Cryptoshi | Bulk

Continuous price discovery is the product! Markets are not break loops.. they are feedback loops; depending upon your signal - you change your quote, you hedge, you resize. The moment that loop pauses past the horizon you care about, you are not trading with confidence, you're trading with stale information. Information asymmetry is the core problem in distributed markets. A participant far from where state changes (the event) pays higher coordination costs -mostly latency.. if you want strong consistency across a wide footprint, you pay in synchrony (waiting for others). There is nothing wrong with that tax for a general-purpose chain - but trading has a different variable: how fast your inclusion will turn into an execution outcome that you can utilise A trader has to bear the tail risk after inclusion -> the order can be filled, rejected, limited by the engine, because everything downstream (hedging, inventory, the next quote) keys off that Trad fi venues already solved "local truth" for price discovery - exchanges like CME, NYSE, TSE, SGX, HKEX lead discovery in their products because participants co-locate or cluster near the matching engine. Proximity buys lower coordination cost, predictable fills, and faster feedback between signal and origin This is not perfect but market structure follows physics here. last few years, many solana perp designs worked on global inclusion as the main win, but continuous execution and continuous discovery are not the same thing as "included in a block".. Even if inclusion is fast but execution feedback is late, traders still operate blind for the interval that matters for hedging and inventory.. Gap is execution risk here! For example - post order -> execution result (400ms +) They do win the composability narrative, without actually winning the part that competes with centralised / regional exchanges: fast inclusion to execution feedback.. Different designs I've been reading about: 1. Localised ingestion with MCP - optimise for global inclusion: many paths into the pipeline, censorship resistance, and a canonical global view after batches settle. The tradeoff is that execution is not continuous in the sense market makers care about - work is still batched, ordered inside the batch, and final economic state (filled vs reject, at what price) arrives on a slower clock than inclusion.. You must quote and hedge the next interval without full feedback from the previous one. On a price-time book, priority is only meaningful when execution is continuous enough that "priority now" maps to risk now 2. Colocated / regional matching - feedback is fast enough that makers can adjust, hedge, and model risk on a shorter clock That does not make regional venues "better" in every class: liquidity splits (arbs are needed), routing complexity, and non-atomic portfolios across regions are real costs with cross margining. A good summary: colocation offer stronger local coordination for continuous discovery loop; MCP is a different offering better for unified global state, bad at continuous market micro-structure loop. We love Solana for inclusion, asset breadth, and liquidity but markets still need a structure built for trading, not only for global settlements.. BULK is our bet that regional coordination (low latency, continuous deterministic execution feedback) is the right layer for complex risk engine products (derivatives), and with settlement anchored where it belongs. What BULK does today (single-region deployment): A distributed validator set within one region (EU), with 20ms end-to-end for the trading path, without generic smart contract overhead or gas complexity for the hot path. A purpose-built execution system that provides fastest information propagation to participants Consensus path: ingest -> broadcast to all validators -> minisketch based reconciliation of pending sets -> vote deterministic execution. No leader compiling the batch; the quorum intersection of pending sets is the agreement object, with compact sketches instead of resending full transaction lists everywhere.. Offers the best local optimum for continuous discovery in a decentralised network - while offering better censorship guarantees than MCP and keeping the loop of inclusion to execution very tight! Our testnet has been an eye opener for us - 15ms avg consensus / inclusion latency + 5ms execution stage latency, all while 160k active accounts (accounts with positions), scaling and optimising this infra makes us fall in love deeper with how we can solve cross-region next Today we have reliable inclusion, fast execution feedback, and less "batch tail risk" than designs where you are included (MCP) but economically unresolved across long windows But that's not enough - to accelerate end state of on chain markets we must look further! Cross-region (without false guarantees) Here's our plan: > Localised ingestion where users enter through the nearest / best node. > Regional books with continuous discovery inside the region on the same kind of pipeline as above.. Cross region behaviour becomes complex but not impossible nor will it lose any continuous price advantage / best bid. Best execution becomes a routing problem; portfolio views across regions need fault tolerance Best exchanges in the world do this for fx markets today - where trillions of dollars are traded. We are working towards a world where it doesn't matter where you are when you place an order -> you get the best price for it. In the backend the order depending on continuous discovery will route to regions that have the best liquidity for it Crypto in Japan/Asia, Equities in USA, Fx in Europe -> all while tight looped execution is preserved! We are not selling the Continuous Discovery dream as a hoax - real lives depend on market micro-structure, it needs to be fast feedback between belief, order, fill, and the next quote. If you are dependent on architecture making inclusion easy and resolution slow, you have not fixed markets - you have repackaged uncertainty as a feature! BULK is our chance to solve this, and we will give everything we have to build the layer where the end state of markets will live. Tick by tick 🫡