← All articles

JIT Liquidity: Two Economic Regimes, One Block

@0x3hu79
@0x3hu79
April 6, 2026 · MEV Research
#MEV#JIT-liquidity#PBS#block-building#Uniswap#order-flow

Just-In-Time (JIT) liquidity is one of the more elegant MEV strategies in DeFi: a searcher adds a concentrated liquidity position immediately before a large swap, captures fees from that swap, and removes the position immediately after - all within a single block. It’s precision market making, compressed into one atomic sequence.

A new study by Koshi Ota (University of Tokyo) and Davide Rezzoli (PBS Foundation) provides the first comprehensive empirical mapping of JIT liquidity under PBS, analyzing approximately 442,000 JIT bundles on Uniswap v3 from January 2024 to September 2025. The dataset includes around 52,700 private bundles - JIT activity that never appeared in the public mempool.

The findings reveal that public and private JIT execution aren’t just different channels - they’re different economic regimes, with measurably different profit-sharing dynamics between searchers and builders. Here’s what the data shows.

How JIT Liquidity Works

Uniswap v3 allows liquidity providers to concentrate their capital in specific price ranges. A JIT provider exploits this by:

  1. Mint: Adding a very tightly concentrated liquidity position around the price where a large incoming swap will execute
  2. Swap: The large swap executes, paying fees to all LPs in range - but the JIT position is so concentrated it captures a disproportionate share
  3. Burn: Immediately removing the position, exiting with the earned fees

The entire Mint-Swap-Burn sequence happens within one block. The JIT provider’s position exists for the duration of exactly one trade.

This works because fee distribution in Uniswap v3 is proportional to liquidity in range. A JIT provider who knows the exact execution price can provide far more concentrated liquidity than any passive LP - and therefore capture far more of the swap fee.

The flip side: passive LPs, who provide liquidity continuously through quiet and active periods alike, see their fee income diluted during the highest-value trades. The paper characterizes JIT as “fee sniping” against passive LPs - they bear the adverse selection risk while JIT providers capture the premium moments.

Public vs. Private: Two Different Games

The study classifies each JIT bundle as public (the underlying swap was visible in the mempool) or private (none of the bundle’s transactions appeared in the mempool). The classification uses Flashbots’ Mempool Dumpster as the reference dataset.

Public JIT: competitive, high pass-through

Public JIT accounts for roughly 90% of all observed bundles. Multiple searchers can see the same swap in the mempool and compete to JIT it. This competition drives up the fees searchers are willing to pay builders.

The regression analysis quantifies this: on public routes, for every $1 of realized profit (both fee income and price impact components), approximately $0.75 flows to the builder as payment. That’s a 75% pass-through rate - consistent with competitive auction dynamics where searchers bid away most of their surplus.

Private JIT: exclusive, attenuated pass-through

Private JIT represents roughly 10% of bundles but tells a different economic story. The swap is not visible in the public mempool - the JIT searcher has access to it through private channels (direct builder APIs, relay-based submission).

Here, the economics shift dramatically. For the price impact component, only about $0.32 per dollar of realized profit passes through to the builder - less than half the public rate. The searcher retains substantially more.

The intuition is straightforward: with exclusive or semi-exclusive access to the flow, there’s less competitive pressure. The searcher doesn’t need to bid aggressively because there aren’t five other searchers trying to JIT the same swap.

The gap is in the slopes, not a flat discount

An important nuance: the difference between public and private isn’t a uniform per-bundle discount. Once token-pair composition is controlled for, there’s no statistically significant flat “private discount.” The difference shows up in the slopes - how much of each dollar of profit gets passed through. Private routing changes the profit-sharing relationship, not the base price.

Who’s Doing This

The JIT landscape is remarkably concentrated.

The searchers

jared (jaredfromsubway.eth) dominates public JIT - 69% of all bundles in the filtered dataset. jared operates exclusively through public channels: zero private JIT bundles detected. The primary builder relationship is with Beaver (153,000+ bundles), with Titan as the secondary destination. The Beaver-jared pair averages $238 in total fees per bundle - the highest among all pairs.

SCP dominates private JIT - 100% of private JIT in the dataset originates from SCP. But SCP also operates publicly (roughly 38,000 public bundles). The private activity routes to multiple builders: Beaver receives 50.8%, with Titan (15.8%), Flashbots (7%), BuilderNet operators (6.9%), and Nethermind (5.8%) receiving the rest. This multi-builder routing suggests strategic diversification rather than dependence on a single partner.

The contrast between jared and SCP is structural: jared competes openly and pays more per bundle through competitive bidding. SCP routes privately and retains more profit by avoiding the auction.

The builders

Beaver is the dominant builder for JIT activity. It handles the most bundles from both jared (public) and SCP (private). When weighted by fee contribution and pair ratios, Beaver captures roughly four times more total JIT fee volume than Titan.

Interestingly, the paper notes that searchers tend to pay Titan lower fees than Beaver for comparable JIT activity. The authors suggest this reflects strategic bidding behavior or different deal structures, not lower builder quality.

Flashbots, BuilderNet operators (Flashbots, Beaver, Nethermind), and Rsync round out the builder set, each processing meaningful but smaller JIT volumes.

Integrated Pairs Pay More

The regression includes a dummy variable for known searcher-builder integration (cases where the searcher and builder have a documented operational relationship). The result: integrated pairs pay approximately $47 more per bundle, conditional on the same realized profits.

This suggests that vertical integration between searchers and builders changes the economic relationship. Integrated pairs may have tighter coordination, better execution guarantees, and less principal-agent friction - allowing the searcher to commit to higher fee payments because the execution is more reliable.

The implication for the broader builder market: builder-searcher integration isn’t just about convenience. It’s a measurable economic advantage that generates higher fee flow - reinforcing the concentration dynamics that the blockspace market literature has been flagging.

What This Means at the System Level

The privatization of JIT entrenches builder asymmetries

The paper’s conclusion is direct: reliable JIT execution requires privileged builder access. As a result, value capture concentrates in a tight builder-searcher network. Private JIT in particular creates a feedback loop where builders with established private relationships attract more JIT flow, which helps them win more blocks, which makes them more attractive to JIT searchers.

Two economic regimes, measurably different

The 75% vs. 32% pass-through gap for price impact isn’t just an academic curiosity - it’s a concrete measurement of what exclusive order flow costs the system. On public routes, competitive dynamics push value toward builders and ultimately proposers. On private routes, searchers retain more, and the competitive pressure that would otherwise distribute surplus more broadly is attenuated.

This provides empirical grounding for what the blockspace market paper describes as “fragmented access to transactions across builders” - the JIT data shows exactly how this fragmentation manifests in fee dynamics.

Passive LPs absorb the cost

JIT providers minimize their adverse selection risk by precisely timing entry and exit. Passive LPs, who sustain baseline market depth, receive diminished compensation because JIT providers capture a disproportionate share of fees from the highest-value trades. From a microeconomic perspective, JIT operates as an optimized, event-driven form of market making that maximizes fee capture per unit of capital - at the expense of passive participants.

For Block Builders

JIT is position-sensitive and can’t be commoditized through merging. Unlike non-contentious transactions that can be appended via relay block merging, JIT bundles must surround a specific swap in a specific order. They’re inherently contentious - they can’t be tacked onto the bottom of a winning block. JIT revenue remains exclusively a “winning builder” advantage.

The public/private split creates a two-tier market. Builders who receive private JIT flow get a competitive edge in the relay auction (the bundles add block value), but the searcher pays less for private execution. Builders who rely only on public JIT face more competition but capture a larger share of each bundle’s value. The strategic question is which mix optimizes total revenue.

Searcher relationships are measurably valuable. The $47 integrated-pair premium and the concentration of JIT around two searcher-builder corridors (jared-Beaver and SCP-Beaver) suggest that attracting and retaining JIT searchers is a concrete builder differentiation strategy - not just a theoretical one.

JIT dynamics may shift under future PBS changes. If encrypted mempools reduce mempool visibility, public JIT (which depends on seeing pending swaps) could decline, pushing more activity toward private channels. If FOCIL forces certain transactions into blocks regardless of builder preference, the ordering flexibility that JIT requires could become constrained. Both directions favor established builder-searcher relationships over open competition.

Sources

Original paper:

Data:

Related reading: