Why now

The timing argument: two academic results made in-range IL priceable and hedgeable, and Arbitrum + Chainlink + Stylus made the exact on-chain pricer buildable. The anchors are theory that validates the approach — the Phi-sum is the pricer.

Inflexion could not have been built — as a priced, fully collateralized, non-custodial market — three years ago. Two things had to become true at once: the theory of in-range impermanent loss had to mature from actuarial guesswork into something with a model-light fair value and a concrete hedge, and the infrastructure had to be able to evaluate the resulting closed form on-chain, live, per quote, at a precision that does not leak.

Both happened. This page is the timing argument.

The single distinction this whole page turns on: the two academic results below are theory anchors for why in-range IL is priceable and hedgeable. They are not the pricer. The pricer is an exact closed-form Φ-sum computed on-chain by the Stylus FairValueOracle. Do not conflate the validation with the machinery.

The two theory anchors

The case for a priced, collateralized IL market only became rigorous recently. Two results — one on the cost of the LP’s exposure, one on the hedge for it — turn “LPs lose a lot to IL” into “this specific risk has a model-light fair value and a concrete replication.” They are the reason the approach is sound; they are not how the protocol computes a number.

Lipton, Lucic & Sepp (2025)

An IL-protection claim is statically replicable by a strip of vanilla options — so it has a model-light fair value and a concrete hedge. This is the anchor for why the claim is priceable and hedgeable.

Milionis, Moallemi & Roughgarden (2022)

Automated Market Making and Loss-Versus-Rebalancing (LVR). The AMM’s adverse-selection cost has a closed form proportional to instantaneous variance — the theta of the replicating short option. The anchor for the cost of short-gamma exposure.

Anchor 1 — the hedge exists: static replication (Lipton–Lucic–Sepp, 2025)

The structural problem with IL has always been that it felt like an exotic, path-heavy payoff that you could only reserve against actuarially — guess a distribution, charge a fat uncertainty premium, hope. Lipton, Lucic & Sepp showed that an IL-protection claim is statically replicable by a strip of vanilla options. Static replication means the hedge is a fixed portfolio set at inception — you do not rebalance into the storm.

Two consequences fall straight out, and both are load-bearing for Inflexion:

• It has a model-light fair value. If the payoff is a combination of vanillas, its fair value is the same combination of vanilla prices. That is what makes a closed-form premium even conceivable.
• It has a concrete hedge. A market maker who underwrites the claim is not stuck holding naked short gamma; they can export the risk to the global options market. This is precisely what Path B market makers do — and why a hedged MM makes the whole system’s risk smaller rather than just relocating it.

Anchor 2 — the cost is closed-form: LVR (Milionis–Moallemi–Roughgarden, 2022)

The complementary result quantifies what the LP is paying for being in the pool. Milionis, Moallemi & Roughgarden showed that the AMM’s adverse-selection cost — loss-versus-rebalancing (LVR) — has a closed form proportional to instantaneous variance, equivalently the theta (time-decay) of the replicating short-option position.

This is the precise framing of the problem Inflexion solves. A Uniswap v3 LP holds a position whose value V_lp(P) is strictly concave in price in range — the LP is structurally short gamma, and IL is the realization of that short-gamma / LVR cost. The concavity is not hand-waving; it is the second derivative of the in-range value function:

$$V_{lp}(P) = L\left(2\sqrt{P} - \sqrt{P_a} - \tfrac{P}{\sqrt{P_b}}\right), \qquad \frac{d^2 V_{lp}}{dP^2} = -\tfrac{1}{4}\,L\,P^{-3/2} < 0 .$$

A concave value function is a short-option payoff. The LP pays LVR (theta) for it. What Inflexion sells is the long-convexity leg that offsets it — capped at the in-range maximum, MaxIL — priced as the exact risk-neutral expectation of that capped payoff. The two anchors meet here: LVR says short gamma has a closed-form cost; static replication says the offsetting claim has a model-light fair value and a hedge. Inflexion is the market that connects them.

Read these two anchors as one sentence: a v3 LP is short gamma with a closed-form cost (LVR), and the hedge for it is a strip of vanillas (static replication). That is the entire theoretical license to build a priced, collateralized IL market. The theory-anchors page carries the full derivation chain.

Why the anchors are NOT the pricer

This is a seam worth stating plainly.

The anchors say the claim can be priced and hedged. They do not give Inflexion the number it puts on-chain. Inflexion does not reimplement a strip-of-vanillas hedge to back into a price, and it does not run Monte Carlo or look up a calibrated surface. It computes the exact risk-neutral expectation of the actual capped v3 payoff in closed form.

The capped payoff min(IL, MaxIL) is piecewise in the terminal price — a constant arm, a linear-in-P arm, and a √P arm, split by the range edges and the two cap-crossing prices. Each arm integrated against the lognormal density of P_T is a standard interval moment in the normal CDF Φ. So the fair premium is a finite Φ-sum (≈6–10 terms, Black–Scholes class):

$$\text{FairPremium} = \mathbb{E}_{\mathbb{Q}}\!\left[\min(IL,\,\text{MaxIL})\right] = \text{fairRate} \cdot \text{MaxIL}, \qquad \text{fairRate} = \frac{\mathbb{E}_{\mathbb{Q}}[\min(IL,\text{MaxIL})]}{\text{MaxIL}} .$$

No fitted coefficients, no lookup table, no Monte Carlo — evaluated live per quote, with σ_ref as the only stochastic input. The full arm-by-arm construction is on the Φ-sum page.

Limit → Why → Mitigation: the only residual approximation. The Φ-sum is exact given the model. Limit: the one modeling assumption is risk-neutral GBM with μ = 0, r = 0. Why: no on-chain formula can remove the GBM assumption — that would require a forward-vol view the chain cannot see. Mitigation: it is covered two ways — a deliberately conservative σ_ref (a max(σ_short, σ_long, floor) with the floor set above calm ETH vol), and the residual forward-vol premium is left as MM alpha, the very incentive that keeps forward-looking-vol market makers in the two-sided market.

So the lineage is: anchors validate → Φ-sum prices. The anchors establish that the approach is theoretically clean. The Φ-sum is what actually clears a trade.

The infrastructure enablers

Theory is necessary but not sufficient. A model-light fair value is worthless if you cannot evaluate it on-chain, live, and precisely enough that the price does not leak. The second half of “why now” is that the chain finally can.

Arbitrum One

The deepest Uniswap v3 liquidity, where the LPs who need this product actually are.

Chainlink

Fully deployed price feeds — the settlement price source and the input to the realized-vol oracle.

Stylus

Rust/WASM smart contracts that evaluate the compute-heavy Φ-sum at a precision Solidity cannot match cheaply.

Arbitrum One — the liquidity is here

Arbitrum One has the deepest Uniswap v3 liquidity of any rollup. The product hedges a specific v3 position’s in-range IL, so it has to live where the positions are. Building on the chain with the largest population of in-range v3 LPs is the difference between a demo and a market.

Chainlink — the settlement and volatility input

Settlement reads the Chainlink round pinned at expiry T and pays min(realized_IL, MaxIL) from pre-locked collateral; the realized-volatility oracle (σ_ref) is built from EWMAs of Chainlink log-returns at two horizons. Both depend on Chainlink being fully deployed and live on Arbitrum. The oracles page covers the staleness, sequencer-uptime, and lone-spike defenses.

Limit → Why → Mitigation: the testnet has no Sequencer Uptime Feed. Limit: Arbitrum Sepolia — where Inflexion is live today — has no Chainlink L2 Sequencer Uptime Feed. Why: it is a testnet with no SLA; the feed simply is not deployed there. Mitigation: OracleManager skips the sequencer check on testnet by design, and this check is gated to be re-enabled before any mainnet deploy. It is a known, parameterized switch, not an oversight. See oracles.

Stylus — the exact Φ-sum, on-chain

This is the enabler that turns the theory into a trustless product rather than an off-chain quote you have to believe. The Φ-sum is compute-heavy — it leans on the error function — and Solidity’s Abramowitz–Stegun erf is only accurate to ~1.5e-7, which when divided through by MaxIL amplifies to ~1e-3 at the tight tier. That is enough error to be picked off on a price you publish.

Arbitrum Stylus lets the protocol write the oracle in Rust/WASM with integer fixed-point arithmetic — and because Stylus bans WASM floating point at activation, the result is deterministic and machine-precise. The Stylus FairValueOracle matches a 50-digit mpmath reference to ≤ 1e-12 on fairRate (in practice a few wei at 1e18 scale). That precision is why the on-chain published price can be trusted as the fair value rather than a loose estimate.

A precise nuance worth owning: Stylus is production for the compute-heavy FairValueOracle — not everywhere. The tiny ILMath kernel was benchmarked in Stylus and measured ~5.3× more expensive cached, because there is too little compute to amortize Stylus’s per-call overhead. So Solidity ILMath is production; the Stylus ILMath port is a rejected benchmark. “Stylus is faster” is not a blanket claim — it is true exactly where the compute is heavy.

A maturing options surface — the risk has somewhere to go

The final enabler is external to the chain. Static replication only matters if the replicating vanillas trade. The off-chain options market (Deribit, which holds >90% of ETH options) and on-chain options venues (Panoptic) now exist as real venues where a Path-B market maker can export the short-gamma risk it underwrites.

This makes Pillar 3’s risk-export argument concrete rather than aspirational: without a place to hedge, a closed pool of IL writers is just ETH short gamma circulating against itself; with a maturing options surface, hedged MMs make the system’s net risk smaller. The theory anchor (static replication) and the infra enabler (a venue to replicate in) are two ends of the same hedge.

Putting it together

1. The cost became closed-form (LVR, 2022)

   A v3 LP is short gamma; the cost of that exposure is closed-form, proportional to instantaneous variance. The problem is now quantified, not just observed.

2. The hedge became model-light (static replication, 2025)

   The IL-protection claim is a strip of vanillas — so it has a fair value and a concrete hedge. The solution is now priceable and hedgeable, not actuarial guesswork.

3. The chain could finally evaluate it exactly (Stylus + Chainlink, on Arbitrum)

   Arbitrum has the liquidity, Chainlink supplies the settlement and vol inputs, and Stylus computes the exact Φ-sum on-chain at machine precision. The price is now published and trustless.

4. The risk has somewhere to go (a maturing options surface)

   Deribit and Panoptic give market makers a venue to export short gamma — so hedged MMs shrink the system’s net risk instead of relocating it.

Each beat was missing as recently as a few years ago. Together they make a priced, fully collateralized, non-custodial in-range IL market not just possible but the obvious next primitive — which is the whole why-inflexion thesis.

What “now” does NOT mean

Honest scope, because the timing argument is easy to over-sell.

It does not mean the math is novel.

The IL formula, the convexity proof, and the lognormal interval moments are textbook once you have the anchors. Inflexion’s contribution is the synthesis: pricing the exact capped v3 payoff in closed form and publishing it on-chain at machine precision, with a two-rail market around it. The novelty is the product and the dataset, not a new theorem.

It does not mean the no-bad-debt guarantee leans on the timing.

The no-bad-debt guarantee is structural and oracle-independent: in FULL mode collateral = MaxIL ≥ payout by construction, regardless of what σ was or what the anchors say. The theory anchors and σ_ref are load-bearing for pricing and depositor solvency, never for the FULL no-bad-debt invariant. The two are separate claims — see the disclosures below.

It does not mean the data moat is mature on day one.

The timing makes the architecture of the data moat possible now; the dynamics (the pool-vs-MM forward-vol spread) mature with multiple competing market makers and real flow. The structures ship from day one; the dataset begins at the redeploy that ships the moat events. See the maturity disclosure below.

LPs are always paid with no bad debt in FULL mode — qualified: capped payoff + solvent USDC + oracle/settlement liveness.

The no-bad-debt guarantee is exact only under the full clause:

• FULL collateralization — the underwriter has posted exactly MaxIL per swap, locked for the life of the position.
• Capped payoff — the payout is min(realized_IL, MaxIL). The cap is load-bearing: it is what makes payout ≤ collateral hold by construction.
• Solvent USDC — the collateral numeraire (dUSDC) holds its peg and remains redeemable.
• Oracle / settlement liveness — the price oracle stays live and settle() runs within the protocol’s settlement window.
• No rehypothecation breach — locked collateral stays instantly liquid; it is never routed to utilization-gated venues.

Stated without these qualifiers, the guarantee is not correct.

Depositor capital is NOT guaranteed: the junior tranche is first-loss; the senior tranche is structurally protected from underwriting loss only, not systemic tail.

Underwriting capital — whether supplied to the pooled cvAMM vault (Path A) or posted directly by a market maker (Path B) — is at risk:

• Junior tranche — first-loss. Junior depositors absorb underwriting losses first and buffer the senior tranche.
• Senior tranche — protected from underwriting loss, exposed to the systemic tail. The senior tranche is structurally shielded from ordinary underwriting loss while junior capital buffers, but it still takes the systemic tail once the junior buffer is exhausted.

No-bad-debt is a guarantee to the LP being paid (see the FULL-mode payout clause), not a guarantee that depositor or MM capital is preserved.

Structures ship from day one; the full DYNAMICS (pool-vs-MM spread as a forward-vol signal) mature with multiple competing MMs and volume. The dataset begins at the single redeploy; the off-chain telemetry halves begin at the first interaction logged.

What Inflexion offers today is the architecture and the first public view into the microstructure of the DeFi LP volatility-risk premium — not yet a mature dataset:

• Structures are present day-one. All five non-circular behavioral signals are defined and queryable from launch.
• Dynamics mature with volume. The richer signal — the pool-vs-MM spread read as a forward-vol indicator — only sharpens once multiple market makers compete and real flow accumulates.
• The dataset begins at the single redeploy, and the off-chain telemetry halves begin at the first interaction logged. History is served live from the subgraph via the API; if a backing (subgraph URL / RPC / telemetry) is absent, surfaces degrade to a typed pending state — never an error.

Where to go next

The theory anchors, in full

The LVR and static-replication results derived and tied to the protocol’s math.

The Φ-sum (the actual pricer)

The exact closed-form premium, arm by arm — what the anchors are not.

The three pillars

On-chain FairValue, the cvAMM floor, and MM competition — the conceptual spine.

Why Inflexion

The problem, the surplus, and why this is the next primitive.