Published on Sat Nov 01 2025 00:00:00 GMT+0000 (Coordinated Universal Time) by Orkid Labs

The MEV arms race has real costs. Flashbots reports that sandwich attacks extract 500 million to 1 billion dollars annually from retail users who don’t know they’re being attacked. The average MetaMask user pays 87 basis points per swap—8.70 dollars per thousand-dollar transaction—to invisible bots that buy before them and sell after them. We observe that some protocols subsidize execution to compete with centralized exchanges, which compresses margins. Trading desks report getting frontrun on their own arbitrage opportunities, burning gas on reverts and seeing inconsistent profit-and-loss per unit of risk.

We believe there’s a better approach: treating MEV as a coordination problem instead of an extraction problem. Extraction maximizes profit per transaction, which can create negative externalities. Coordination means routing where the path will actually clear, at the size you can safely move, without inviting toxic flow. That requires signals that anticipate safe paths and sizes before you commit capital, private lanes that eliminate pointless exposure to sandwich bots and frontrunners, and execution policy that turns intent into bounded risk. Not just quotes. Not just routing. Orchestration.

The difference between orchestration and extraction is the difference between variance and realizability. Extraction-based MEV strategies maximize variance. Sandwiching means buying before the user and selling after the user to pocket the spread, which costs the user 50-150 basis points in slippage. Frontrunning means seeing the user’s arb and executing it first, which can result in reverts and wasted gas. Backrunning means extracting value from the user’s state change, which can give the user a worse price than quoted. The result is high variance and unpredictable outcomes for users.

Orchestration-based MEV strategies maximize realizability, not variance. Predictive routing means routing where the path will actually clear at the size you can safely move, not where the quote looks good at time T but fails at time T plus thirty seconds when you execute. Private execution means eliminating sandwich risk, frontrunning, and backrunning by never broadcasting to the public mempool where every transaction is visible to bots and validators. Policy guardrails mean profit floors that reject executions below favorable conditions, slippage caps that prevent catastrophic losses from thin pools or stale state, circuit breakers that disable execution on degraded signal freshness or RPC health, and freshness gates that reject executions if state is stale beyond N blocks or simulation age exceeds T milliseconds. The result is low variance and more predictable outcomes for users.

For protocol teams: We observe that users experience high slippage costs and unpredictable execution. Integrating ORKID for private, MEV-protected routing can help you offer better execution to your users. Lower realized slippage means users have better outcomes, which can improve retention and user experience. The math is straightforward: if you’re routing ten million dollars per day and you reduce slippage from 87 basis points to ten basis points, you’re creating 77,000 dollars per day in value for your users, 28 million dollars per year. That’s a meaningful competitive advantage.

For wallets and routers: Reactive routing—best quote at time T—can result in reverts or worse execution by time T plus thirty seconds when state has changed. Predictive routing that optimizes for best fillability at time T plus delta can help reduce reverts and improve user outcomes. Fewer reverts means lower gas costs for users. Lower slippage means better prices. Better execution means happier users.

For trading desks and funds: Right-too-early reverts waste gas and create inconsistent profit-and-loss outcomes. Policy guardrails that only execute when conditions are favorable—profit floors at 0.7 times the 48-hour median profit per chain, slippage caps at the 80th percentile by pair and venue, freshness gates that reject stale state—can help improve consistency. More consistent profit-and-loss because you’re only executing when you have edge. Fewer wasted executions because you’re not burning gas on marginal opportunities. Better risk-adjusted returns because you’re maximizing realizability.

For users: Current execution costs on MetaMask average 87 basis points per swap. Using wallets and protocols that integrate ORKID can offer 2-13 basis points execution cost instead, with private routing and no sandwich risk. The difference is 0.20 to 1.30 dollars per thousand-dollar swap instead of 8.70 dollars. Scale that to your annual trading volume and the savings are material.

The product has four pillars that work together to maximize realizability instead of variance. The first pillar is signals via proprietary detection that quantifies fill probability, safe size bands, and route stability before you commit capital. The reactive approach is scraping quotes from DEXs, picking the best quote at time T, and executing while hoping it clears. That’s how you get reverted or sandwiched. The predictive approach is modeling pool state, depth distribution across price ranges, and microstructure like fee tiers and tick spacing to quantify fill probability at time T plus delta and routing where the path will actually clear. The inputs are on-chain pool data including reserves and fees, depth and shape from liquidity distribution across price ranges showing where liquidity is concentrated and where it’s thin, venue microstructure like fee tiers and tick spacing that affect slippage, and recent fills including execution data, revert rates, and realized slippage that tell you which routes are actually working in production. The outputs are probability-of-fill bias showing which routes are safer based on historical data, safe size bands showing how much you can move without excessive slippage, and route stability metrics showing how likely the path remains valid over the next few blocks. The objective is maximizing Rate of Realizability, which measures the probability that a quoted price will actually be realized after accounting for gas, slippage, and execution risk. A quote that looks five basis points better but has 60 percent realizability loses to a quote that’s three basis points worse but has 95 percent realizability.

The second pillar is predictive routing, which does multi-hop, cross-venue planning with policy constraints and real execution costs baked in from the start. The static approach is best quote on a single venue via direct route while ignoring gas costs and slippage risk. The dynamic approach is multi-hop across multiple venues with cross-venue optimization and real execution costs in basis points including gas and fees. Consider a direct USDC to WETH swap on Uniswap V3 that quotes five basis points better than a USDC to DAI to WETH route via Balancer plus Curve. The direct route has a better quote but the pool is thin, slippage risk is high, and Rate of Realizability is low because the pool might not have enough liquidity at your size. The multi-hop route has a worse quote but the pools are deep, slippage risk is low, and Rate of Realizability is high because both pools have ample liquidity. The winner is the multi-hop route because it delivers a slightly worse quote but a better realized price after accounting for slippage and execution risk. This is the difference between optimizing for quotes and optimizing for fillability. One gets you reverted. The other gets you filled.

The third pillar is private execution via Tycho private lanes exclusively, with no public mempool exposure and simulate-then-execute using REVM and tycho-simulation on every route. The public approach is broadcasting to the public mempool and hoping you don’t get sandwiched while praying your transaction clears. The private approach is routing via Tycho private lanes with no public broadcast and simulating before executing. The public mempool is an MEV honeypot where every transaction you broadcast is visible to searchers who are bots scanning for arbitrage opportunities, validators who can reorder transactions for profit, and block builders who can insert their own transactions before and after yours. They can sandwich your swap by buying before you and selling after you to pocket the spread. They can frontrun your arb by seeing your opportunity and executing it before you. They can backrun your trade by extracting value from the state change you created. Private lanes eliminate this exposure entirely because your transaction goes directly to the block builder via Tycho’s private routing infrastructure. No public broadcast, no MEV extraction, just execution. Before we send anything, we simulate the route to enforce minOut and staleness constraints. If the simulation fails or the state is stale, we reject the execution and re-simulate. No hope-and-pray execution.

The fourth pillar is guardrails via policy that implements profit floors, slippage caps, freshness gates, circuit breakers, and rate limits. The unbounded approach is executing every opportunity with no profit floor, no slippage cap, and no staleness check. That’s gambling. The bounded approach is only executing when conditions are favorable with dynamic profit floors, slippage caps calibrated to market conditions, and freshness gates that reject stale state. Policies are configurable per chain and per venue, allowing you to tune execution parameters to your risk tolerance and market conditions. The impact is turning intent into bounded risk by only executing when you have edge and avoiding right-too-early executions that burn gas without profit.

The technical moat comes from four sources that are hard to replicate even with capital and engineering talent. First, proprietary signals that use advanced modeling of pool state, depth, microstructure, and recent fills instead of reactive spread-scraping or static quotes. This requires deep understanding of AMM math across multiple venue types and execution dynamics. Second, simulation gates where every execution is validated before sending with constraints enforced. This requires integration with simulation infrastructure and careful validation logic. Third, private execution by default with no public mempool exposure, aligned with non-toxic MEV design goals. This requires partnership with private routing infrastructure providers who are selective about who they work with. Fourth, a production-grade implementation with inter-service communication, real-time data feeds, and structured observability. This requires careful engineering and operational discipline.

The traction proves the approach works and the proof is production data, not backtests or simulations. We have production infrastructure executing real swaps with real capital, with zero MEV exposure from private lanes only. The infrastructure is production-ready with execution guardrails enforced to prevent runaway execution, profit floors validated to prevent right-too-early executions, slippage caps enforced to prevent catastrophic losses, freshness gates validated to prevent executing on stale state, and circuit breakers tested to prevent executing on degraded RPC health. Every execution is simulated before sending with constraints validated and minOut enforced to prevent slippage beyond acceptable bounds. No hope-and-pray execution. Simulate then execute.

The business model captures value through SDK licensing where DEXs, wallets, and trading desks pay for routing SDK at starter tier for 99 dollars per month up to 10k monthly volume, pro tier for 499 dollars per month up to 100k monthly volume, and enterprise tier with custom pricing for unlimited volume with SLA and white-label. API subscriptions provide real-time feeds for signals, routes, and execution data at basic tier for 99 dollars per month with one chain and 100 requests per minute, pro tier for 299 dollars per month with three chains and 1000 requests per minute, and enterprise tier with custom pricing for all chains and unlimited requests per minute. White-label routing offers custom integrations for enterprise clients with SLA and dedicated support at setup cost of 10k to 50k dollars one-time and monthly cost of 5k to 20k dollars for SLA, support, and custom features. The unit economics are compelling: 100k monthly routed volume with eight basis points improvement versus status quo creates 80 dollars per month value per user, capture rate of 20-40 percent via licensing and support delivers 16 to 32 dollars per month per user, scale to 1000 users generates 16k to 32k monthly recurring revenue, and expansion via enterprise features, white-label routing, and custom integrations scales revenue without scaling headcount proportionally.

We believe there are meaningful differences between extraction and orchestration approaches. Extraction maximizes profit per trade while orchestration maximizes realizability. Extraction can create high variance and unpredictable outcomes while orchestration creates low variance and stable outcomes. Extraction can have negative user impact while orchestration can have positive user impact. We believe orchestration creates more sustainable value because it aligns incentives and focuses on coordination rather than competition.

The call to action for teams is becoming a design partner with done-for-you integration at zero dollars for limited slots because we want proof points and case studies, not revenue from early adopters. You get white-glove integration with zero engineering lift on your side, shadow mode testing to audit performance versus your current router with full transparency into what’s happening, production deployment when KPIs hit targets that we define together, and a case study to prove better execution with data that you can use for marketing and investor updates. DEXs with more than ten million dollars daily volume, wallets with more than 10k active users, and protocols with execution quality problems should apply. For desks, pilot the SDK with SDK access plus white-glove support. You get Rust SDK with gRPC and WebSocket feeds, policy configuration for profit floors and slippage caps and venue allowlists tailored to your risk tolerance, shadow mode testing with no real executions so you can audit performance without risk, and production deployment when you’re ready. Trading desks with more than one million dollars monthly volume, funds with execution quality problems, and market makers with MEV exposure should apply. For investors, request the data room for a small round to expand coverage and signal depth. You get data room with performance data and case studies and financials, live demo to see ORKID in action with real swaps on real chains, pitch deck with market opportunity and competitive analysis and roadmap, and due diligence with technical deep dive and team background and traction. Seed and Series A investors in DeFi infrastructure, strategic investors like DEXs and wallets and protocols, and angels with domain expertise in MEV and DeFi should apply.

Orchestration beats extraction. We’re live, we’re proven, and we’re scaling. Contact jacob@orkidlabs.com for design partner slots, SDK pilots, or investor data room access.

Written by Orkid Labs