Last week in Hong Kong, I joined a fireside chat at @Equinix’s Tech Leaders Exchange to discuss the convergence of TradFi and DeFi and how @TheHUB is creating an optimized environment for this convergence to materialize.
A couple of highlights from the talk: a lot of Digital Assets trading still runs in the cloud (both exchanges and trading firms), while in traditional finance, traders have long colocated their servers alongside markets to minimize latency. Looking three years ahead, I expect to see more institutional participation in digital assets, increasingly AI-driven markets, and a broader shift from cloud infrastructure to bare metal as trading becomes more latency-sensitive.
Shout out to the Equinix APAC team, who brought together an extremely well curated audience for this conversation.
Last but not least, I came away impressed by Hong Kong, both by the energy of the city and the progress it’s making on digital asset regulation. Already looking forward to being back!
Do je, Hong Kong! 🇭🇰
The new @TheHUB website shows what we’ve been building: a facility in Tokyo where exchanges, trading firms, bare metal providers and node operators can deploy their infrastructure, connected through one network with equalized latency.
If you want to be part of it, DMs are open.
Yesterday I joined @wenaltseason’s Space and talked about why agentic frameworks are an important growth area for @TheHUB.
Big thanks to @sensible for hosting and to everyone who tuned in live!
If you missed the conversation, you can catch the recording below:
TL;DR: If you're price sensitive, you're latency sensitive. Whether you realize it or not.
When I pitch @TheHUB as ultra-low-latency trading infrastructure, I constantly hear the following feedback: "We're not super latency sensitive."
My answer is always the same: "Well, are you price sensitive? Because latency and jitter have a direct impact on pricing."
I think there's a common misconception in crypto markets regarding latency and jitter (latency volatility, in simple terms). People think it's only about speed, but let me use an analogy to make my point.
Think about credit operations. When a bank prices a loan, it considers a large number of variables in its risk model: the borrower's creditworthiness, probability of default, expected loss given default, collateral quality, loan duration, among others.
Everything impacts the final price of the credit line. Every uncertainty needs to be broken down into risk inputs in the model. The bigger the risk, the higher the interest rate required to compensate for it.
Back to the trading world.
Analogous to credit assessments, when you quote a trade, all the relevant risk factors (aka uncertainties) of that operation should be inputs to a risk model that ultimately impacts the price.
For a market maker quoting a price, the capacity to cancel or update orders is arguably one of the most critical components of its risk management framework. One of your main jobs as a market maker is to provide liquidity to the books while taking the smallest amount of risk possible.
You wanna be able to cancel orders as fast as you can if there's a big directional movement in the market. And you wanna be back providing liquidity as soon as the market stabilizes.
But that's just one part of the equation. Inventory exposure, market volatility, adverse selection, hedging costs, order fill probabilities and the risk of getting picked off on stale quotes are all factors that ultimately influence how a market maker prices its liquidity.
And latency and jitter directly affect several of these risks. If you know you can reliably cancel or update a quote in 100 microseconds, that's one risk profile. If it takes 10 milliseconds, that's another. And if it sometimes takes 100 microseconds and sometimes 10 milliseconds, that's yet another problem entirely. That last part is jitter, and it's often overlooked. It's not just about how fast you are, but how predictable/deterministic your infrastructure is.
The less predictable your execution and cancellation times, the harder it becomes to accurately model your exposure. And when uncertainty increases, you either have to price that risk into your quotes, reduce your exposure, or accept a higher probability of getting picked off.
This is why ultra-low-latency infrastructure isn't just about being faster than the next guy. It's about reducing risk, improving pricing efficiency and enabling tighter spreads.
And those benefits aren't limited to HFTs. Better pricing from liquidity providers can translate into better execution for everyone trading against them.
Which brings me back to my initial point:
You might not be latency sensitive. But the price you trade at is.
And while you might not have realized it yet, the big HFTs certainly have and they might just be eating a bite of your lunch every day without you even noticing.
Perps on stocks, indices and commodities from @tradexyz on @HyperliquidX keep trading after the underlying assets’ markets close every Friday, and the price they reach by the end of the weekend has turned out to be a good forecast of where those markets reopen.
An interesting report by @Blockworks says on 614 market-weekends found that in 95% of the cases where an asset moved at least 1% over the weekend, the move pointed in the same direction as the market when it reopened. Since the perp price is public throughout the weekend, it gives anyone a basis to mark risk or set margins before the open.
Traditional exchanges may still determine when an asset officially opens and closes, but they no longer have exclusive control over when its price is discovered.
Many of the trading venues we talk to don’t fit into the industry’s usual split between crypto and TradFi.
Think of a perps exchange that uses a traditional centralized order book for matching but settles transactions onchain. Or a fully centralized RFQ based exchange to trade crypto and TradFi perps, that accepts onchain assets as collateral.
These venues combine familiar financial instruments and trading mechanisms with different approaches to what people can trade, when they can trade and how transactions are settled. Calling them crypto doesn’t capture that, but neither does calling them TradFi. The name I like for them is neomarkets.
As these exchanges grow, they need what traditional stock exchanges have around their matching engines, including colocation for the firms that trade on them, without giving up the round-the-clock trading and open access that set them apart.
We built @TheHUB to be the home of these neomarkets.
Nodes for Solana and Hyperliquid are already being deployed at @TheHUB by @Enter_Elysium and @StakingFac.
Trading firms quoting on onchain exchanges often rely on infrastructure like these to access pre block information, balances, positions, and order book data.
An exchange can also run its own matching engine at TheHUB, with its market makers and node operators in the same facility instead of spread across different locations.
These are the 3 main verticals for TheHUB.
In market making, average latency only tells part of the story. You don’t get picked off on your average order. You get picked off when you’re in the tail.
A system doing 1ms most of the time but occasionally taking 10ms can be worse than one consistently doing 2ms.
That’s why reducing jitter matters as much as reducing latency itself. And why deterministic infrastructure all the way from market data ingestion to order entry matters.
Whether you’re a solo trader, prop shop, market maker, or hedge fund, I strongly recommend giving this series a read.
Shoutout to @Tristan0x. It’s hard to explain complex subjects in such a clear, and practical way.
How I made $6K PnL in 30 days on $10K principal market making perps ($158m+ volume).
The strat was entirely built by me and my quant, Claude (he won a national math competition in China).
Last night I asked if you'd want a guide. Hundreds of you said yes, so here it is: HFT for Dummies.
The honest timeline when I did it:
- Days for basic harness and venue integrations.
- Weeks to get it semi-competitive.
- Months to really tune and get it properly profitable.
This series can hopefully shortcut the learning curve for you.
I've been writing trading bots since 2017. It's how I got into crypto (free market data!), and how I met my co-founder.
Back then the barrier was high. You had to know how to code, to study data to find alpha, and to grind devops to have systems not blow up in production. Thankfully I taught myself Python and later Rust, but not everyone has a CS degree.
Nowadays with LLMs, anyone can have a skilled quant in their pocket. I use it for data analysis, strategy tuning, research and alpha generation and infra optimisation. It's a ridiculous time saver.
I'm extremely bullish on one-person quant shops. I think this will be a huge trend this next decade as finance comes on chain and market data and connectivity are accessible to anyone. I've already met many hobbyists who've made a living trading on e.g. Hyperliquid, enough so to quit their day jobs.
What we'll be building
A delta-neutral, reference-price market maker.
Binance is where most price discovery happens. We stream its market data, use it as a reference price, and quote a bid and an ask around it on another venue. I'm using @Bulletxyz because it's the one I know best.
Market making is the business of picking up pennies. Mine are under half a basis point per dollar. Enough pennies add up at the scale of millions of volume.
The catch is speed. If the market moves and you're slow to update, someone lifts your stale quote and you're the exit liquidity. It's a fun game of algorithmic cat and mouse.
On Bullet, takers get a 100ms speed bump. Why? It gives soon-to-be market makers like you a safety buffer so you don't get steamrolled by billion dollar firms with microwave towers.
The goal
The goal: tick to order sent in under 10ms. Stretch goal: under 1ms, which is very achievable if you're determined. We'll also focus on minimising jitter (randomness) because your slowest orders are the ones that will lose you the most pnl.
Agenda:
I'll aim to post one topic a day, going from easy wins, and getting progressively more deep in the weeds. If you work at any HFT firm this will all be elementary stuff, but hopefully it's helpful for newcomers.
Saving seconds
1. Basic intro to algo-trading (this post)
2. Running a simple market maker from your laptop
3. Measure once, cut twice: telemetry and Amdahl's law
4. AWS: Infra and colocation
Saving millis
5. Stop polling: websockets are your friend
6. Optimising order execution
7. How to not get rekt by API rate limits
8. Linux tuning: CPU pinning, IRQs and other nerd stuff
Saving micros (diminishing returns, but fun)
9. Optimising the hot path with CPU profiling in Rust
10. AWS: the last mile
11. Bonus: nonces (and other DeFi quirks)
I'll try share some open source code along with it. Don't expect to be beating Jump in the orderbooks just yet. It should be a good baseline that you can extend and take it as far as you want (you can take it pretty far!).
Tomorrow: setting up a basic MM in Rust, the architecture of which will roughly follow the diagram attached below.
We are very happy to announce that @StakingFac is deploying @solana and @HyperliquidX nodes at @TheHUB.
Node operators are a big part of the infrastructure onchain trading depends on, and Staking Facilities has been running node infrastructure across different networks for years. That’s the kind of experience we want inside TheHUB!
Shout out to the Staking Facilities team!