-sandwich aggregate. Individual sandwich extractions range from tens to hundreds of ETH-equivalent for large high-slippage swaps on Uniswap V2/V3 and PancakeSwap; the per-sandwich extraction scales with the victim's authorised slippage band — a 5% slippage tolerance on a $100K swap authorises up to $5K of value to be extracted by an attacker who pushes the pool price to the amountOutMin boundary. No single headline-grabbing loss figure anchors the class because the extraction is distributed across thousands of individual sandwiches rather than concentrated in a single incident. EigenPhi bundle-level data records the per-sandwich extraction amounts across the MEV ecosystem.
OAK Techniques observed: OAK-T9.013 (Slippage-Manipulation Sandwich Attack) — canonical anchor case. Distinct from OAK-T5.004 (Sandwich MEV Extraction) at the profit-mechanism level: T5.004 profits from the bid-ask spread of the sandwiched trade; T9.013 profits from the victim's slippage-tolerance parameter — the victim's amountOutMin is the attacker's profit boundary. The two Techniques co-occur in practice; the distinguishing signal for T9.013-specific extraction is whether the victim's output amount exactly equals the amountOutMin (boundary reached — T9.013) versus exceeding it marginally (standard spread-only sandwich — T5.004).
Attribution: pseudonymous Known MEV bot addresses (jaredfromsubway.eth and equivalents on Ethereum mainnet) execute slippage-manipulation sandwiches; the bot addresses are on-chain attributable and the sandwich bundles are identifiable at the block-builder layer. The human operator behind the bot address is typically not identified.
Key teaching point: The slippage-tolerance parameter — intended as a protection against adverse price movement — is the attacker's profit surface. The user who sets a wide slippage tolerance to ensure their trade executes under volatile conditions inadvertently authorises the MEV searcher to extract up to the full slippage band. The wallet-side default-slippage setting and the user-education surface ("your slippage tolerance of X% allows up to Y ETH in extraction") are the load-bearing mitigations for T9.013.

Summary

Slippage-manipulation sandwich attacks exploit the user-configurable slippage-tolerance parameter on AMM swap transactions. The victim's swap transaction includes amountOutMin (the minimum output amount the user will accept) or sqrtPriceLimitX96 (the price limit for the swap); the difference between the current pool price and the user's limit defines the authorised slippage band. A wide slippage band — chosen by the user to ensure execution under volatile conditions or simply left at the wallet/DEX default — defines a large profit surface for an MEV searcher who can observe the transaction in the mempool.

The attack sequence: (1) the attacker observes a swap transaction in the mempool with a wide slippage tolerance; (2) the attacker front-runs the victim with a trade that moves the pool price to the exact boundary of the victim's amountOutMin; (3) the victim's swap executes at the worst-acceptable rate; (4) the attacker back-runs with a trade that unwinds the position at the post-victim-swap price. The attacker's profit is the full slippage band — the victim receives exactly the amountOutMin they specified, never more. The victim's "protection" (the minimum output amount) was precisely honoured; the attacker extracted the difference between the pre-front-run fair price and the boundary price.

The pattern is operationally prevalent across Uniswap V2/V3 (Ethereum), PancakeSwap (BSC), and increasingly on Solana via Jupiter DCA/limit-order slippage surfaces. Known MEV bots execute the sandwich at scale; the bot's profit per sandwich scales linearly with the victim's authorised slippage band and the victim's trade size.

Timeline (UTC)

When	Event	OAK ref
2020–2021	MEV sandwich literature characterises the slippage-parameter-as-attack-surface dimension; Uniswap V2 token-deployment wave creates the first operational slippage-manipulation sandwich surface	T9.013 (slippage-band surface)
2021–2025	jaredfromsubway.eth and equivalent MEV bots execute slippage-manipulation sandwiches on Uniswap V2/V3 at scale; EigenPhi bundle-level data records per-sandwich extraction amounts	T9.013 + T5.004 (spread + band)
2021–2025	PancakeSwap high-slippage sandwiching on BSC; BSC's transparent mempool and short block times make high-slippage swaps economically attractive targets	T9.013 (BSC surface)
2023–2025	Solana Jupiter DCA/limit-order slippage surface exploited; MEV searchers target wide-slippage Jupiter orders on Solana; the absence of a public mempool shifts observation to the leader-scheduling layer	T9.013 (Solana surface)
2024–2025	Wallet-side slippage warnings begin to deploy (MetaMask, Rabby, Phantom); auto-slippage calculation based on pool depth is discussed but not yet standard	(defender-side maturation)
Continuing	Slippage-manipulation sandwiches remain an operational MEV extraction pattern across all major AMM chains; wallet-side default-slippage reduction is the highest-leverage per-user mitigation	(ongoing surface)

Realised extraction

Distributed across thousands of individual sandwiches. Per-sandwich extraction scales with the victim's authorised slippage band: a 5% slippage tolerance on a $100K swap authorises up to $5K in extraction. The aggregate extraction across the MEV ecosystem is material but not concentrated in a single headline-grabbing incident. EigenPhi bundle-level data is the primary source for per-sandwich extraction quantification.

Public references

See citations in corresponding technique file.

What this example tells contributors writing future Technique pages

The slippage tolerance is the attacker's profit ceiling. Future T9.013 worked examples should record the victim's authorised slippage band and the attacker's realised extraction as paired values — the extraction should not exceed the band (the victim's amountOutMin bounds the attacker's profit), and the extraction approaching the band is the T9.013-specific distinguishing signal.
T9.013 and T5.004 co-occur — the distinguishing signal is the extraction-to-band ratio. A sandwich where the victim's output amount exactly equals the amountOutMin is a T9.013 signal; a sandwich where the victim's output amount exceeds the amountOutMin reflects a standard T5.004 spread-only extraction. Future worked examples should record both the amountOutMin and the victim's realised output amount to distinguish the two.
Wallet-side slippage warnings are the load-bearing mitigation. Future T9.013 mitigation work should focus on wallet-side pre-trade simulation that displays the maximum extractable value at the user's chosen slippage tolerance.

References

[daian2019flashboys] — Daian et al., "Flash Boys 2.0" (2019): foundational MEV characterisation.
[zhou2023sok] — academic taxonomy covering MEV extraction and slippage-manipulation surfaces.
EigenPhi bundle-level MEV data (slippage-band sandwich detection and per-sandwich extraction quantification).
Flashbots MEV-Boost relay data (bundle-level transparency for slippage-manipulation sandwiches).
Wallet-vendor documentation: MetaMask slippage warnings, Rabby pre-trade simulation, Phantom transaction preview.