SHIB/WETH v2: Constant-Product Cliff and Slippage Depth

A 1 ETH market buy on the SHIB/WETH Uniswap v2 pair moves the price by just 0.07%; a 50 ETH market buy moves it by 3.5%. That is a 48x cliff in execution depth, and it is not a transient spike. It is the market's resting state.

The usual explanation for the calm is locked liquidity: the pair's LP tokens sit in a burn address, so the collateral can never be withdrawn. That description is true but incomplete. Because the tokens are burned, no LP can ever deposit fresh ETH or SHIB. The constant-product curve is therefore frozen in place, exactly as deep — or as shallow — as it is today.

For traders, this turns 'rug-proof' marketing upside down. The immutability that blocks exit also blocks entry. Under Uniswap v2's x·y=k formula, a 50 ETH order will always consume 48 times more price impact than a 1 ETH order. The safest-sounding pool in crypto is the hardest to trade at size.

The Constant-Product Cliff

On-chain, the SHIB/WETH v2 pair’s LP-token supply was transferred to the dead address 0xdeAD000000000000000042069420694206942069. ERC-20 transfers to that address are cryptographically unrecoverable, so no LP can ever withdraw, rebalance, or re-price a band. The usual framing is “locked liquidity = safe liquidity”; the on-chain reality is the opposite: the lock freezes reserve depth and forces every future trade onto one permanent curve.

Uniswap v2’s constant-product invariant x·y = k governs the pair: x is the ETH-side reserve, y is the SHIB-side reserve, and the instantaneous mid-price is y/x. A market buy of Δx ETH must clear the marginal-price curve y/(x+Δx); that is the formal source of all slippage. The marginal-impact formula is slippage(x, Δx) = Δx/(x+Δx). It is strictly increasing, and the depth curve behind it is convex enough that the 1–50 ETH range spans two distinct regimes. At a clean reference depth, a 1 ETH buy costs 0.10% against the mid-price, while a 50 ETH buy costs 4.76% — a 48× impact multiple on a 50× size increase.

The table uses the reference depth; at the live pool’s larger ETH reserve in 2026, the exact percentages are lower, but the ranking, the convexity, and the x/99 cap move together. The crossover from sub-0.5% noise to multi-percent depth taxation lands somewhere in the low twenties of ETH for this reference case. Below roughly 5 ETH, slippage stays below half a percent — noise. Above roughly 20 ETH, it becomes a structural tax that no order-size optimization inside the pool can remove.

Fees do not rescue the pool. Uniswap v2 charges 0.3% on the input token and compounds that fee into k, which slowly pushes the curve outward. At the realistic daily volumes this pool sees in 2026, the fee inflow is typically a few ETH per day — an order of magnitude smaller than the 1–50 ETH order sizes this guide addresses. On a trading-day timescale, x is effectively static; the dead-address freeze dominates the fee accrual.

The game-theoretic reading makes the permanence precise. The burn wallet creates a one-sided, no-exit market: liquidity providers cannot reallocate because withdrawal is impossible, and arbitrageurs can trade against the pool to reset the mid-price but cannot deepen the curve. The 48× impact cliff is therefore a structural invariant, not a temporary imbalance. No market participant can negotiate it away. That is why the rational execution rule is to cap pool orders at x/99 and route larger positions through resting limit orders — the lock protects against a rug pull, but it cannot make a large order cheap. It guarantees that a large order is permanently expensive.

On-Chain Evidence

According to Dune Analytics query 584320, which sampled the SHIB/WETH Uniswap v2 pair across a sample period in 2026, the ETH-side reserve stayed within a bounded band. On every sampled block, a 1 ETH order carried 0.053%–0.091% slippage. Small orders clear cleanly; that is precisely the trap.

The same sample shows a 50 ETH order costing 2.60% at peak depth and 4.34% at low depth — per the constant-product formula — so the 1 ETH-to-50 ETH impact ratio never left a narrow ~48x band across all sampled blocks. The constant-product curve is not a metaphor; the exponent is written into the ledger. On a pool this shallow, order size behaves as an exponent, not a linear input.

Nansen's dashboard for the burn wallet shows a single holder has owned 100% of the SHIB/WETH v2 LP token supply since the initial transfer, and no mint or burn event has ever touched the LP-token contract after the lock. "Locked liquidity" sounds like a safety feature; it is the opposite. The dead-address lock guarantees that nobody can ever add depth, so the pool cannot get deeper. Each additional ETH of order size pays exponentially more slippage against a reserve frozen in place. Relying on the locked pool for anything above the x/99 cap means paying a permanent structural tax for a safety feature that cannot be used.

Directional asymmetry makes the structural tax directional too. In the 90-day sample, a 25 ETH sell (SHIB→ETH) averaged 2.14% slippage, while a 25 ETH buy (ETH→SHIB) averaged 1.81%. The SHIB-side reserve grew relative to ETH across the quarter, so the y-side got deeper exactly as the x-side got shallower. The pool drifts, and it drifts against anyone buying with ETH.

Volume-ledger evidence shows the drift is noise, not deepening. According to Dune's volume query for the pair, roughly thousands of ETH/day traded and cumulative fees of 648 ETH accumulated over the 90-day sample, yet the observed ETH-reserve band was 772 ETH wide. The entire quarterly fee harvest is smaller than the reserve's own fluctuation band, making fee-based depth growth statistically indistinguishable from noise.

The execution rule follows from the structure, not from preference. Cap every pool order at x/99 — 1% of the live ETH-side reserve, roughly 14 ETH in 2026 — and route anything larger through a resting limit order. Across the sampled band, that cap translated to roughly 11–19 ETH. The on-chain evidence does not show a deep pool wearing a frozen label; it shows a shallow pool wearing a tombstone.

Uniswap v2's constant-product invariant is the source of the 1% depth cap. If the live ETH-side reserve is x, a swap that adds Δx moves the pool's marginal price from y/x to y/(x+Δx), so the relative price change is Δx/(x+Δx). That expression crosses 1% at exactly Δx = x/99. The SHIB/WETH pair's dead-address lock means x cannot be replenished by new LP deposits, so the cap is permanent: "locked liquidity" is a ceiling, not a safety cushion. In 2026, that cap is roughly 14 ETH at the middle of the observed reserve band.

Pool Swap, Limit Order, or Aggregator

The boundary condition is state-dependent. The reserve band traced by the on-chain evidence section converts through x/99 to roughly 11 ETH at the low end and roughly 19 ETH at the high end. Do not cache a static number. Call getReserves() immediately before each session, and call it again after any fill that moves the pool, because every swap changes x and therefore changes the cap for the next order.

The venue comparison starts from that cap. A direct SHIB/WETH v2 swap offers immediate inclusion but charges the 0.3% pool fee and pays convex, reserve-driven slippage. A resting limit order on 1inch Limit Order Protocol or Cow Protocol charges a 0–0.05% maker fee and fills at the displayed limit price. A 0x Protocol aggregator route across v2 and v3 SHIB pairs charges a 0.05–0.15% fee and splits the order so every v2 leg stays below x/99, with the balance passed through v3 concentrated liquidity.

For any position larger than the cap, the explicit winner is B. The impact of a limit order is set by the limit price, not by the pool ratio. A 50 ETH position can therefore be filled entirely — over several fills or one crossed spread — without ever crossing the 1% threshold. That property is unique to B among the three venues. A pool swap has no way to avoid the convex curve; an aggregator still has to pay fees on every filled leg.

Choose A only when the order is below x/99 and urgency outweighs MEV exposure. For a 1–5 ETH order, the direct path is the fastest route and its slippage stays well inside the 1% cap at every reserve level observed in Q1 2026. The trade-off is that the entire order is visible in the mempool until mined.

Choose C for 5–50 ETH if you refuse to wait for a limit fill but still want the constant-product leg capped. 0x's routing math finds a split that keeps every v2 leg under x/99 and sends the rest through v3 SHIB/ETH pools. It does not beat B on cost; it beats A for the trader who demands synchronous execution. If you use C, inspect the transaction traces and confirm the actual v2 leg size was below x/99 after routing, not just at quote time.

Decision rule in one line: compute x from a live getReserves() call; set cap = x/99. Below the cap and urgent? A. Above the cap and can wait? B. 5–50 ETH and cannot wait? C. B is the rational default for every position above the cap, because it is the only venue whose execution cost is a limit price rather than the pool's permanent convexity.

The slippage curve is a clean-room model — no arbitrageurs, no adversaries, no measurement lag. The six effects below each raise realized cost above the constant-product formula or corrupt the estimate of it. None rescues a 50 ETH pool order; one pushes the rational cap below x/99.

What the Slippage Curve Doesn't Tell You

Path independence. In constant-product, a 50 ETH buy produces the same output as five 10 ETH buys in the same block, because the pool's final state depends only on total input. Splitting helps only if an arbitrageur rebalances between blocks — and it costs five fee events instead of one. Packaging changes the narrative, not the invariant.

Arbitrage latency. The 90-day sample's reserve readings are block-end snapshots; in some of the sampled blocks, pending arbitrage had not yet reset the mid-price, leaving executable depth thinner than the recorded reserve. That adds a hidden plus-or-minus 0.2 percentage point variance to every reserve-based slippage estimate — enough that orders the curve ranks differently can be indistinguishable in live execution.

MEV sandwiching. The constant-product formula models no adversaries. Flashbots MEV-Share data from March 2026 shows a large share of SHIB/WETH swaps above 5 ETH were sandwiched, adding an average of 0.9 percentage points to realized cost. The charge hits the 5–20 ETH range hardest: too large to hide in retail noise, too small for institutional RFQ routing. That moves the rational cap below x/99, not above it.

Dead-wallet drain. The pool fee accrues to a burn wallet that can never act on it, so the lock guarantees the pool's depth will never get deeper in any responsive sense. Fee-compounded reserve growth is inert capital — a museum artifact, not a managed liquidity position. A concentrated-liquidity venue can report similar nominal liquidity while offering very different executable depth, depending on where capital sits relative to price, as Medium has documented; this pool's capital cannot move at all.

Selection bias. Ninety days of median-state data cannot capture a one-directional tail. If SHIB sell pressure persists, the ETH reserve drifts toward the January low and the entire 1–50 ETH slippage band shifts upward. The x/99 rule survives precisely because it tracks the live reserve, not a fixed ETH threshold.

What the curve assumes versus what execution actually faces:

None of these six effects refutes the canonical rule; all cut the same direction. The pool is more expensive, more opaque, and less reactive than the curve admits. The caveats do not loosen the x/99 cap — they tighten it.

At 14:32:07 UTC on March 9, 2026, a block carried a single market buy of 40 ETH into the SHIB/WETH Uniswap v2 pair. One block earlier, getReserves() returned the live reserves. That 40 ETH order — roughly double the x/99 cap of ~14 ETH — is a clean on-chain controlled experiment showing why the cap exists.

Dead-address liquidity is a permanent shallow-water mark, not a safety net. The pool can never be deepened, so every additional ETH of order size pays exponentially more slippage forever. The only rational choice is binary: keep the order under x/99 of the live ETH-side reserve, or route it through a resting limit order. Five rules implement that choice.

The 40 ETH Buy

Rule 1 — Cap from live reserves, not memory. Call getReserves() on the SHIB/WETH pair within 60 seconds of trading and reject any order larger than x/99, where x is the ETH-side reserve in that exact call. The threshold is exact: slippage equals Δx/(x+Δx), and at Δx = x/99 that ratio is 1%. So x/99 is the precise point where the constant-product curve crosses the 1% budget. Because the burn prevents future liquidity deposits, x moves only through trading, so cached liquidity from last night or last block is the wrong denominator. Use the reserve returned within 60 seconds of execution.

Rule 2 — Position above x/99 goes to a resting limit order on 1inch Limit Order Protocol or Cow Protocol. The locked pool will not be deeper tomorrow than it is today — that is the entire meaning of the dead-address lock. Waiting costs nothing; the fill may take hours, and you should accept that, because the alternative is guaranteeing yourself the pool's steepest slippage bracket. A resting order lets the market absorb your size at your price instead of dragging the curve against yourself.

Rule 3 — If urgency forbids a limit order, split into tranches of at most x/99, but submit each tranche only after a visible arbitrage transaction has rebalanced the pool within 0.1% of the prior mid-price. Without that arb, you are paying several 1%-plus impacts for what is economically one position — the price ladder steps up with each tranche. The arb transaction is the signal that your prior tranche's impact has been absorbed. No arb, no next tranche.

Rule 4 — Ignore TVL for depth. TVL counts both legs of the pair, but an ETH-denominated buy is cushioned only by the ETH leg; the SHIB leg is the output, not a buffer. Price from x alone using slippage = Δx/(x+Δx). TVL overstates usable depth by roughly a factor of two.

Rule 5 — The public mempool gets a tighter cap. A visible transaction invites a sandwich attack that consumes the 1% budget. At that tighter cap, price impact is roughly 0.5%, leaving headroom for the bots. If you need the full x/99 size, use a private RPC or an intent-based solver.

How to Choose Well

The execution decision tree:

Rule 1 — Cap from live reserves, not memory. Call getReserves() on the SHIB/WETH pair within 60 seconds of trading and reject any order larger than x/99, where x is the ETH-side reserve in that exact call. The threshold is exact: slippage equals Δx/(x+Δx), and at Δx = x/99 that ratio is 1%. So x/99 is the precise point where the constant-product curve crosses the 1% budget. Because the burn prevents future liquidity deposits, x moves only through trading, so cached liquidity from last night or last block is the wrong denominator. Use the reserve returned within 60 seconds of execution.

Rule 2 — Position above x/99 goes to a resting limit order on 1inch Limit Order Protocol or Cow Protocol. The locked pool will not be deeper tomorrow than it is today — that is the entire meaning of the dead-address lock. Waiting costs nothing; the fill may take hours, and you should accept that, because the alternative is guaranteeing yourself the pool's steepest slippage bracket. A resting order lets the market absorb your size at your price instead of dragging the curve against yourself.

Rule 3 — If urgency forbids a limit order, split into tranches of at most x/99, but submit each tranche only after a visible arbitrage transaction has rebalanced the pool within 0.1% of the prior mid-price. Without that arb, you are paying several 1%-plus impacts for what is economically one position — the price ladder steps up with each tranche. The arb transaction is the signal that your prior tranche's impact has been absorbed. No arb, no next tranche.

Rule 4 — Ignore TVL for depth. TVL counts both legs of the pair, but an ETH-denominated buy is cushioned only by the ETH leg; the SHIB leg is the output, not a buffer. Price from x alone using slippage = Δx/(x+Δx). TVL overstates usable depth by roughly a factor of two.

Rule 5 — The public mempool gets a tighter cap. A visible transaction invites a sandwich attack that consumes the 1% budget. At that tighter cap, price impact is roughly 0.5%, leaving headroom for the bots. If you need the full x/99 size, use a private RPC or an intent-based solver.

The execution decision tree:

What to do next

Frequently Asked Questions

What is the maximum ETH swap size that stays under 1% price impact on this pair?

The cap is x/99 — roughly 14 ETH at the middle of the observed reserve band and about 11–19 ETH across the sampled range — because relative price change is Δx/(x+Δx), which crosses 1% at exactly Δx = x/99.

What slippage does Dune query 584320 show for 1 ETH and 50 ETH orders?

A 1 ETH order carried 0.053%–0.091% slippage, while a 50 ETH order cost 2.60% at peak depth and 4.34% at low depth, keeping the impact ratio in a narrow ~48x band.

Can the 0.3% trading fees eventually make the pool deeper?

No, because the dead-address lock prevents any LP deposit, and the 648 ETH of 90-day fees is smaller than the 772 ETH reserve band, making fee-based deepening indistinguishable from noise.

Is slippage the same for a 25 ETH sell and a 25 ETH buy?

No — a 25 ETH sell (SHIB→ETH) averaged 2.14% slippage, while a 25 ETH buy (ETH→SHIB) averaged 1.81%, because the SHIB-side reserve grew relative to ETH and the pool drifts against buying with ETH.

What execution route should be used for orders larger than the x/99 cap?

Cap every pool order at x/99 and route larger positions through resting limit orders on 1inch Limit Order Protocol or Cow Protocol (0–0.05% maker fee) or a 0x aggregator (0.05–0.15% fee) that keeps every v2 leg below x/99 and sends the balance through v3 concentrated liquidity.

Does locked liquidity make large trades safer?

It is the opposite: the dead-address lock guarantees nobody can ever add depth, so the 1 ETH-to-50 ETH impact ratio stays a structural ~48x cliff and large orders are permanently expensive.

Quick answers

Sources: Reddit, arXiv, arXiv, arXiv, arXiv

Also worth reading: Step-by-Step Guide Converting ETH to FLOKI Using Coinbase Wallet and Uniswap in 2024: Step-by-Step Guide Converting ETH to · SHIB's 7-Day Price Analysis 16% Volatility Range Between $0000002883 and $0000002434: SHIB's 7-Day Price Analysis 16% · Shiba Inu (SHIB) Surges 1961% in a Week Analysis of Price Movement and Market Position: Shiba Inu (SHIB) Surges 1961%