Pinned Tweet
7 roadmap eras. Here's the rest of the iceberg:
Nillion Dusk, now. Sealed orders on Ethereum that fire on price or time. No bots to run, no one to trust. An SDK built for agents, apps in a few lines of TypeScript, and anyone can run a node by staking NIL. Mainnet early Oct
Darkfall, Nov to Feb. One network across Hyperliquid, Base, Robinhood and the rest of EVM, native on each. Stake NIL once on Ethereum, earn fees from every chain you serve. Misbehaving nodes get jailed and slashed
Skyglow, Feb to May. Watch a condition on one chain, settle on another. Nothing bridged, nothing wrapped, no value locked. Goes past EVM too, Solana and Move chains
Nightside (my fav), May to Aug. Custom conditions of any complexity, resolved by any feed the nodes can verify, checked at high frequency. A wallet moving, a court ruling, a CPI print, a shipment. Nodes specialise in feeds and get paid for the facts they confirm
Neon, Aug to Oct. Covenants get their own language and VM, so conditions become programmable. Nest them, do maths on them, loop them. The design space blows wide open
Aurora, Oct to Jan. Long-dated covenants get a second, quantum resistant seal on top of the first. FHE coprocessors compute on conditions without ever decrypting them
Moonshine, Dec to Feb. Ciphertext moves to blobs at about a tenth of the cost, reveals get verified with a single ZK proof, cent-priced covenants, and you can stake NIL for capacity instead of paying per covenant. As volume grows the network starts paying for itself, so it can print less new NIL
Era one of covenants is live. It's called Dusk
A price and a clock. That alone kills stop hunting, launch sniping and bid leaking, and it's just the tip of the iceberg
Underneath sits everything the market doesn't see coming. Six more eras of capability, unlocking use cases nobody has even thought to want yet, surfacing on a public schedule between now and 2028
Dusk broke the surface yesterday. Get your hands on it while everyone else is still staring at the waterline
Mark Mac retweeted
Five ways to use Covenants as a layer on top of prediction markets:
(1) Events fire bets. CPI comes in above forecast and your YES on a Fed hike in October fires. Same for an exit on a position you hold, or a bet that waits for two events
(2) Odds fire bets. The market on LeBron playing Game 7 resolves NO and your YES on the Celtics winning the series fires. Same for a stop loss on your bet, a bet that fires when a correlated market’s odds cross a level, or an arb between Polymarket and Kalshi
(3) Odds fire trades. The market on a 2026 recession crosses 40% and your ETH short opens. The same market resolves NO and the short closes
(4) Resolutions fire payouts. The market on a hurricane hitting Florida resolves YES and the insurance pays out, no claims desk. Bounties, escrow, refunds and DAO buybacks work the same way
(5) Agents run all of it. An agent seals fifty of these across five venues and goes offline. They still fire
Every one sits sealed off venue till it fires, so nobody can read it or copy it
Been looking at who actually makes money on Polymarket when big news drops. About 3% of accounts, and most of them are bots
London Business School and Yale went through 1.72m Polymarket accounts. The ones making money buy the right side of the market in the first few minutes after the news, before the odds catch up. To do that you need a VPS, the CLOB API, a websocket feed, a private key on the box and someone awake at 4am to restart it when it crashes. Which is why it's 3% and not 30%. The study looked at Fed announcements, but it's the same on a CPI print, a court ruling, an ETF approval, anything
You can't beat them by hand on the headline markets. And the long tail, the 1,700+ markets under 1m of volume that reprice slower after news because no bot is watching them, is too many for a human to watch either. That's the one place a normal account can win, and nobody's set up to take it
With Covenants you write the trade before the news. Buy YES on a Fed hike if CPI comes in hot, buy NO on a December cut if payrolls come in 100,000 above forecast. Both sit sealed, nobody can read them, and they fire on their own, no server needed. On the headline markets your order goes in the same second as the bots. On the small markets the bots aren't even there, so if you have alpha on one, you're first in line
Gonna spend this week going through everything a prediction market can trigger. A lot more than I expected tbh
Private transactions ≠ private trading strategies.
Onchain markets have 3 layers: transaction, execution and intelligence.
Even if a transaction is private, your stop losses and limit orders can still reveal your next move.
Nillion's Covenants keep them encrypted.
All five of these sit in front of existing venues like Polymarket and Kalshi. There’s another category where a venue builds Covenants into how its own markets resolve. Gonna post about that one later this week
Five ways to use Covenants as a layer on top of prediction markets:
(1) Events fire bets. CPI comes in above forecast and your YES on a Fed hike in October fires. Same for an exit on a position you hold, or a bet that waits for two events
(2) Odds fire bets. The market on LeBron playing Game 7 resolves NO and your YES on the Celtics winning the series fires. Same for a stop loss on your bet, a bet that fires when a correlated market’s odds cross a level, or an arb between Polymarket and Kalshi
(3) Odds fire trades. The market on a 2026 recession crosses 40% and your ETH short opens. The same market resolves NO and the short closes
(4) Resolutions fire payouts. The market on a hurricane hitting Florida resolves YES and the insurance pays out, no claims desk. Bounties, escrow, refunds and DAO buybacks work the same way
(5) Agents run all of it. An agent seals fifty of these across five venues and goes offline. They still fire
Every one sits sealed off venue till it fires, so nobody can read it or copy it
Been looking at who actually makes money on Polymarket when big news drops. About 3% of accounts, and most of them are bots
London Business School and Yale went through 1.72m Polymarket accounts. The ones making money buy the right side of the market in the first few minutes after the news, before the odds catch up. To do that you need a VPS, the CLOB API, a websocket feed, a private key on the box and someone awake at 4am to restart it when it crashes. Which is why it's 3% and not 30%. The study looked at Fed announcements, but it's the same on a CPI print, a court ruling, an ETF approval, anything
You can't beat them by hand on the headline markets. And the long tail, the 1,700+ markets under 1m of volume that reprice slower after news because no bot is watching them, is too many for a human to watch either. That's the one place a normal account can win, and nobody's set up to take it
With Covenants you write the trade before the news. Buy YES on a Fed hike if CPI comes in hot, buy NO on a December cut if payrolls come in 100,000 above forecast. Both sit sealed, nobody can read them, and they fire on their own, no server needed. On the headline markets your order goes in the same second as the bots. On the small markets the bots aren't even there, so if you have alpha on one, you're first in line
Gonna spend this week going through everything a prediction market can trigger. A lot more than I expected tbh
Mark Mac retweeted
Been looking at who actually makes money on Polymarket when big news drops. About 3% of accounts, and most of them are bots
London Business School and Yale went through 1.72m Polymarket accounts. The ones making money buy the right side of the market in the first few minutes after the news, before the odds catch up. To do that you need a VPS, the CLOB API, a websocket feed, a private key on the box and someone awake at 4am to restart it when it crashes. Which is why it's 3% and not 30%. The study looked at Fed announcements, but it's the same on a CPI print, a court ruling, an ETF approval, anything
You can't beat them by hand on the headline markets. And the long tail, the 1,700+ markets under 1m of volume that reprice slower after news because no bot is watching them, is too many for a human to watch either. That's the one place a normal account can win, and nobody's set up to take it
With Covenants you write the trade before the news. Buy YES on a Fed hike if CPI comes in hot, buy NO on a December cut if payrolls come in 100,000 above forecast. Both sit sealed, nobody can read them, and they fire on their own, no server needed. On the headline markets your order goes in the same second as the bots. On the small markets the bots aren't even there, so if you have alpha on one, you're first in line
Gonna spend this week going through everything a prediction market can trigger. A lot more than I expected tbh
Mark Mac retweeted
Used the weekend to build out the protection side of Latch and based on the feedbacks I got from the team and community
You can now set an automatic stablecoin exit: “Exchange my USDT for USDC if USDT falls below $0.99, but only if I receive at least the amount I choose.”
Paid depeg cover is here too.
Choose a provider, request private terms and review the payout, premium and protection period before paying.
The provider reserves the full payout upfront and activates the policy. If the agreed condition qualifies, the payout can happen automatically.
Providers now have a workspace to manage capital, reply to requests, activate policies and track payouts. Sentinel lets them review exposure and pause new purchases while keeping existing commitments protected.
I also added a one-time email alert for sealing availability. Turn it on, confirm your email, and Latch will notify you when the required three-node committee can be formed. After sending, the alert turns itself off (love this one a lot, helped a couple of times to know when nodes are up more than 3 😄)
The trading agent, chat and MCP tools remain part of the app.
Built with @nillionnetwork Covenants. Available on Sepolia with test assets fully available on the platform to test.
Try it: latch.femdev.site
Feedback welcome, especially on the cover and exit flows.
I wasn’t done cooking. Two ideas from Nillion’s wishlist, one app 👨🍳
Meet Latch
A trading agent that can go offline while your sealed instructions keep working, plus cover that pays automatically if USDT loses its peg.
Trading
“Buy WETH if BTC reaches $90k.”
“Take profit at $4k, or stop loss at $2.5k.”
Describe your plan in chat, review it and approve through your wallet. Your conditions stay sealed until release. Paired exits share one allocation: when one sale succeeds, the other closes.
Depeg cover
“If USDT falls to $0.98 or below, pay this wallet 10 USDC.”
A provider reserves the payout upfront. The beneficiary accepts the terms. Once activated, the sealed condition can trigger payment automatically, no claim form or waiting for someone to approve your payout.
MCP lets compatible AI assistants like Claude, chat-gpt, cursor, codex etc prepare plans, check funding and explain their status. Your wallet authorizes activation; receipts show the outcome.
Built with @nillion Covenants
Live on Sepolia with test assets.
Try it: latch.femdev.site
Been looking at who actually makes money on Polymarket when big news drops. About 3% of accounts, and most of them are bots
London Business School and Yale went through 1.72m Polymarket accounts. The ones making money buy the right side of the market in the first few minutes after the news, before the odds catch up. To do that you need a VPS, the CLOB API, a websocket feed, a private key on the box and someone awake at 4am to restart it when it crashes. Which is why it's 3% and not 30%. The study looked at Fed announcements, but it's the same on a CPI print, a court ruling, an ETF approval, anything
You can't beat them by hand on the headline markets. And the long tail, the 1,700+ markets under 1m of volume that reprice slower after news because no bot is watching them, is too many for a human to watch either. That's the one place a normal account can win, and nobody's set up to take it
With Covenants you write the trade before the news. Buy YES on a Fed hike if CPI comes in hot, buy NO on a December cut if payrolls come in 100,000 above forecast. Both sit sealed, nobody can read them, and they fire on their own, no server needed. On the headline markets your order goes in the same second as the bots. On the small markets the bots aren't even there, so if you have alpha on one, you're first in line
Gonna spend this week going through everything a prediction market can trigger. A lot more than I expected tbh
Mark Mac retweeted
few understand how widely applicable covenants can be -- when they said encrypted markets, they mean every market -- perps, yield, PMs and more -- the TAM is fucking massive
ethereum:0x7cf9a80db3b29ee8efe3710aadb7b95270572d47
Spent the weekend deep diving into the stuff Covenants can do in prediction markets and it’s super exciting
Will be sharing some of that this week. Stay tuned 👀
Mark Mac retweeted
This is wild. It fired!
19:56 UTC, minutes after I posted. All 250 ZEC sold at $1,049, 53k loss locked in. ZEC printed $1,048.40 for one minute and two hours later it’s $1,100
Brutal. This guy did the responsible thing, set a stop to protect himself, and the venue put it on display for anyone to aim at. 53k gone, and ZEC ripped straight back up without those 250 coins
This needs fixing. With Covenants nobody knows where your line is
Oh damn, found another one. ZEC this time
Someone’s long 625 ZEC at 10x from $1,264, with a public stop loss on 250 ZEC at $1,050. ZEC’s at $1,057 as I type, so that’s under 1% from his line. A Sunday arvo job for a desk with size
Wick it there, his stop dumps 262.5k into the dip and locks in a 53k loss, then price bounces without him. Liq is at $928. His ETH, XRP and NEAR stops are all public too. Whole playbook, free to anyone with the address. Hunters gonna hunt
With Covenants the stop stays sealed until $1,050 actually prints. Nobody can read it while it waits
Mark Mac retweeted
He set a stop to protect himself and the venue showed it to everyone. Taken out at $1,049, and ZEC is $1,100 two hours later
Every trader on a public venue is in the same spot. That’s what we’re building Covenants to fix
This is wild. It fired!
19:56 UTC, minutes after I posted. All 250 ZEC sold at $1,049, 53k loss locked in. ZEC printed $1,048.40 for one minute and two hours later it’s $1,100
Brutal. This guy did the responsible thing, set a stop to protect himself, and the venue put it on display for anyone to aim at. 53k gone, and ZEC ripped straight back up without those 250 coins
This needs fixing. With Covenants nobody knows where your line is
This is wild. It fired!
19:56 UTC, minutes after I posted. All 250 ZEC sold at $1,049, 53k loss locked in. ZEC printed $1,048.40 for one minute and two hours later it’s $1,100
Brutal. This guy did the responsible thing, set a stop to protect himself, and the venue put it on display for anyone to aim at. 53k gone, and ZEC ripped straight back up without those 250 coins
This needs fixing. With Covenants nobody knows where your line is
Oh damn, found another one. ZEC this time
Someone’s long 625 ZEC at 10x from $1,264, with a public stop loss on 250 ZEC at $1,050. ZEC’s at $1,057 as I type, so that’s under 1% from his line. A Sunday arvo job for a desk with size
Wick it there, his stop dumps 262.5k into the dip and locks in a 53k loss, then price bounces without him. Liq is at $928. His ETH, XRP and NEAR stops are all public too. Whole playbook, free to anyone with the address. Hunters gonna hunt
With Covenants the stop stays sealed until $1,050 actually prints. Nobody can read it while it waits
Your trading strategy can be exposed before your trade even happens.
And that creates a problem many crypto users overlook.
On transparent blockchains the logic behind a trade or automated strategy can reveal what someone plans to do Other market participants may use that information to position themselves ahead of it.
This is one of the problems @nillion is exploring with Encrypted Markets.
Its Covenants let users seal an instruction onchain so it remains encrypted until a specified condition is met.
The interesting part isn't just keeping information private.
It's the possibility of protecting a strategy until the moment it's meant to act.
Because in competitive markets privacy isn't only about hiding who you are.
Sometimes it's about keeping your next move your own.
I'm exploring how @nillion approach could change the way we think about onchain markets.
Research: nillion.com/news/the-road-to…
#Nillion #Web3
Oh damn, found another one. ZEC this time
Someone’s long 625 ZEC at 10x from $1,264, with a public stop loss on 250 ZEC at $1,050. ZEC’s at $1,057 as I type, so that’s under 1% from his line. A Sunday arvo job for a desk with size
Wick it there, his stop dumps 262.5k into the dip and locks in a 53k loss, then price bounces without him. Liq is at $928. His ETH, XRP and NEAR stops are all public too. Whole playbook, free to anyone with the address. Hunters gonna hunt
With Covenants the stop stays sealed until $1,050 actually prints. Nobody can read it while it waits
PUMP's the only big alt green on Hyperliquid today so I went looking at who's long it and how
One account is long 249m PUMP, $1,087,000 at 3x, in from 0.005265. Price is 0.004363, so he's already down $225,000. And sitting under it, in public, a stop on the whole position at 0.003949. That's 9.5% below price. PUMP has moved 12 to 23% in a day three times this week, so pushing it there is not a big ask for anyone with size
If it fires, the stop sells all 249m PUMP into the dip and he's locked in a $328,000 loss. His stop allows fills down to 0.003554, so a hard wick can fill him lower and the loss runs to $426,000. He put up about $510,000 for this trade, so that leaves him $84,000. Then price bounces back, he has no position left, and he watches the coin he just got kicked out of go green
Market makers and big accounts do this for a living. Thin weekend book, a stop everyone can see. They sell PUMP down until 0.003949 prints, his stop dumps $984,000 at market, they buy it off him at the low and let price drift back up. His $328,000 loss is their profit. It's a perp venue, so the price that fires his stop is the perp price, not spot, and perp prices wick far harder. On Binance or Bybit a stop can fire off that one exchange's last trade, so one wick on one book is all it takes
Finding this took zero skill. Hyperliquid's public API returns every open order and position for any address, trigger price included. The screenshot is the raw response for his wallet, and his liquidation price at 0.003388 is sitting right there too. A desk doesn't even do this by hand. A script polls the API, maps every stop loss on the book, and ranks them by how cheap they are to hit. He's basically a row in a spreadsheet
A stop loss is the one order you can't afford to show, and the venue shows it to everyone. With Covenants the same stop sits sealed until 0.003949 actually prints, then fires on its own. Nobody can read the level while it waits
The quoted post shows one that already happened. A ZEC stop at $773.94, taken to the exact tick, then reversed. ZEC is $1,158 today
Mark Mac retweeted
PUMP's the only big alt green on Hyperliquid today so I went looking at who's long it and how
One account is long 249m PUMP, $1,087,000 at 3x, in from 0.005265. Price is 0.004363, so he's already down $225,000. And sitting under it, in public, a stop on the whole position at 0.003949. That's 9.5% below price. PUMP has moved 12 to 23% in a day three times this week, so pushing it there is not a big ask for anyone with size
If it fires, the stop sells all 249m PUMP into the dip and he's locked in a $328,000 loss. His stop allows fills down to 0.003554, so a hard wick can fill him lower and the loss runs to $426,000. He put up about $510,000 for this trade, so that leaves him $84,000. Then price bounces back, he has no position left, and he watches the coin he just got kicked out of go green
Market makers and big accounts do this for a living. Thin weekend book, a stop everyone can see. They sell PUMP down until 0.003949 prints, his stop dumps $984,000 at market, they buy it off him at the low and let price drift back up. His $328,000 loss is their profit. It's a perp venue, so the price that fires his stop is the perp price, not spot, and perp prices wick far harder. On Binance or Bybit a stop can fire off that one exchange's last trade, so one wick on one book is all it takes
Finding this took zero skill. Hyperliquid's public API returns every open order and position for any address, trigger price included. The screenshot is the raw response for his wallet, and his liquidation price at 0.003388 is sitting right there too. A desk doesn't even do this by hand. A script polls the API, maps every stop loss on the book, and ranks them by how cheap they are to hit. He's basically a row in a spreadsheet
A stop loss is the one order you can't afford to show, and the venue shows it to everyone. With Covenants the same stop sits sealed until 0.003949 actually prints, then fires on its own. Nobody can read the level while it waits
The quoted post shows one that already happened. A ZEC stop at $773.94, taken to the exact tick, then reversed. ZEC is $1,158 today
Last night I pulled 600 Hyperliquid wallets at random. 182 were running a live position, and 36 of those had a stop loss or a take profit resting with the trigger price coming back in plain text from a public API call
Here's what that costs. On 22 Aug someone was long 104 ZEC with a stop at 773.94. It sat readable for four hours. Price came down, took every coin at exactly 773.94, and reversed
A day later ZEC was 812. A week later 805. It's 1,461 now
Same position today would be worth $72,000 more if they stayed in. But the one number that could take them out was published
The app below is the first of its kind where that number stays sealed until it fires
PUMP's the only big alt green on Hyperliquid today so I went looking at who's long it and how
One account is long 249m PUMP, $1,087,000 at 3x, in from 0.005265. Price is 0.004363, so he's already down $225,000. And sitting under it, in public, a stop on the whole position at 0.003949. That's 9.5% below price. PUMP has moved 12 to 23% in a day three times this week, so pushing it there is not a big ask for anyone with size
If it fires, the stop sells all 249m PUMP into the dip and he's locked in a $328,000 loss. His stop allows fills down to 0.003554, so a hard wick can fill him lower and the loss runs to $426,000. He put up about $510,000 for this trade, so that leaves him $84,000. Then price bounces back, he has no position left, and he watches the coin he just got kicked out of go green
Market makers and big accounts do this for a living. Thin weekend book, a stop everyone can see. They sell PUMP down until 0.003949 prints, his stop dumps $984,000 at market, they buy it off him at the low and let price drift back up. His $328,000 loss is their profit. It's a perp venue, so the price that fires his stop is the perp price, not spot, and perp prices wick far harder. On Binance or Bybit a stop can fire off that one exchange's last trade, so one wick on one book is all it takes
Finding this took zero skill. Hyperliquid's public API returns every open order and position for any address, trigger price included. The screenshot is the raw response for his wallet, and his liquidation price at 0.003388 is sitting right there too. A desk doesn't even do this by hand. A script polls the API, maps every stop loss on the book, and ranks them by how cheap they are to hit. He's basically a row in a spreadsheet
A stop loss is the one order you can't afford to show, and the venue shows it to everyone. With Covenants the same stop sits sealed until 0.003949 actually prints, then fires on its own. Nobody can read the level while it waits
The quoted post shows one that already happened. A ZEC stop at $773.94, taken to the exact tick, then reversed. ZEC is $1,158 today
Last night I pulled 600 Hyperliquid wallets at random. 182 were running a live position, and 36 of those had a stop loss or a take profit resting with the trigger price coming back in plain text from a public API call
Here's what that costs. On 22 Aug someone was long 104 ZEC with a stop at 773.94. It sat readable for four hours. Price came down, took every coin at exactly 773.94, and reversed
A day later ZEC was 812. A week later 805. It's 1,461 now
Same position today would be worth $72,000 more if they stayed in. But the one number that could take them out was published
The app below is the first of its kind where that number stays sealed until it fires
For anyone wondering how often this actually happens, I pulled 1,500 Hyperliquid wallets last month. More than half the stops that fired reversed within a day, median 2%. The thread 👇
xitter.cf/thatmarkmac/status/210…



























