Jun | bulk trade

Crypto made asset issuance permissionless It never made derivatives creation permissionless A futures market is not just a ticker and a pool. It is an agreement between contract design, reference pricing, liquidity, matching, margin, clearing, collateral, liquidation and distribution In the traditional derivatives stack, those functions are split across institutions: The exchange defines the contract and operates the matching engine The clearing house becomes the central counterparty, sets margining rules, nets obligations, manages settlement and maintains the default waterfall. FCMs, brokers and prime brokers provide market access, financing, collateral management and house margin It is expensive and slow to coordinate, but robust! CEXs compressed that stack, but kept the most valuable gate for themselves. Everyone has heard what a listing actually costs 😄 Ten years of CEX innovation also failed to bring TradFi-level capital efficiency to defi. Most perp venues still margin positions individually or use static margin tables that barely respond to changing market conditions Portfolio margin is the most powerful tool in derivatives and the most gatekept. It is why a fund runs a hedged book for a fraction of the capital a retail account posts on the same two legs. In TradFi you need an ISDA, a prime broker and a six figure account to access it. We built it into the base layer of the exchange and opened it to everyone.. Margin should scale with the net risk of the whole book, not the gross size of each position alone. Required margin scales with portfolio variance -when one position offsets another, the required collateral should reduce. A real hedge should release capital instead of being margined twice CME is the Industry standard benchmark - SPAN evaluates a discrete risk array and grants cross-position relief through exchange-defined spread credits! Every crypto offset CME offers, is a contract against itself in another wrapper. Bitcoin against Micro Bitcoin. Ether against Micro Ether. Futures against options on the same underlying. Long Bitcoin, short Ether is two different assets, and you need to use a third product, the Ether/Bitcoin Ratio future to access the benefit. Here's how BULK risk engine (inspired by SPAN) performs better: Netting: every position is evaluated against every other position through a live correlation matrix. The measured relationship is the credit, there is no eligible-spread list Regime: maintenance margin changes with the live volatility and trend state across nine market regimes. A position should not require the same margin in a calm trend and a correlated crash Resolution. SPAN is 16 scenarios. BULK runs a 500,000 path Monte Carlo at every point of a 3,024 point grid of threshold by leverage by regime by side. Roughly 1.5 billion simulated paths per asset (and more), each path 25k+ steps, with regime transition dynamics. It prices the real tail and the path into it.. the resulting risk surfaces and correlations are protocol state. Every validator applies the same margin, liquidation and settlement logic. There is no private risk desk granting one account a better margin agreement Same principle CME pioneered, margin the net risk of the book. A generation newer in execution, and transparently decentralised! BIP-1 brings that engine to market creation BULK compresses functions divided between the exchange, clearing house, broker and prime broker into one deterministic base layer. For deployers, the opportunity is larger than launching another long-tail crypto perp. A deployer can build markets around commodities, rates, indices, sector baskets, regional assets or any other product with a defensible reference price and real liquidity. You can prove demand for your idea without persuading any committee! And once that market matures, its traders enjoy the fruits of portfolio margining. The market can become part of a wider capital network, where positions hedge against existing markets and collateral is allocated to the net risk of the whole portfolio Innovate. Deploy. Prove demand. Calibrate the risk. Unlock Portfolio margining. That is BIP-1

Gud tek 🫡

We are so early.. back to work, I keep reminding myself perps are 0.1% of traditional derivatives We have to build systems that trad finance offers, without the clunkiness of crypto to accelerate tam 🫡

Elliptical cryptography is built on math assumptions and tricks, if indeed open ai or anthropic has found a (log p)^k solution to discrete log then it’s also plausible to assume that other open source labs or frontier research groups will find it too High performance cryptography needs a better way to do hash based crypto, they are computationally demands and mssg size will balloon

BULK is one of the fastest growing derivatives exchanges A 8% compounded daily growth, on avg $350k in net OI added per day since launch Global derivatives represent roughly $1 quadrillion in outstanding notional. At $100B+ in open interest, perps represent only around 0.01% of that scale In the last few months Perps have already demonstrated substantial demand and price discovery during non trading hours. As more of traditional finance catches on to perpetuals, I believe capturing more than 10% of the broader derivatives market over the next few years is possible But to get there, we (onchain) need the risk infrastructure that tradfi has spent decades developing: that allows institutions to manage complex portfolios. SPAN is part of that history. Used since 1988 across exchanges and clearing orgs worldwide, it puts portfolio risk and eligible hedge offsets at the centre of margining Cme's portfolio-margining programme alone averaged $8.4B in daily margin savings during 2025, with savings exceeding $10B per day at the beginning of 2026 That is billions of dollars in lower collateral requirements through better portfolio risk management In the next few years, there wont be a distinction between trad and defi. But if we want to make that happen, we have to do better We, as a industry should offer the beautiful features of blockchain to risk infrastructure built through decades That is the direction we believe onchain derivatives should take to grow: recognise useful hedges, charge for the risk that remains, and account for what it would cost to close positions under stress. Ultimately, that means helping traders use their capital more efficiently BULK brings that portfolio-risk philosophy onchain through its own margin model. A long BTC / short SOL portfolio can receive margin credit because the positions partially offset. The credit depends on sizing, correlations and risk conditions, while leverage constraints and closeout reserves remain in place We are so early, To bring the world of derivatives onchain, we need to learn, adapt and build a better system For us at BULK, the job is clear: build and share the best risk systems onchain, and shove it down tradfi's throat Over time, we want to bring that same margining philosophy to a broader set of instruments: swaps, forwards, options and other OTC derivatives, alongside interest-rate and FX futures and options Accelerating, tick by tick.

Tokenisation and trading are the strongest PMF we have for global state machines like Solana and Ethereum. Global adoption of derivatives built natively on blockchains is still in its early days We are building BULK towards that future: the best margining + execution stack onchain. The benchmark is not a crypto exchange - it should be systems like CME SPAN. Tradfi derivatives markets already use sophisticated portfolio margining, but blockchains give us the opportunity to go further by unifying collateral, risk and execution in a programmable, globally accessible state machine. Derivatives trade trillions of dollars every day, yet access to the underlying market remains fragmented. Blockchains can change that in the same way they opened up global access to spot assets. Opportunity to deploy trillion dollar derivatives for RWA / Equities, rests entirely with BIP1 BIP-1 is about enabling people to craft entirely new derivatives - markets with pricing, settlement and risk structures that would be impossible to build on traditional rails Excited to be working with BIP-1 deployers already pioneering new ways to price markets across every kind of asset. The great “let’s make a derivative of everything” era is upon us!

🧱🧱

Propamms could replace the expensive quoting part of a price time ladder that an orderbook holds. You can reprice bunch of levels with a very small txn and that’s best way to reprice on Solana currently.. The complexity is a layer deeper - matching and risk checks, A matcher or core engine of this perps exchange has to touch multiple propamms to find the best price to the users order but also has to do bunch of risk checks that are expensive ie calculate margin, leverage, portfolio impact, etc Exchange lifecycle is also cost that we always forget when we talk about these prop amm vs clob ideas - the expense still remains where you have to read and write literally thousands of users to keep them updated at all times (even moreso during volatility or when accounts are closer to liquidations) All these costs can be shifted but the costs remain nonetheless Clobs are beautiful as they hide this complexity- matching is usually O1, Risk and lifecycle can be 01 too Majority costs of clob are in downstream that needs to be compressed and represented on a Solana block

beginning - but lot of work left to do 🧱

BTC Market Update.. Between 14:15 and 15:15 UTC, low BTC liquidity caused a price dislocation in Bulk’s orderbook while the external oracle remained near $79,600 Several highly leveraged accounts traded aggressively through the available depth. One account, 8dDu…q1Wh, repeatedly rebuilt a large BTC short while undergoing partial liquidations. It generated approximately $1.5 million of sell volume across 400 fills and was partially liquidated eight times. This concentrated activity materially reduced available book liquidity. Approximately 6.50 BTC of the account’s flow was absorbed by the protocol liquidation strategy too At 14:36:45 UTC, a protective short-close order activated near a mark price of $81,450. It became a reduce-only market buy, consumed the limited ask liquidity, and reached a resting ask at $115,115. Three additional protective short-close orders activated shortly afterward, near $82,000, $82,239, and $82,500. The resulting dislocation triggered a wider liquidation and ADL cascade. Within the primary event: > Multiple highly leveraged short accounts were force-closed > Open interest fell from approximately 26.21 BTC to 6.82 BTC > The protocol liquidator absorbed substantial inventory and incurred a material loss while containing the event. > ADL covered approximately $48,480 of account shortfall >Protocol liquidator/risk-vault capital covered approximately $66,149 that ADL could not recover Across the full incident, the protocol processed approximately 26.55 BTC of forced liquidations and approximately 17.69 BTC of matched ADL reductions. The liquidation, ADL, and risk-vault mechanisms contained the account deficits and prevented them from developing into exchange-wide insolvency This incident was driven by low orderbook depth, concentrated highly leveraged activity, and multiple execution and risk mechanisms interacting during a low-liquidity bootstrap period. We are reviewing every affected liquidation and ADL account and will reimburse eligible users for losses due to the dislocation 🙏 The next network upgrade introduces three immediate protections: 1. A mark price that is harder to move We are removing mark price’s sensitivity to a shallow or heavily skewed local book. During the liquidity-bootstrap phase: > Mark will remain more strongly anchored to robust external market data. > Small amounts of distant liquidity will not materially move mark. > Local book inputs will require sufficient depth before influencing mark. > Mark-versus-oracle divergence will be explicitly bounded 2. User equity will no longer depend on instantaneous local liquidity User equity currently is evaluated using an estimated price-to-close. This is conservative in a deep market, but in a thin market a missing price level can make the estimated closeout cost deteriorate suddenly! During the bootstrap period: > Equity and liquidation eligibility will use market price instead of instantaneous local price-to-close. > Sparse depth will no longer immediately reduce account equity. > A temporary liquidity gap will not by itself trigger a broad liquidation cascade. > Liquidation eligibility will remain aligned with the market reference used by users and market makers. Liquidity-aware equity will return when the books have normal, reliable depth, with explicit impact caps, minimum-depth requirements, and stronger oracle anchoring These protections are designed for the current low-liquidity phase and will remain active while deeper, more resilient market liquidity is established Bulk has zero tolerance for market manipulation. Any deliberate attempt to exploit low-liquidity conditions, manipulate the order book, or externalise losses onto users or the protocol will be thoroughly investigated and will not be tolerated User safety, platform security, and the protection of customer trades come before every other priority. Eligible users affected by this dislocation will be reimbursed following the incident review today xitter.twitterapi.xyz/Rutto02/status/2096647…

tick by tick now!

On a proof-of-stake network, rewards are a bundle of incentives for three things: capital, consensus participation and block production Ethereum issues ETH to validators for consensus work for attestations, block proposals and sync-committee duties. The proposer also earns from priority tips and mev flow. On Solana, epoch staking rewards depend on delegated stake and vote credits. The block-producing leader receives half of the base transaction fee, all of the priority fee and may also earn mev These systems are answering three basic questions: i) Decentralisation through stake ii) Who participated correctly in consensus? iii) Who produced the block? All of these are important. But for a trading network, there is another economically useful role that sits outside the protocol’s accounting: Who actually brought the fee-paying order into the network? The validator reward design we are preparing for BULK mainnet adds this as a separate contribution. 50% of validator rewards are assigned to txn orgination - it rewards the validators who introduce economically productive order flow to the rest of the network Economically productive does not mean transaction count or packets sent. It focuses on where and when a transaction actually executed and paid a positive fee An invalid transaction contributes zero, as well as orders that enter the book but never fills. A fill where the fee is fully consumed by a funded maker rebate contributes zero. The accounting unit is the actual integer fee atoms posted by execution. It is not an estimate of the transaction’s value reconstructed later from notional, order size or message volume. This choice controls network spam! Simply put, origination credit is assigned after deterministic execution and fee settlement, meaningless activity has no reward surface. A message only becomes relevant to origination once it produces a real fill that actually pays a fee in the network. Spam can still consume bandwidth, so the network has rate limits, canonical transaction identities, deduplication, self-trade prevention and normal account and risk checks. The point is that the reward function itself does not turn spam into a positive-return strategy. Even manufactured fee-paying fills are not a free loop. Attribution is also not based on a validator reporting its originated transaction. There is no single globally true first-arrival timestamp on a wide-area network. A validator in Singapore may first receive a transaction from one peer while a validator in Frankfurt receives the same transaction from another Each validator therefore records which peer first delivered the canonical transaction to it At the end of the epoch, validators commit signed attribution reports through consensus. Once those reports are committed, every executor derives the same origination payout from the same inputs. Self-attribution is excluded. A validator’s origination weight is calculated from what other validators report first receiving from it Duplicate delivery cannot create duplicate economic value because the canonical transaction can only produce its committed execution result once. This gives us deterministic settlement over non-deterministic network observations. Consensus makes every validator calculate the same reward function. Peer witnessing prevents a node from directly awarding itself It does not eliminate collusion, which remains part of the bft security model, but it limits the influence of any one report and makes the resulting payout auditable The economical effect is huge.. A validator earns more not by producing spam, but by operating better ingress: low-latency regional gateways, reliable api infrastructure, direct application integrations, strong peering and consistent delivery of orders that actually execute and pay fees. Transaction distribution therefore becomes a protocol-accounted contribution rather than only a private agreement or an off-protocol payment-for-order-flow deal The network should gradually direct economics toward nodes that make the system work better. But the signal cannot be a node self-reporting that it is fast, important or well connected. Byzantine nodes can lie about their own performance, and raw activity is trivial to manufacture. The signal has to be peer-observed, backed by actual execution and settled through consensus. Origination tells us who brought the network something worth executing Read more: docs.bulk.trade/bulk-exchang…

BULK testnet network did halt during this - we run a small 20 nodes cluster in tokyo but with different localities. Once things were back to normal - network performed a self heal and continued with block production! With mainnet on the horizon we are decentralising the network with different operators and localities to prevent a disruption. Very cool to see that we are one step closer to mainnet with no issues on network healing Gbulk

Order execution priority is expected to change on @bulktrade At present, orders queued for execution after consensus agreement between nodes go through a deterministic shuffle. This prevents anyone from inserting a front-running transaction after seeing the block contents When the block executes, actions are processed based on protocol priority: 1) Cancels execute first 2) Post-only (ALO) maker orders execute second 3) Regular orders, including market, GTC and IOC, execute third This structure provides fairness guarantees independent of the shuffled ordering, If a cancel and a matching order are in the same execution block, the cancel executes first regardless of its shuffle position. Makers seed the book before takers can cross it. Post-only orders establish liquidity before aggressive orders become executable There is also no front-running within a block. Once consensus fixes the transaction set, nobody can insert a transaction after observing the block contents! As we head towards mainnet (big news next week), we have been having meaningful conversations with the community and market makers. We have come to a few conclusions: 1. BULK network latency between nodes is currently around 20ms at p99 2. Block creation and order collection are incredibly fast - 8ms avg! 3. Execution latency for a dense batch remains below 5ms at p99, keeping the full order path within roughly 20-25ms While this makes BULK one of the most performant decentralised trading networks in the industry, raw speed alone is not enough.. We also need to provide makers with protection against adverse flow, similar to what they have become comfortable with on other venues. Different forms of taker protection already exist across crypto exchanges, including Hyperliquid, Backpack, Lighter and others. These protections make it easier for market makers to quote tighter because they are less likely to have stale liquidity swept when the market moves against them We believe permanent speed bumps prevent true price discovery, which should be a sign of mature markets! But we also understand the current market microstructure. Today, makers often feel safer quoting tightly on an exchange with a bump than quoting wider on an exchange without one. Current crypto industry standards have become important enough that we cannot ignore them.. To create the best market structure, where every participant can thrive, we have to meet the market where it is today and then push hard to innovate beyond it As BULK begins as a new network, we want to provide the same safety to market makers until liquidity and markets mature. This should also be a net positive for our retail users too, who want tighter spreads, deeper books and better fills So today, we are introducing a 25ms taker speed bump: Transactions containing only provably maker-only (ALO) orders can proceed immediately. Taker-capable transactions wait in the delay pool for 25ms before becoming eligible for deterministic ordering and execution. This architecture is strictly temporary. It is intended to protect and encourage market makers during BULK’s first 90-120 days while they become comfortable quoting consistently on the network During that period, we will introduce a novel protective order type that provides better and more targeted protection without delaying all taker flow. The direction is an order that can cancel or constrain its resting exposure when deterministic short-horizon flow signals indicate elevated adverse-selection risk. Similar ideas exist in equities through mechanisms such as IEX D-Limit and Cboe Quote Depletion Protection, You can learn more about how transactions are classified and held before deterministic ordering here: docs.bulk.trade/architecture… If you are a market maker or a taker fund, please reach out. We want know your thoughts! We will push hard to innovate market structure while offering makers the protection they need, so the overall net gain to BULK remains green Tick by tick!

Build with bulk - now live!

Whey market on @bulktrade anyone?

Last 2 weeks have been huge for us, stepping closer to mainnet - tiered consensus: backup and active set for more robust network participants - snapshots and chunking: bringing fast catchup to the network and state saving - Matching engine performance improvements - liquidation vault as a function of the exchange - frost based withdrawal and deposits for the mainnet network Next few things we are working on - new order types; scale, twap etc - audit for the program in process - validator rewards - vaults - BIP 1 A product maturation cycle is always a slow process and audits slow a product down taking up huge timelines with little to no movement As we mature the security matures with it - we have seen this with Hyperliquid too from 3/5 multisig bridge to network native circle mint Bulk today is also secured by same setup - any new upgrades to the program go through a time lock and can be tracked publicly All user funds have their own individual accounting pda, trusting the security Solana offers and secures billions of dollars today. All fud is fuel - we will do better to educate users in the mean time Tick by tick!

Higher

Aura maxxing in these market conditions