We @bulktrade fundamentally believe that in a financial system like derivatives trading, it is critical that transaction ordering and execution are independent -that multiple operators arrive at the same conclusion without a single authority influencing the outcome An exchange that enables the human urge to trade should not have unilateral control over who wins and who loses Crypto is a weird place to run unregulated exchanges that can censor, reorder, or delay our transactions with zero accountability Real ex: you have a position open, liquidation is 10bps away, you try to close - on every CEX and every single sequencer DEX.. there is no guarantee your close is processed without any bias Last 5 years the industry has suspected various chinese cexs of doing exactly this for liquidation fees! Operator controls the order of execution and you have to trust them not to abuse it Scaling an exchange without a single operator is a hard problem. You need to offer real advantages to different participants - maker protections against toxic flow, low-latency ticks for price discovery, fair execution for retail. The industry assumed you needed a centralised sequencer to deliver this! Our thesis and code proves you don't BULK delivers 20ms ticks and price discovery while being operated by a distributed set of Solana validators, A decentralised L0 network running a purpose-built BFT consensus that commits in a single round! To the people saying you need to defy physics to do this, you haven't touched low-latency distributed systems... Solana with MCP and Alpenglow already proves that lower latency and different market structures are possible at the L1 level BULK takes this to the extreme: leaderless consensus, multi-operator, Institutional grade risk engine and still faster than any true dex! We spent a long time building this infrastructure - from consensus, to recovery, to a risk engine that does real time margining with true portfolio margin unlocks that have never existed on-chain before. All this code boils down to one thing: cementing a great product at early.bulk.trade that traders love! Fastest fills and Best risk management with no compromises! An example of how different BULK's risk engine is from what exists today on a onchain dex : Liquidations are triggered by off-chain keeper bots.. These are external agents are incentivised to call a liquidation instruction when your margin ratio drops below threshold, the bots evaluate each position independently, close it via market order, and collect a fee There is no awareness of your portfolio. No consideration of whether that position was hedging another. No optimisation of market impact or how wider book will react! This is destructive. Closing a position that provides correlation benefit to the rest of your portfolio "a hedge" can actually increase your portfolio risk! Triggering further liquidations in a downward spiral, this is exactly what happens during cascading liquidation events, and it's why every major blowup in crypto exchanges follow the same pattern.. On the other hand - BULK's liquidation engine is an optimiser, not a simple closer. When a portfolio breaches its maintenance margin, the system evaluates every position and computes an "impact ratio": how much margin is reduced per $ of liquidation cost. Positions that reduce the most risk for the least market impact are liquidated first. Positions that provide hedging benefit: where closing them would increase portfolio margin are explicitly preserved This is how worlds biggest clearing houses manage defaults (CME, OCC.. etc) It has never existed on-chain before! And this is just one part of a wider, deeply integrated risk system: portfolio margin, regime-adaptive margining, continuous risk surfaces, real-time trade checks - all running deterministically on every validator Building a product that gives every trader access to best tools, the same tools that (in tradfi) need you to have high 6 figure accounts and prime brokerage relationships (ex- CME SPAN requires 50% of last clearing price of an contract as margin) brings me great joy. We are here to win with our traders..and we will do it tick by tick, transparently and by building a great product. Gbulk

🛠️Contributor ⭐️OG 👀Testnet Who will you be? Don't be sidelined - join now

I fell in love with Solana's vision: a single, global state machine capable of hosting the world’s financial markets. Today, we're building our part. Introducing bulk-agave: a new validator client delivering the ultra-low latency and high throughput a global orderbook demands in a credibly neutral environment. Makers get the fastest execution and Takers get the best prices. This is just the start - jobs not done

Two years ago, when we started building, our vision was simple: make markets better. That remains the cornerstone of our ideology today. After months of exploring orderbook architectures, one thing became clear - performance always came with tradeoffs. We have set out to define the best: blending the highest levels of performance with a truly decentralised, multi-operator architecture. Our architecture unlocks the deepest liquidity environment, coupled with the best user experience, all while staying true to first principles. What makes a perps dex special isn’t just the orderbook or the matching engine - it’s the risk engine. We are building state of the art risk systems, capable of handling any collateral type, managing margining safely, and controlling exposure risk at scale. None of this works alone. Going to battle with comrades at your side is the best feeling. Thanks to our team @kdotcrypto @glowburger @tr8dr @levbeta + others and to our amazing supporters @tarunchitra @ani_pai @mdudas @mdcieri37 @CarlKVogel @ReliableNarr @Jonaskristensen @isc_mirana @jmj and many more. Tick by tick, we’ll make BULK the best place to trade.

We’re 7 strong at @bulktrade trying to win the war of the century- building the world’s best decentralised exchange If you’re obsessed with trading, performance, and being the best… join us: > Frontend / Full-stack Craft elite trading UX, working closely with @glowburger and the wider engineering team (React, Node.js) >Mobile Help us bring infinite markets to your pocket (Kotlin MP / Flutter) If you don’t fit in above, but want to help us expand our propaganda don’t feel shy. Dm me or @kdotcrypto with proof of work 💪

By the time you finish reading this - BULK has already processed and finalised 9 ticks (blocks) with avg latency of 18ms between FRA <> AMS on a 10 node cluster. We will push this to 16ms by mainnet and potentially cutting more by optimising minisketch -moving to something different than BCH. This means you get best exp on your trades and makers don't have to sacrifice perf for decentralisation

.@bulktrade has a cool new txn type: Bundle! Think atomic baskets - an arbitrager or hedge algo can submit a set of orders across books that either executes together in the next tick or not at all. Bundles preserve leg order, enable cross-book arbs, OCOs, and complex multi-leg strategies all at tick speed

Just fueling the autism in me

basically you're really should stake with BULK validator, 0% fees (and 0% MEV fees), so you keep all the yield main motivation is to help support the global trading engine + also increase your APY through fee share (juicy USDC) And (yes there's more) earn BULK [redacted] go to your fav wallet -> stake -> type BULK or BULKzVM41WAyQZfL34vxqdsYwEYH9mJAJyzRS4xraf8b

Thinking more about how you can execute a swap on Jup / titan and end up using @bulktrade orderbook for execution You get the best pricing of async 20ms ticks plus MCP so you don’t have to get sandwiched - while the asset transfer will take place on solana block in next 400ms This is a better market structure than PropAmms as you get to quote various price levels and pairs without having to worry about today’s constraints.

There have been many questions around our consensus mechanism. Our design is something we have put tremendous amounts of thought into and one which we believe will be used by many protocols in the future. BULK achieves fast consensus by first agreeing on the set of orders received in each tick using Minisketch-based reconciliation (a compact, sketch algorithm that efficiently synchronizes order differences between peers). Rather than agreeing on the full chain state immediately, BULK relies on its deterministic execution so that every node can later process the same agreed-upon orders and independently arrive at the same resulting state. In later ticks, the system reaches state agreement through a BFT flavoured consensus, ensuring that all nodes converge on the same finalised state. This split-tick model separates fast order agreement from slower state finalisation, which allows us to fast track execution (20ms) while maintaining deterministic secure consistency. These processes were built to support one giant orderbook. While it doesn't operate like a standard appchain or L1; its a state machine that enables any action on BULK to be cryptographically verifiable, replayable and stored in a ledger Why does all this matter for your trading experience? 1. You can choose your commitment level for confirmations - (a) minisketch accepted (8ms), (b) Executed (25ms) and (c) State finalised (40ms); any latency sensitive participant like makers will have the best exp as trades are acknowledged almost instantly. 2. You get to choose the nodes you want to send your orders (MCP) so there is no node censorship or sandwich attacks on your orderflow. 3. BULK is a neutral player in the network - all the validators determine what happens to your orders, matching results and asset changes. 4. You can replay and verify what happened on the chain - best use case is during liquidations; verify you were not cherry picked for sweeps or that your orders were neglected. 5. Resilient performance - As a state machine; fault tolerance comes naturally. Oct 10 events have taught us that a single machine can go down under heavy load. With multiple nodes participating to process the orderbook, your experience never degrades. There are more modules (executor, order broadcast, risk models etc) which will be going live with bulk-agave, and I can cover those in the near future. This post is to hopefully answer a lot of the questions we have been receiving from fellow consensus maxis . Thank you for your attention to this matter!

Bros in my dm who asked for LST - here you go! Liquid and more integrations are coming soon.. It goes without saying BulkSOL, will be a key part in the exchange asset composability and yield unlocks! Time to BULK up SOL

Execution can make or break a exchange, to put it simply: > how fast prices move / gap risk - how long does it take to update quotes on the book? If it’s 400ms, that’s the gap between this venue and a liquidity leader - makers quote wider just to avoid holding quotes longer. BULK takes an async approach with continuous pipelining (20 ms ticks), keeping gap risk as low as possible > true book depth - this matters most to both retail and pros. Orders on the book should exist and be executable. Many venues have phantom quotes that vanish when hit. During liquidation events in the last few days prop AMMs simply pulled liquidity to let users suffer(prices worsened and ux declined); having every action exist and reproduced on multiple operators matters a lot. On Bulk, every order lives onchain across participating validators - creating the best UX with real prices and zero hidden slippage. > how orders clear - what price does an order actually execute at? what’s the median delta of a cross? as we know mechanisms like auctions and RFQs are inefficient. Price-Time Priority unlocks the most robust + efficient market structure. Bulk executes thousands of orders across hundreds of levels in just 3–5 ms, giving users the best possible price that is robust and deterministic. There’s more beyond execution: the margining system An old friend once told me running a risk engine is the toughest job in the world, and I agree. The engine acting as the heart of the exchange is where real differentiation lies. After thousands of hours, we have built a next-gen risk optimisation engine that does: 1. Effective haircuts based on market regimes 2. Intelligent order treatment logic 3. Risk-optimised ADL through pre-emptive haircuts Plus more - Our goal is to be the most robust, highly performant and capital efficient exchange that also protects its users risk without any bias.. Next few weeks we will piece all these fundamental systems together. As always - we will get there, all in, tick by tick. 🫡

If you are a big solana whale or stake holder and want more APY on your precious sol; DM @bulktrade for some special APY 🥩 (note- this is not expected apy - but cold hard usdc sent to sol stake) Gbulk

We love all things Solana but waiting for 20ms blocks and mcp till 2027 is not feasible.. Bulk allows a validator to run Solana, Jito and bulk all together - while bulk component only does orderbook execution. For this specific orderbook we built order propagation, mcp and consensus all while carefully profiling performance. Other key thing to note is you need Solana stake to part take in orderbook consensus- so bulk is backed by real economic value and it pays the operators back in bulk rewards (usdc)

Payment for order flow (pfof) is underrated and underutilised by perp dexs. Huge liquidity unlocks from either side of the flow. TradFi has been doing this for years now - your favourite stocks app, route orderflow to specific participants who will use the underlying exchange (Nasdaq, Sgx). BULK is building that exchange with infinite markets - imagine many apps and frontends curated for specific users, they get the best execution venue and liquidity.

Job isn't done until we can offer the best performance + censorship-resistant orderbook on Solana. We're building this with BULK Alpha invites are going out to the traders and builders who believe in this mission 🫡 BULK is meticulously designed to perform no matter what - lowest latencies; you got it. Highest throughput, you get that too - Censorship resistance at day 0 - thats our goal too. For traders, this means best pricing. For makers, flawless execution. Next up: we're designing a next-gen risk engine to outperform any exchange during peak market volatility. The work continues..

We want every single @bulktrade enjoyer to have exposure to maxx [redacted] even before testnet.. Best place to do that right now is on @ExponentFinance while earning 18% and only 50% capacity left 😱 exponent.finance/liquidity/b…

500 likes on this and @kdotcrypto will pull this off @SolanaConf stage😄

Few thoughts from this convo: I agree with @Tristan0x on the problems perp dexs face - congestion is a serious issue. Markets depend on fast movement, and not being able to close positions or process liquidations is a huge risk. One of the biggest pain points, especially from our convos with MMs and traders, is the infra build-up required just to land transactions - and fee markets are brutal. Realistically, it takes 2 slots to cancel and place an order, and that’s best case. As someone who market-makes, I feel that pain. Huge respect to the Bullet team, I get the trade-offs you’ve made: performance > decentralisation. But I still can’t get behind sequencers. Our industry is still recovering from “not your keys, not your crypto,” and honestly, a single-box sequencer is no different from an offline matching engine posting results on-chain. Ordering and inclusion depend entirely on the sequencer, right? And posting fills on a DA is honestly a later-stage problem. So what happens if a user has an open position but the sequencer is censoring their txns? how do they exit? If their orders never reach the book, who takes on that counterparty risk? (Just sharing my concerns.) If centralisation = performance, why not just use a CEX? At least they’re regulated and offer better guarantees. I have similar fears about hyperliquid - yes, there are 10 validators now, but the core assets are still secured by a 3/4 multisig. As @armaniferrante said, we’re still an industry mostly run on multisigs. I also understand what @toly is pushing for - why should solana divert energy to building a separate DA scaling layer when the economics are unproven and could end up leaking revenue? In the end, CLOBs will either improve or die. Solana will improve too, but it’s a hard fight: general-purpose vs. purpose-built, and enshrined infra like HL. The market structure, ordering, and guarantees are all different. It all comes down to risk profile. But the industry doesn’t seem to care about blind boxes and we’ll find out how that ends. In the short term, let’s keep building solana into what we’ve always dreamed of. At @bulktrade, our vision follows that same line of thinking. How do we make sustainable trade-offs to unlock performance without sacrificing decentralisation? Simple goals: High performance -> high decentralisation. We want to offer the lowest-latency distributed systems allows - while maintaining full cryptographic provability. Every trader experiences sub-5 to 25 ms latency because solana’s majority stake already relays within those bounds. We built a custom propagation layer on top of turbine (inspired by rotor) to disseminate orders cluster-wide. All validators match every 20 ms tick deterministically. Then, BLS quorum signs the net delta after N ticks. Since matching is deterministic, all honest validators arrive at the same output - no need for a leader or block-by-block logic. Best part: Bulk will be built as a solana client, modified and custom-implemented to support this. For validators: it’s extra REV from exchange fees. for traders: native solana custody, best pricing, no significant trade-offs. The end goal is simple: A global, sub-25 ms orderbook -fast enough to quote into, secure enough to trust, decentralised enough to run forever. Let’s keep growing the pie.

The goal was never to replicate Nasdaq onchain, but to fundamentally upgrade its core principles. Blockchains solved the first foundational problem: settlement. We've moved from T+1 clearing houses to multiple-second finality. Bulk takes this to its logical extreme, with settlement occurring every 400 ms. Latency is the next biggest unlock - its usually game of wire length. However you don't get the benefit of multi operator matching; Bulk is engineered to win both. Our entire matching cycle remains under 25ms. Nasdaq processes 3-5 million order messages / sec; no decentralised network has come close. BULK is designed for this scale. Within each tick, a rapid pipeline of pre-flight risk checks feeds into a highly-optimised deterministic sort, allowing us to process 2-3 million orders / sec ,this ensures the market is always in its most current + efficient state for all participants. Combined with few other killer features that improve liquidity and capital efficiency - we are redefining what nasdaq means for financial markets. Time to gBULK!

All you need is a dream now! gBULK

Update:

宗旨利滚利 对应 好运八方来 散了才能聚 你不出手? 我们出gBULK

Jobs not done Lets get back to 🥩🥩

Please gib kobie your sol

Solana validators are bulk validators Same hardware with modified client

Tick by tick

I like to think @megaeth is the proof that single main frame scaling (Jane street style) can be applied to cryptographic state machines. One beefy box that is optimised to sequence perf I’m pretty sure they also have replication and other tolerance built

Centralised sequencer L2s are literally just centralized exchanges with different custody setup. If the operator wants to censor you - it will censor you. Flow and actions of the network should be censorship resistant, preventative rather than verifiable after the fact. It’s a hard battle to fight, but we will keep pushing towards the end state of defi

We just spent last 48hrs upgrading the experience for alpha net! Our goal with BULK is to build one exchange with infinite markets and this is just step 0. When I say infinite markets I mean everything - from penny stocks to ETFs. To onboard these Internet capital markets we need a robust high performance orderbook - that can enable vast varieties of market structures. Most users assumed alpha net was a simulation of orders, but we had to break it to them this was a real orderbook-the performance and the experience just blew their minds. Traders and players are used to slow loaders and bad ui so far - we have spent significant amount of time on this; loader free instant confirmation on your trades plus a very inviting ui. Performance is just gonna improve from here - more market listings $PUMP was our high risk listing with 10x leverage and it proves that risk engine can handle these volatile markets with pure confidence. We will keep building - you keep trading 🫡

One sol + a nokia phone

We don’t use Solana shreds tho I defo took a lot of inspo from how shreds are distributed BULK takes a different approach using set reconciliation between nodes then push to execution (can be called mcp / pbs) Experience will be continuous and feels streaming

BULK is the most egalitarian thing I've ever seen. It's insanely ambitious, and if it works, it can reshape the fabric of society.

Bulk mentioned 👉👈🥹

Buybacks are bullish af Pump on alphanet!

As we don’t execute orders on the program - Solana shreds aren’t needed I am working on a new post about consensus and building separation soon But closest source related to execution is here nitter.app/bulktrade/status/19498…

Sub1ms is the computational latency between place order to some result right? not the wire latency Impressive work 🫡

I love minecraft and mods make it fun - just like how BULK makes your trading better

Banger

Appreciate you like bulk design 🙏 Our thesis is we are aiming for execution that isn’t bound to restrictions and Solana’s REV thesis a major part of that Validators will earn more revenue with this execution and these validators are the same ones who run Solana Any composability Bulk has is through Solana applications and assets. And I am a firm believer that there are many apps and use cases we can’t offer that Solana is the best place to build on

🦀🦀

In the last 24hrs since launch, we have more than 80k staked Time to BULK up your sol 🥩🥩

I don't think composability matters if excution is not worth quoting into. At least for a derivatives exchange composability should be treated as second hand effect. Features like borrowing, looping, or hedging strategies, which rely on composability, are secondary. They are ineffective if users cannot get the best and most reliable trade execution first.

The company behind @bulktrade is literally called North Star Labs! And we are literally building the one exchange to rule them all 😃

BULK team is unbeatable - we have been constantly pulling 20hrs a day to ship alpha net and I have never heard anyone complain; all motivated by one simple goal "Make markets better".

GbulkAgave

A man and his friend walked into a bar. "So," the first man said, swirling his drink, "I just had the most enlightening conversation. Apparently, the future of finance is all about being... transparently controllable." His friend choked on his beer. "Transparently controllable? You mean, like, they can see everything you do, but they can also stop you from doing it?" "Precisely!" the first man beamed. "But don't worry, you can verify that they stopped you! And you don't even have to show them your ID before they decide to pull the plug. It's a truly free market, in the sense that they're free to do whatever they want with your funds, and you're free to verify it happened." The friend stared into the bottom of his glass. "So, it's like a really fancy, complicated version of putting all your money under your mattress, except someone else holds the mattress, and they can decide to light it on fire, but you get a perfectly verifiable receipt of the ashes?" "Exactly!" the first man clapped him on the back. "And the best part? The ability to prevent anyone from messing with your money? Apparently, that's 'the least interesting property to optimise for.'" The friend just slowly slid off his stool, muttering, "I think I need a different bar."

Yes, we will have LST too Now let us cook

you can also stake through validator site directly here: stakewiz.com/validator/BULKE…

The best trading venue starts with best custom built infrastructure

BULK ME UP

Can't believe it has been a year, I started working on Avian Network. Met these amazing people on the way @TihamiTahmid @craig @ArkantiK_ @coinminerz @alamshafil @primitt @wdevico @RoberttBertton and @mathias Happy birthday to all Avians

time to 🥩🥩

Cant wait to long $GLXY @bulktrade

hedge / arb folks dm me to learn more

Running a robust risk engine that can handle the best margining and market regimes is the hardest part of a perp dex

Finally landed in London, but they lost my bag 😭

Tick by tick brother

Bulk isn’t just fast. It’s a new execution primitive. We’re building the first fully verifiable, low-latency, on-chain CLOB that runs inside the validator layer itself. Cant wait to talk more about the design philosophy that we are pioneering!

use this as personal liquidation stress relief

Is there a open source implementation of this? or any perf optimised ed25519 The best option I have found is github.com/firedancer-io/fir… Curious if any closed source agave runners use this c lib.. Interested to know performance gains with overall slot processing times? any bottlenecks too

Tick by tick

@SuperteamUK @SuperteamSG @superteam @solana_devs @SuperteamVN @mtndao Tagging all my other frens 🙏

Soon

Remember when we talked about sophisticated risk engines, best clearing houses, and funding rate mechanisms? Now, it feels like the sole race is to launch the next CLOB offering marginally less latency than the last. Markets aren't built on latency alone. Bringing internal capital markets home needs significant research, dev, and rigorous testing of these underlying systems. Instead, we're often just copying opaque CEX contracts or their liquidation engines, relying on their best for a reason status without understanding their proprietary, resource-intensive development. Every system has to be unique and serve its purpose. Even basic choices like contract pricing drastically alter how funding operates, yet the majority of users remain unaware of how these critical systems are truly designed or to what extent they can be trusted. Ultimately, getting users to adopt your product is paramount, far more so than a cleverly gamed system BULK is solving for the user, not for internal wars.