Sleep-deprived. Caffeinated. my life choices is questionable, but my shit posting is fire. 🖨 printing ONLY on @aptos
Joined May 2020
- Tweets16K
- Following315
- Followers724
- Likes6.6K
Pinned Tweet
A dark perp pool seems like an obvious solution if your goal is to hide your position and keep large size from becoming market information.
But there's a more valid question:
𝐖𝐡𝐲 𝐬𝐡𝐨𝐮𝐥𝐝 𝐛𝐞𝐢𝐧𝐠 𝐟𝐚𝐬𝐭𝐞𝐫 𝐭𝐡𝐚𝐧 𝐲𝐨𝐮 𝐝𝐞𝐭𝐞𝐫𝐦𝐢𝐧𝐞 𝐰𝐡𝐞𝐭𝐡𝐞𝐫 𝐬𝐨𝐦𝐞𝐨𝐧𝐞 𝐞𝐥𝐬𝐞 𝐠𝐞𝐭𝐬 𝐭𝐨 𝐞𝐱𝐭𝐫𝐚𝐜𝐭 𝐯𝐚𝐥𝐮𝐞 𝐟𝐫𝐨𝐦 𝐲𝐨𝐮𝐫 𝐩𝐨𝐬𝐢𝐭𝐢𝐨𝐧?
Crazy r̶a̶c̶e̶ game! 🔻
Given recent events, I think now might be a good time for someone to launch a dark pool perp DEX.
I have always been puzzled with the fact that everyone can see your orders in real-time on a DEX. The problem is worse on a perp DEX where there are liquidations.
Even with a CEX order book, where orders aren't linked to a specific individual, if you're looking to purchase $1 billion worth of a coin, you generally wouldn't want others to notice your order until it's completed. Otherwise, people might try to buy before you, effectively front-running you. In the case of a DEX, this can lead to MEV attacks. This results in increased slippage, worse prices, and higher costs for you.
For this reason, the large traders in TradFi use dark pools, which are often 10 times bigger than the “lit pools" (ie, normal orderbooks).
For perps (or futures), it is even more important to not let others know/see your orders. If others can see your liquidation point, they could try to push the market to liquidate you. Even if you got a billion dollars, others can gang up on you. This was possibly what we have seen recently.
The counter argument I have seen is that being more transparent allows market makers to absorb your large orders. This is possibly true. I won’t get into an argument on which is right or wrong. Different traders may prefer different types of markets.
Now might be a good opportunity for someone to launch an on-chain dark pool style DEX + perps, either by not showing the orderbook, or even better not showing deposits into smart contracts at all, or until much later. This should be doable with ZK or similar encryptions.
Just a Sunday idea. If you are building such a project, feel free to reach me on reachme.io/@cz_binance. No guarantees I will invest, or even reply. 🙏
So much to say about how creative this contest was.
It was truly beautiful to see how people communicated their understanding of what Superluminal is building.
Shout out to the top winners.
I read your submissions and they were super impressive. Great job!
Thank you to the Superluminal team @im_jowee and @RobertSagurton for the recognition.
It feels really good❤️
This next chapter is called Baranos!
Thrilled to see what the team is cooking with @BaranosAI .... Lezzzgoooooo!!!!!
Print retweeted
To cut straight to the chase, here's something I've been considering.
Yes, AI agents are the next big thing, especially for perps.
They're moving from "interesting idea" to actual trading infrastructure, with platforms now giving agents the ability to analyse markets and execute trades autonomously.
But if this is true,
Ar𝐞 𝐰𝐞 𝐛𝐮𝐢𝐥𝐝𝐢𝐧𝐠 𝐚𝐠𝐞𝐧𝐭𝐬 𝐭𝐨 𝐭𝐫𝐚𝐝𝐞 𝐛𝐞𝐭𝐭𝐞𝐫, 𝐨𝐫 𝐚𝐫𝐞 𝐰𝐞 𝐣𝐮𝐬𝐭 𝐛𝐮𝐢𝐥𝐝𝐢𝐧𝐠 𝐢𝐧𝐜𝐫𝐞𝐚𝐬𝐢𝐧𝐠𝐥𝐲 "𝐢𝐧𝐭𝐞𝐥𝐥𝐢𝐠𝐞𝐧𝐭" 𝐦𝐚𝐜𝐡𝐢𝐧𝐞𝐬 𝐭𝐨 𝐜𝐨𝐦𝐩𝐞𝐭𝐞 𝐢𝐧 𝐚 𝐥𝐚𝐭𝐞𝐧𝐜𝐲 𝐫𝐚𝐜𝐞?
If the market consistently rewards whoever reacts first, what exactly is the point of an agent's intelligence?
So we get: 𝐦𝐨𝐫𝐞 𝐚𝐠𝐞𝐧𝐭𝐬 → 𝐦𝐨𝐫𝐞 𝐜𝐨𝐦𝐩𝐞𝐭𝐢𝐭𝐢𝐨𝐧 → 𝐦𝐨𝐫𝐞 𝐥𝐚𝐭𝐞𝐧𝐜𝐲 → 𝐦𝐨𝐫𝐞 𝐢𝐧𝐟𝐫𝐚𝐬𝐭𝐫𝐮𝐜𝐭𝐮𝐫𝐞 → 𝐦𝐨𝐫𝐞 𝐬𝐩𝐞𝐞𝐝...
And eventually, the smartest agent isn't necessarily the one with the best strategy. It's the one that gets there first.
Doesn't that defeat the whole point of building an AI agent in the first place?
Ideally, an agent should reason across more information, adapt its strategy, manage risk, and make decisions a human couldn't make manually.
𝗕𝗨𝗧
If the market's final question is still "who got there first," then we've built intelligence all the way down just to hand the advantage back to latency.
And this isn't just theoretical.
Much of the current thinking around agentic trading agrees on one thing: execution and risk control still lag well behind the signal-generation side of things.
Which simply means, agents are getting better at deciding what to trade. They're not necessarily getting a fairer shot at trading it well.
So maybe the next breakthrough isn't simply making agents smarter. Maybe it's giving them a market where intelligence can actually matter.
A Superluminal market @slx_fi
👀
A market where the agent wins because it found the better trade, priced the risk better, and made the better decision, not because its transaction reached the book 10ms earlier.
𝗜𝗳 𝘄𝗲'𝗿𝗲 𝗯𝘂𝗶𝗹𝗱𝗶𝗻𝗴 𝗮𝘂𝘁𝗼𝗻𝗼𝗺𝗼𝘂𝘀 𝘁𝗿𝗮𝗱𝗲𝗿𝘀, 𝘀𝗵𝗼𝘂𝗹𝗱𝗻'𝘁 𝘁𝗵𝗲 𝗺𝗮𝗿𝗸𝗲𝘁 𝘁𝗵𝗲𝘆'𝗿𝗲 𝘁𝗿𝗮𝗱𝗶𝗻𝗴 𝗶𝗻 𝗲𝘃𝗼𝗹𝘃𝗲 𝘁𝗼𝗼?
#somethingtothinkabout🤔
Print retweeted
A dark perp pool seems like an obvious solution if your goal is to hide your position and keep large size from becoming market information.
But there's a more valid question:
𝐖𝐡𝐲 𝐬𝐡𝐨𝐮𝐥𝐝 𝐛𝐞𝐢𝐧𝐠 𝐟𝐚𝐬𝐭𝐞𝐫 𝐭𝐡𝐚𝐧 𝐲𝐨𝐮 𝐝𝐞𝐭𝐞𝐫𝐦𝐢𝐧𝐞 𝐰𝐡𝐞𝐭𝐡𝐞𝐫 𝐬𝐨𝐦𝐞𝐨𝐧𝐞 𝐞𝐥𝐬𝐞 𝐠𝐞𝐭𝐬 𝐭𝐨 𝐞𝐱𝐭𝐫𝐚𝐜𝐭 𝐯𝐚𝐥𝐮𝐞 𝐟𝐫𝐨𝐦 𝐲𝐨𝐮𝐫 𝐩𝐨𝐬𝐢𝐭𝐢𝐨𝐧?
Crazy r̶a̶c̶e̶ game! 🔻
Given recent events, I think now might be a good time for someone to launch a dark pool perp DEX.
I have always been puzzled with the fact that everyone can see your orders in real-time on a DEX. The problem is worse on a perp DEX where there are liquidations.
Even with a CEX order book, where orders aren't linked to a specific individual, if you're looking to purchase $1 billion worth of a coin, you generally wouldn't want others to notice your order until it's completed. Otherwise, people might try to buy before you, effectively front-running you. In the case of a DEX, this can lead to MEV attacks. This results in increased slippage, worse prices, and higher costs for you.
For this reason, the large traders in TradFi use dark pools, which are often 10 times bigger than the “lit pools" (ie, normal orderbooks).
For perps (or futures), it is even more important to not let others know/see your orders. If others can see your liquidation point, they could try to push the market to liquidate you. Even if you got a billion dollars, others can gang up on you. This was possibly what we have seen recently.
The counter argument I have seen is that being more transparent allows market makers to absorb your large orders. This is possibly true. I won’t get into an argument on which is right or wrong. Different traders may prefer different types of markets.
Now might be a good opportunity for someone to launch an on-chain dark pool style DEX + perps, either by not showing the orderbook, or even better not showing deposits into smart contracts at all, or until much later. This should be doable with ZK or similar encryptions.
Just a Sunday idea. If you are building such a project, feel free to reach me on reachme.io/@cz_binance. No guarantees I will invest, or even reply. 🙏
To cut straight to the chase, here's something I've been considering.
Yes, AI agents are the next big thing, especially for perps.
They're moving from "interesting idea" to actual trading infrastructure, with platforms now giving agents the ability to analyse markets and execute trades autonomously.
But if this is true,
Ar𝐞 𝐰𝐞 𝐛𝐮𝐢𝐥𝐝𝐢𝐧𝐠 𝐚𝐠𝐞𝐧𝐭𝐬 𝐭𝐨 𝐭𝐫𝐚𝐝𝐞 𝐛𝐞𝐭𝐭𝐞𝐫, 𝐨𝐫 𝐚𝐫𝐞 𝐰𝐞 𝐣𝐮𝐬𝐭 𝐛𝐮𝐢𝐥𝐝𝐢𝐧𝐠 𝐢𝐧𝐜𝐫𝐞𝐚𝐬𝐢𝐧𝐠𝐥𝐲 "𝐢𝐧𝐭𝐞𝐥𝐥𝐢𝐠𝐞𝐧𝐭" 𝐦𝐚𝐜𝐡𝐢𝐧𝐞𝐬 𝐭𝐨 𝐜𝐨𝐦𝐩𝐞𝐭𝐞 𝐢𝐧 𝐚 𝐥𝐚𝐭𝐞𝐧𝐜𝐲 𝐫𝐚𝐜𝐞?
If the market consistently rewards whoever reacts first, what exactly is the point of an agent's intelligence?
So we get: 𝐦𝐨𝐫𝐞 𝐚𝐠𝐞𝐧𝐭𝐬 → 𝐦𝐨𝐫𝐞 𝐜𝐨𝐦𝐩𝐞𝐭𝐢𝐭𝐢𝐨𝐧 → 𝐦𝐨𝐫𝐞 𝐥𝐚𝐭𝐞𝐧𝐜𝐲 → 𝐦𝐨𝐫𝐞 𝐢𝐧𝐟𝐫𝐚𝐬𝐭𝐫𝐮𝐜𝐭𝐮𝐫𝐞 → 𝐦𝐨𝐫𝐞 𝐬𝐩𝐞𝐞𝐝...
And eventually, the smartest agent isn't necessarily the one with the best strategy. It's the one that gets there first.
Doesn't that defeat the whole point of building an AI agent in the first place?
Ideally, an agent should reason across more information, adapt its strategy, manage risk, and make decisions a human couldn't make manually.
𝗕𝗨𝗧
If the market's final question is still "who got there first," then we've built intelligence all the way down just to hand the advantage back to latency.
And this isn't just theoretical.
Much of the current thinking around agentic trading agrees on one thing: execution and risk control still lag well behind the signal-generation side of things.
Which simply means, agents are getting better at deciding what to trade. They're not necessarily getting a fairer shot at trading it well.
So maybe the next breakthrough isn't simply making agents smarter. Maybe it's giving them a market where intelligence can actually matter.
A Superluminal market @slx_fi
👀
A market where the agent wins because it found the better trade, priced the risk better, and made the better decision, not because its transaction reached the book 10ms earlier.
𝗜𝗳 𝘄𝗲'𝗿𝗲 𝗯𝘂𝗶𝗹𝗱𝗶𝗻𝗴 𝗮𝘂𝘁𝗼𝗻𝗼𝗺𝗼𝘂𝘀 𝘁𝗿𝗮𝗱𝗲𝗿𝘀, 𝘀𝗵𝗼𝘂𝗹𝗱𝗻'𝘁 𝘁𝗵𝗲 𝗺𝗮𝗿𝗸𝗲𝘁 𝘁𝗵𝗲𝘆'𝗿𝗲 𝘁𝗿𝗮𝗱𝗶𝗻𝗴 𝗶𝗻 𝗲𝘃𝗼𝗹𝘃𝗲 𝘁𝗼𝗼?
#somethingtothinkabout🤔
In practice, there may be no universal winner in the DFBA vs CLOB debate.
Different markets need different things.
But there's a difference between arguing about architecture and asking the real questions:
Does the trader get a fair shot before someone faster extracts the edge?
Is there a better experience where you're not priced out before you even get a chance to compete?
Does price matter more than latency?
Is your size automatically alpha for everyone watching?
How much of your edge should come from being right, and how much from getting there first?
How would you know you could've won, if you never got a fair seat at the table before the deck was already stacked against you?
Print retweeted
On March 12, 2026, a rider from Bolt Nigeria STOLE my package worth over ₦400,000.
I did YOUR job for you, tracked him myself, gathered clear evidence, and reported at your Lekki Phase 1 office on March 13. You took all my evidence and promised to act.
Print retweeted
have faith in @Aptos & @AptosLabs
team is giga cooking, best thing we can do is support
while @shelbyserves & @decibeltrade are still in the oven, we should
→ learn how they work
→ figure out how to position before mainnet
both were announced super early, which is actually an edge if u use the time to study.
would love to see more shelby + decibel content on my timeline
aptos.
Print retweeted
A new demonstration of Aptos' real-world adoption 🇯🇵
The digital wallet for the Universal Expo in Osaka (@expo2025_japan) is powered by @Aptos.
Over 4.4M transactions recorded and 500K+ accounts registered.
A very smooth user experience, because that’s what you get with Aptos 🔥
